| <<<Back 1 day (to 2014/06/30) | 2014/07/01 |
pedro_mac | morning folks | 06:32.52 |
kens | Hi chrisl, how was the IoW ? | 07:02.05 |
chrisl | morning kens. IoW was really nice - almost surprisingly so | 07:03.49 |
kens | At least you (mostly) had decent weahter | 07:04.04 |
chrisl | The weather was all great - only rain was ~3am, and we weren't out at 3am! | 07:04.29 |
kens | Ah that was handy, it was wet here on Satuday afternoon | 07:04.47 |
chrisl | That's probably the rain we had - it was 3am on Saturday morning | 07:05.32 |
kens | Hmm GMail marked Zoltan's mail as spam :-) | 07:08.34 |
pedro_mac | chrisl: thatâs pretty good going :) | 07:08.52 |
chrisl | pedro_mac: yeh, we were pleasantly surprised - we kept thinking we'd get caught in a deluge miles from anywhere, but it never happened | 07:15.46 |
jogux_mac | good morning | 08:54.56 |
kens | Morning | 08:55.03 |
jogux | there's something fabuously wrong with Glasgow's weather this summer | 08:55.49 |
Robin_Watts | morning | 08:55.50 |
pedro_mac | hi Robin | 08:55.59 |
kens | I take it you can see the Sun ? | 08:56.08 |
jogux | kens : I can barely look in the direction of the sun | 08:56.36 |
kens | :-) | 08:56.44 |
chrisl | Got to get all that good weather out the way before the Games start! | 08:56.58 |
sebras | good morning! | 08:58.45 |
jogux | chrisl : hehe | 08:59.55 |
Robin_Watts | pedro_mac, jogux: So, are either of you at all interested in the font engine? | 09:00.13 |
| I think I want to bend it a bit to make handling spaces better. | 09:00.35 |
pedro_mac | Robin_Watts: thatâs leading question⦠;) | 09:00.41 |
jogux | remembers what happened when I mentioned knowing about styling | 09:00.45 |
Robin_Watts | Every Font_Handle is basically a transform + set of OpenFonts. | 09:01.09 |
jogux | nods. | 09:01.20 |
Robin_Watts | where each OpenFont is the actual object that corresponds to a real font. And that set is used as the list of fonts to 'fall back' through. | 09:01.57 |
| One of those fonts is a 'Spaces' font, that exists purely to offer widths/metrics/null outlines for the common unicode space chars. | 09:02.28 |
pedro_mac | yup | 09:03.07 |
Robin_Watts | In the current code we have hardwired widths for the spaces that it supports, but there are 6 space chars that it does not support, supposedly because their size is should be calculated from the actual fonts in use. | 09:03.44 |
| for example, the width of the 'ideographic space' should be set to be the same as an ideograph. | 09:04.03 |
| likewise there is punctuation space and figure space. | 09:04.32 |
| Now, it was the ideographic space that was causing the [] in Artifex 2014. | 09:05.01 |
| so that's the one I need to get supported. | 09:05.11 |
chrisl | Wouldn't it be better to add those space glyphs to the actual fonts, instead of relying on some fallback? | 09:06.02 |
Robin_Watts | chrisl: what if the fonts involved aren't ours? | 09:06.24 |
kens | ideographic languages don't generally use spaces at all..... | 09:06.34 |
chrisl | Then there's a missing glyph, and the file is wrong | 09:06.46 |
kens | And a fixed width space probably onl;y works if its a fixed width font | 09:06.52 |
Robin_Watts | kens: I have many examples here of files that do. | 09:07.02 |
kens | Robin_Watts : do what ? Use spaces ? | 09:07.16 |
Robin_Watts | yes. | 09:07.20 |
kens | Hmm, for what purpose ? | 09:07.30 |
| Though since this is MS Offcie, nothing surprsies me | 09:07.47 |
Robin_Watts | specifically I have lots of files with japanese/chinese fonts in that have ideographic spaces in the middle of runs of ideographs. | 09:08.08 |
paulgardiner | jogux: I believe I can see what difference between master and the branch is affecting this spreadsheet-ruler problem. | 09:08.10 |
kens | In any event, a fixed width space is only correct for a fixed width font. THere are proportional CJKV fonts too | 09:08.13 |
Robin_Watts | kens, sure. | 09:08.29 |
kens | ie the space may not be the size of an ideogram | 09:08.36 |
| Robin_Watts : How would you have one 'space' which works for all fonts ? | 09:09.12 |
Robin_Watts | kens, well, presumably if people are using proportional fonts, then the ideographic space will be some value near the average of the ideograms. | 09:09.18 |
paulgardiner | jogux: We have the DOCVIEW_ZOOMSET feature enabled. It's part of removing simplelib, but master has it independently of REMOVE_SIMPLELIB. | 09:09.23 |
Robin_Watts | kens: I wouldn't, that's the point. | 09:09.25 |
kens | OK then I misunderstand, I thought you were saying that you would add an ideographic space to the 'spaces' font | 09:09.47 |
pedro_mac | Robin_Watts: are you thinking of synthesising the spaces metrics on open of each font? | 09:10.00 |
paulgardiner | jogux: it completely changes how rulers are controlled and it doesn't currently allow for separation between the focus area and where the rulers are, AFAIKS | 09:10.12 |
Robin_Watts | My first attempt was to hardwire a value in for the ideographic space in the spaces font. And that works, but it's not ideal as you just predicted. | 09:10.25 |
| pedro_mac: Kinda. | 09:10.38 |
| My new plan is to: | 09:10.44 |
kens | Robin_Watts : can't you just look in the font for a space glyph ? Then you only need to synthesise one if its not present | 09:11.06 |
Robin_Watts | kens: Fallback already gives us that. | 09:11.27 |
kens | I guess I don't understand the problem then..... | 09:11.44 |
Robin_Watts | If there are spaces in the original fonts then those will be supplied and everything is fine. | 09:11.51 |
kens | is unsurprised by that | 09:11.56 |
Robin_Watts | The problem is if the font we are using does not have an ideographic space. | 09:12.13 |
kens | Oh, then I think I'm with Chris, but this isn't printing so.... | 09:12.16 |
| Can't you check for other kinds of spaces first ? | 09:12.28 |
Robin_Watts | kens: how would that help? | 09:13.01 |
paulgardiner | jogux: so possibilities are - 1) update the zoomset feature to allow for the separation between rulers and focus area, 2) turn off DOCVIEW_ZOOMSET until we have more time, 3) go with my current fix that changed the focus area behaviour to a small extent | 09:13.06 |
kens | Well presumably *a* space is more likely to be correct in a given font, than a random synthetic space | 09:13.23 |
Robin_Watts | kens: Not really, no. | 09:13.41 |
kens | Well, you have the examples, not me.... | 09:13.53 |
Robin_Watts | The width of a vanilla space in (say) Arial, is likely to be completely unrelated to the widths of an ideograph in that font. | 09:14.17 |
| or at least will be related in a non-standard way. | 09:14.28 |
kens | Turue, but in such a font the average of all widths seems no more likely (to me) to be correct either | 09:14.53 |
| It will presumabl;y only be correct in a font mostly composed of ideographs | 09:15.09 |
Robin_Watts | kens: right, but I wasn't suggesting we'd use the average of all widths. | 09:15.15 |
kens | I should shut up and let you finish | 09:15.33 |
Robin_Watts | So each openfont has a 'getWidths' method that is called to fill in the widths for the given used unicode values. | 09:16.08 |
| Each openFonts getWidths method is called in turn until all the widths for the required string are filled in. | 09:17.19 |
| The spaces one, just looks up the widths for the space chars that it knows about in its hardwired table, and fills those in. | 09:17.44 |
jogux | paulgardiner: ah, cool... I guess that kinda of makes sense. | 09:18.33 |
| paulgarinder: I'm not 100% sure I know what the 'focus area' actually is | 09:18.43 |
| my temptation would be to go with 3), but I don't know if end users might see that as a regression. | 09:19.08 |
Robin_Watts | I propose to try and change it so that for things like 'ideographic_space' or 'punctuation_space' etc, it will look up the width of an ideograph/punctuation in the earlier fonts and will then use that. | 09:19.09 |
| So it'll still be a guess, but it'll be a guess based on the characteristics of the actual font(s) in use. | 09:19.48 |
paulgardiner | jogux: it would be a change only in the iOS version | 09:19.53 |
| a regression I mean | 09:20.01 |
mattchz | morning | 09:20.07 |
jogux | robin_watts : well it can't be worse than a box. :) | 09:20.25 |
kens | Robin_Watts : Well its a reasonable last ditch attempt I guess. | 09:20.52 |
paulgardiner | I'm just trying out turning off DOCVIEW_ZOOMSET and that may be problematic. Looks like there have been changes since it was enabled that rely on it | 09:21.13 |
pedro_mac | Robin_Watts: that sounds like a reasonably decent option in this situation | 09:21.15 |
chrisl | Seems to me this stuff (M$ Office, not SO) pretty much guarantees varying degrees of wrong, and never right, in portability terms....... | 09:22.00 |
jogux | paulgardiner: I'm not sure I'd be overly keen on turning it off, it's moving the code in the wrong direction. and you already found out it probably won't work straight off either :-) | 09:22.08 |
Robin_Watts | jogux, pedro_mac: I probably need to be prepared to look up several ideograms, and several pieces of punctuation until I find one that is there. | 09:22.10 |
| which means it might be a bit slow. | 09:22.15 |
| which means I need to cache the results. | 09:22.26 |
jogux | robin_watts : yeah, I was about to ask | 09:22.27 |
paulgardiner | I could have a brief look at option 1 | 09:22.41 |
Robin_Watts | which means, I think, that I need to have some storage in the Font_Handle. | 09:22.42 |
jogux | robin_watts: wouldn't the cache go in OpenFont ? | 09:23.26 |
Robin_Watts | and it means that the getWidths function needs to be able to get to the other openfonts in the Font_Handle, rather than just being an entirely local thing. | 09:23.36 |
| so it harms the modularity a bit. | 09:23.44 |
| but I think it's the 'right' thing to do. Just wanted to float the idea in case anyone could immediately think of a better option. | 09:24.06 |
| kens, chrisl: It would, of course, be much nicer to assume that all fonts properly gave 'a' value for all the different spaces in unicode, cos that should be at least as accurate as any on-the-fly guess we can make. | 09:27.21 |
paulgardiner | jogux: turning off DOCVIEW_ZOOMSET does seem to work, provided I remove a few asserts (that in any case look wrong anyway), but as you say it's a backwards step. | 09:27.22 |
Robin_Watts | but sadly, we aren't as in control of the fonts as we'd like. | 09:27.34 |
kens | wonders what MS do.... | 09:28.09 |
chrisl | Use their own fonts..... | 09:28.17 |
pedro_mac | canât think of an obvious way to make this more modular just now. What you suggest sounds quite logical and shouldnât introduce any real binding between modules | 09:28.19 |
Robin_Watts | jogux: We have one 'spaces' openfont for the entire system, I believe. | 09:28.31 |
| and the width of the spaces depends on the other fonts that are "fallback"ed through | 09:29.00 |
| (fallen-back? falled-backen? hmm) | 09:29.17 |
| hence the space widths are a property of the set of openfonts, not a single openfont, I think. | 09:29.40 |
| thus Font_Handle is the place to store them. | 09:29.49 |
jogux | robin_watts : oh. hrm. but you can cache the first punctation / ideograph character of your search into the openfont though? | 09:33.03 |
| ie. the slow bit of trying to figure out which character you want actually exists? | 09:33.14 |
| feel free to tell me I'm being stupid if I missed some part of the logic :) | 09:34.00 |
| (my belief is there there are 10s or less of OpenFonts and potentially 100s of Font_Handle) | 09:35.07 |
Robin_Watts | jogux: Not into the spaces openfont, not. | 09:42.05 |
jogux | oh, yes, I was thinking to cache the resultts into the *other* openfont(s) | 09:42.26 |
| intermediate results, even. | 09:42.44 |
Robin_Watts | Possibly, though that's nastier. | 09:45.57 |
jogux | isn't sure quite how slow it'll be tbh. we cache all the width metrics in RAM anyway? | 09:46.58 |
Robin_Watts | yes, looking up char widths can't be that slow. | 09:49.26 |
| and we'll only do it where we actually need them. | 09:49.37 |
| I'm not suggesting doing it on every single font_handle open. | 09:49.49 |
kens | For TTF you just consult the HMTX table | 09:49.51 |
jogux | robin_watts: nod. | 09:50.33 |
paulgardiner | pedro_mac: is there anything we want to be asking the new Good contact relating to the Android app? | 09:55.22 |
jogux | robin_watts : I think I'm essentially happy with your plan | 09:57.54 |
pedro_mac | paulgardiner: I think its primarily how we go about editing - either through sample apps from them or access to other editing apps (probably more difficult). Iâve still had no response from support on the issues with singleTop/singleTask activity model, and whether they have plans to remove the restriction, despite prodding them | 09:57.55 |
| paulgardiner: file transfer works fine, which I think is all the iOS app supports so the editing stuff *could* be deferred in order to get android available (but obviously needs the task model problem sorted) | 09:59.10 |
paulgardiner | I think the iOS app supports editing. | 10:00.00 |
| And several picsel-specific services | 10:00.50 |
| Just telling the new contact that we have had no response to the support requests sounds worth doing. | 10:01.59 |
| You could give me something to send on to him, or probably just as good to email him directly. | 10:02.35 |
pedro_mac | paulgardiner: Iâve reverted to an âoldâ SDK model of having a separate singleTop activity for good authentication and invoking the singleTask PicselViewer activity through a (cached) intent but we still have issues with switching between tasks | 10:03.39 |
| paulgardiner: ok, Iâll email him. | 10:03.49 |
| paulgardiner: the marketplace listing for iOS has edit file, save edited file but not file-transfer or picsel doc service | 10:04.30 |
paulgardiner | Oh! I knew it didn't have file-transfer. That's because SDK 1.4.x had the GDSecureDocs class for passing docs between apps. The file-transfer service is something I added as part of the update to the later SDK | 10:07.05 |
| I'm surprised though that it doesn't mention the picsel services. I think they show up in the console. | 10:07.53 |
pedro_mac | they show as available, but theyâre not associated | 10:08.13 |
paulgardiner | :-) at some stage I should probably try to understand the distinction between associated and available | 10:09.39 |
| pedro_mac: is that the case for editing too? | 10:10.08 |
pedro_mac | nope - edit/save edit are associated. I guess weâll need to add file-transfer once we publish | 10:10.42 |
paulgardiner | yes. Probably the lack of that is why my tests failed. Brad had told me to use com.picsel.smartoffice2gd when testing file transfer. | 10:11.52 |
| I had been intending to point that out to him, but then I didn't want to mention too many thing at once when I couldn't even authenticate the apps. | 10:13.24 |
pedro_mac | grins - joined up thinking is a step too far... | 10:13.49 |
paulgardiner | :-) | 10:14.46 |
| Separate activity for authentication sounds like a good thing to try, but are you saying that required using an older Good SDK? | 10:16.47 |
| I'm probably in a position to help on the Android side now, but the whole single-top single-task business is something I know little about. | 10:17.55 |
pedro_mac | paulgardiner: Iâm using the newest SDK, but I had updated the app using the current âbest practiceâ as it was completely broken at first; Good android apps used to use a separate activity for auth, but recommend using the new state manager (rather than old event system) and implement the auth handling interface in each activity | 10:25.00 |
| paulgardiner: its much nicer, but has the issue that their auth lib doesnât support being launched as singleTask (which we use for the PicselViewer activity to guarantee single instance) | 10:25.55 |
| so⦠Iâve reatianed all the other SDK mods and reverted the auth to a separate singleTop activity, but still see crashes in certain circumstances switching in and out of the app. | 10:26.45 |
| the crashes are actually *worse* with the seaparate-activity model :-/ | 10:27.07 |
| basically weâre seeing alien calls with no PicselContext set up | 10:27.35 |
| its quite possible thereâs a simpler explanation so another pair of eyes on it would be good | 10:28.03 |
| if you have time | 10:28.07 |
paulgardiner | Worth a go at least I guess | 10:28.27 |
pedro_mac | I have the ânewâ auth model of android-sog branch in ~pete/private-repos/sot.git | 10:29.21 |
| the fallback model I havenât pushed yet, but can do that now | 10:29.46 |
paulgardiner | Okay thanks | 10:30.26 |
tor8 | hi chrisl. so, I've looked at the glyph lists now. | 10:49.45 |
chrisl | tor8: oh, yeh, I'd forgotten about that.... | 10:50.16 |
tor8 | I've assembled WGL4, W1G, UFST (from the existing URW fonts and your list of diffs) and MES-2 glyph lists | 10:51.53 |
| and extracted the metrics from the microsoft fonts | 10:52.05 |
| the MES-2 is some official subset of unicode for european languages | 10:52.37 |
| W1G is something that Linotype and Monotype and other font foundries use | 10:53.03 |
chrisl | Ugh, nice to have "a standard" :-( | 10:53.35 |
tor8 | chrisl: yeah... | 10:54.58 |
| your glyph list has several duplicates with the same glyph under different names | 10:55.26 |
| like Dslash and Dcroat | 10:55.32 |
| not sure how you handle that in the ps interpreter, but in the worst case I think some scripting with fontforge can solve that trivially | 10:56.00 |
| in case you want the same glyph with both names | 10:56.12 |
| chrisl: if I make supersets of everything but MES-2 I end up with 748 glyphs, with MES-2 then 1137 glyphs. the current fonts all have 242 glyphs. | 10:56.43 |
| my question now is -- which subset do we ask for? | 10:56.53 |
| we could take one subset and do the intersection with what the microsoft fonts have | 10:57.40 |
| which means we can supply them with the exact metrics we want | 10:57.50 |
chrisl | Well, it really depends on what we want..... my work was all based on having a fully UFST compatible alterative, which I've no clue whether that's still relevant | 10:58.04 |
tor8 | I see two motivations for the fonts -- one as a microsoft compatible font set, and the other as a more general substitute font | 10:59.08 |
| the microsoft fonts all have arabic and hebrew and 2500 glyphs give or take a few hundred | 10:59.34 |
chrisl | I don't see much value beyond what we're required to have for PDF support | 10:59.50 |
tor8 | which is overkill for us I think | 10:59.51 |
| chrisl: yeah, in that case I'd pick the smaller subset | 11:00.56 |
| chrisl: do you have access to the UFST fonts? | 11:01.32 |
| there are roughly 50 glyphs which are in your UFST set but not in the microsoft fonts | 11:01.46 |
| so I can't provide metrics for those | 11:01.58 |
| the arrows, set and math symbols | 11:02.23 |
chrisl | tor8: depends what you means "access" - the microtype fonts aren't "fonts" as such. | 11:02.40 |
tor8 | nothing that you can get fontforge to open up or somehow extract the metrics from? | 11:03.20 |
mattchz | Re: http://bugs.ghostscript.com/show_bug.cgi?id=693224 | 11:03.46 |
chrisl | No, not even close. You can only access them via the UFST | 11:03.47 |
| tor8: ^ | 11:03.52 |
tor8 | urgh | 11:04.01 |
chrisl | That's why I produced the list in the way I did | 11:04.15 |
mattchz | Tor8/Robin_Watts/paulgardiner: I can only partially reproduce this. Iâll bring in my HTC desire tomorrow to test it. | 11:04.40 |
tor8 | http://pastebin.com/NAAAW5Ym that's the list of UFST glyphs that are missing from the microsoft fonts | 11:04.47 |
mattchz | but, the main thing seems to be that mupdf is started as part of the Download apps task. | 11:05.07 |
| i.e. we donât have a separate entry in the task manager for mupdf, because itâs being as a viewer within Downloads | 11:05.31 |
| QuickOffice does the same. | 11:05.41 |
| SmartOffice and Amazon Kindle both open a new task, and the app appears as a separate entry. | 11:06.06 |
| I think either is valid. Was there an intention of how MuPDF should behave? | 11:06.55 |
chrisl | tor8: the font metrics files only contain metrics for a subset of the glyphs in the Microtype fonts...... | 11:07.11 |
mattchz | A side effect of the way things are right now is that it is possible to have multiple documents open at once â one in the MuPDF app âproperâ and others as an activity running in other tasks. | 11:07.35 |
paulgardiner | thinks for second time today that he must get to grips with Android's task vagaries | 11:07.49 |
mattchz | Note that despite the separate UI tasks, all actually run within the same linux process | 11:08.04 |
| Was MuPDF designed that should she safe? And what behaviour do we actually want? | 11:08.22 |
| paulgardiner> paul, yeah, I have to do that every time I revisit one of these issues :( | 11:08.39 |
| itâs not entirely clear how it should work. | 11:08.49 |
| (from a usability pov) | 11:08.58 |
pedro_mac | amen to that | 11:09.00 |
paulgardiner | Robin did some work on allowing multiple simultaneous tasks amd that seemed to work. | 11:09.40 |
tor8 | chrisl: right, so unless you can get the metrics for that list I guess we'll have to live with whatever URW come up with? | 11:09.51 |
mattchz | paul> ah, so thatâs intentional then? | 11:10.31 |
| If so, I might just close this bug. | 11:10.38 |
tor8 | nobody I know knows how android tasks are supposed to work. it's all black magic and never does what you expect... | 11:11.06 |
paulgardiner | I'm not sure we knew what the alternatives were | 11:11.08 |
mattchz | tor8: indeed | 11:11.15 |
chrisl | tor8: I can probably get things like the advance width accurately-ish using a Postscript program | 11:11.18 |
tor8 | ...from a user perspective | 11:11.22 |
| chrisl: if you could get that (just enumerate all the fonts and do it by glyph name) I can merge that in with what I'm assembling | 11:12.00 |
| just the advance width, I don't think anything else really matters | 11:12.27 |
mattchz | paul: there are potentially a few alternatives. But the most obvious one would be just to have a single instance of MuPDF, showing only a single document. And that would run in its own task. | 11:12.32 |
tor8 | the advance width is the crucial bit not to get messed up layouts | 11:12.37 |
paulgardiner | If all Android devices had the words "Anything could happen" written on the back button, things would be fine, | 11:12.39 |
mattchz | I notice sebras reported that bugâ¦maybe he knows more | 11:13.00 |
tor8 | paulgardiner: indeed! | 11:13.00 |
chrisl | tor8: FWIW, the monotype supplied TTF equivalents of the MT fonts ("for screen use") also only contain a subset of the glyphs in the MT fonts - which I think is "enlightening" | 11:13.09 |
mattchz | paul: hehe | 11:13.13 |
| youâll be pleased to know, itâs all changing a bit in Android L! | 11:13.29 |
tor8 | mattchz: I just pinged him on ICQ, but I think he might be away being furious in a meeting. | 11:13.32 |
mattchz | hehe | 11:13.38 |
| thanks | 11:13.41 |
| http://www.androidpolice.com/2014/06/30/android-l-feature-spotlight-the-new-document-centric-multitasking-interface-finally-makes-sense/ | 11:13.57 |
paulgardiner | mattchz: in that case, what would happen if someone tried to open a document in MuPDF when it already had an open document? | 11:14.16 |
mattchz | presuambly itâd close it and open the new one, possibly warning if a save was needed. | 11:14.37 |
chrisl | tor8: I'll see about getting the advance widths when I get a bit of time | 11:14.38 |
mattchz | Iâll see what QO does, ooi | 11:14.44 |
| I think the current behaviour is fine, assuming its intentional. | 11:15.08 |
tor8 | chrisl: I was thinking sending them the list of unicode characters the fonts need, and lists of advance widths for each character for each font would be good enough | 11:15.50 |
mattchz | oh, weird. QO does the same as us too. Iâm sure it was doing something different before. | 11:16.20 |
chrisl | tor8: yeh, that should be enough - we easily provide other specifics if they ask | 11:26.56 |
tor8 | chrisl: okay. when do you expect to get a bit of time for the advance widths? I think henry is itching to get this done. | 11:27.56 |
chrisl | tor8: I'll have it done for tomorrow at the latest | 11:28.57 |
tor8 | chrisl: fab, thanks | 11:29.17 |
| just mail me the lists and I'll massage them into the same format the rest is in and see if there are any incompatibilities | 11:29.40 |
chrisl | Can't be that important, though, given how long it's festered....... | 11:29.48 |
tor8 | well, I'd like to see it happen so I can close some of my bug reports :) | 11:30.05 |
Robin_Watts | mattchz: If you start mupdf as an app, then hit the home key, and back out, then open a file that launches MuPDF you can have 2 instances running. | 11:33.39 |
| That used to confuse MuPDF because we had globals in the lib. | 11:33.59 |
| That's why we now have a 'globals' object. | 11:34.05 |
| I wonder if this bug has been solved now. | 11:34.35 |
| sebras? You awake? | 11:34.39 |
pedro_mac | paulgardiner: Iâve just rebased my current stuff - on the âandroid-sog-gdauthâ branch in my private repo. | 11:35.18 |
paulgardiner | Right. I'll grab that. Thanks. | 11:36.23 |
pedro_mac | should be able to build using the solution - thereâs an android-sot-gd project in the solution on that branch | 11:37.25 |
pedro_mac | suspects it is probably having similar issues to the muPDF multi-instance stuff. Thatâs what Iâd been looking at sorting in the alien. Itâs a bit of a mess based on when it was originally done though (the alien that is) :( | 11:39.12 |
| for now Iâd much rather do the simpler option of getting the existing singleTask activity working with whatever-good-auth-method-works | 11:40.04 |
sebras | Robin_Watts: hi! | 11:49.39 |
| mattchz: Robin_Watts: ok I've read the logs now. | 11:55.24 |
| I'm not sure how android apps ought to behave in these situation, but this was not what I expected. | 11:56.44 |
| so I reported it. | 11:56.48 |
| tor8: btw, did you find a way to improve the parse speed of those destination links? | 11:57.11 |
tor8 | sebras: haven't looked yet, been busy digging with font character sets | 11:57.52 |
Robin_Watts | sebras: Does MuPDF still render the wrong document in step 13? | 11:58.11 |
sebras | tor8: ok, let me know if you want me to report is as a bug proper. | 11:58.20 |
| Robin_Watts: I have my old HTC sensation at home and I can update mupdf later tonight and test again. | 11:58.53 |
| Robin_Watts: in the mean time I can test on my HTC One. | 11:59.02 |
tor8 | sebras: "interesting" /Outlines tree. "/Count -65" | 11:59.50 |
Alex90 | Hi guys, could someone point me to a way of geting the coordinates of an image from a pdf? I have the fz_image and the fz_pixmap but it seams that the left and top coordinate is always 0 (note in the rendered pdf almost all the images start from a diferent position not 0). | 12:00.25 |
tor8 | sebras: I don't think it's a bug, it's just a file with way too much to parse | 12:01.15 |
| it might make sense to parse lazily though? | 12:01.37 |
sebras | Robin_Watts: mattchz: on my HTC One (using a later version of android and HTC Sense) I can no longer do step 7. when I try I immediately go back to the mupdf task/view/whatever that is displaying a.pdf | 12:02.17 |
Robin_Watts | Alex90: You want to get the position/scale etc of each image within a PDF page? | 12:02.41 |
sebras | Robin_Watts: mattchz: could this be a problem in HTC's Download app in that you could restart it even if you were displaying other documents before..?\ | 12:02.49 |
| tor8: strictly we only care about the links on the current page. | 12:03.08 |
Alex90 | i am interested in the position | 12:03.18 |
sebras | tor8: I don't remember if we can find out which ones are on the current page though. | 12:03.31 |
Robin_Watts | Alex90: Sounds to me like you need to use a new fz_device. | 12:03.32 |
Alex90 | i already have the list with the images | 12:03.44 |
Robin_Watts | What list? | 12:03.53 |
tor8 | sebras: it might be slow because that's what we do... we look up stuff in the name tree lazily | 12:04.05 |
Alex90 | i have the fz_page_block and each fz_image_block | 12:04.29 |
tor8 | and the name tree is *gigantic* | 12:04.29 |
Robin_Watts | Alex90: Ah, right. Just a mo then. | 12:04.37 |
sebras | tor8: oh, ok. | 12:04.44 |
| tor8: it was just a dramatic difference in startup time, so I got curious and bisected a bit to figure out what happened. | 12:05.09 |
tor8 | yeah, it used to fail on the first entry and then abort | 12:05.33 |
Robin_Watts | Alex90: OK, so you've run the text extraction device, and that's kept you a list of images too. | 12:06.14 |
| It doesn't keep positions/scales etc though. | 12:06.26 |
tor8 | sebras: we have a function to pre-load (and flatten) the name tree into a single pdf_obj | 12:06.26 |
| but that isn't being called anywhere in the code other than pdfclean | 12:06.36 |
Alex90 | hmm it keeps only the images width and hight? | 12:06.54 |
Robin_Watts | If you look in source/fitz/stext-device.c at the fz_text_fill_image_mask function, for example. | 12:06.55 |
| Alex90: It doesn't even keep the displayed width/height. | 12:07.12 |
| That function is part of the page device interface. | 12:07.26 |
tor8 | sebras: adding any sort of pre-loading and caching becomes problematic now that we allow editing of files | 12:07.31 |
Robin_Watts | When you run the page through the text extraction device, that gets called. | 12:07.37 |
| fz_image *img is the source image | 12:07.50 |
| That clearly has a width/height. | 12:07.56 |
tor8 | though I guess we could have a set of 'preloaded' objects that get invalidated on every pdf editing operation | 12:08.08 |
| the problem is knowing the set of pdf edits, since pdf_obj's aren't immutable | 12:08.27 |
Robin_Watts | conceptually that fills a square from (0,0) to (1,1) and that is scaled onto the page by applying the fz_matrix *ctm. | 12:08.41 |
tor8 | or maybe that's been solved already with various dirty flags etc | 12:08.42 |
Robin_Watts | what that code does is to store img into the list of page blocks. | 12:09.07 |
| What you want to do is to make it store the ctm too. | 12:09.23 |
Alex90 | aha thx will read a bit more about the fz_matrix to see how should i store it or retreive it | 12:11.59 |
rayjj | morning, all | 12:19.25 |
kens | Hi rayjj | 12:19.38 |
Robin_Watts | morning rayjj. Still in a strange timezone? | 12:19.46 |
chrisl | morning rayjj | 12:19.51 |
rayjj | Robin_Watts: yes, still in CDT (St. Louis area) | 12:20.09 |
| two hours closer than CA | 12:20.29 |
| kens: are you replying to Chitra re: fax 2.1 ? | 12:23.32 |
Robin_Watts | rayjj: marcos is back, so presumably he is? | 12:24.28 |
rayjj | it looks like Wipro could (should) become a customer | 12:25.58 |
| since if they need JBIG2 (which isn't clear) they would need to commercial version in order to get the Luratech encoder | 12:26.37 |
pedro_mac | excitement! Just received an email from Good support | 12:28.41 |
| âHi Pete, | 12:28.53 |
| I am getting this case up to our Good Dynamics Engineering and Development teams. | 12:28.55 |
| On a side note, have you been to the Good Community development network forum?â | 12:28.56 |
| now thatâs impressive. issue raised on the 24th, and after a week of unresponsiveness its going to be handed off to engineering. | 12:29.47 |
| I wonder where engineering requests usually go | 12:30.01 |
rayjj | pedro_mac: be careful not to bite your tongue when it's in your cheek ;-) | 12:30.32 |
pedro_mac | :) | 12:30.42 |
henrys | pedro_mac: laurel and hardy err⦠richard and brad | 12:31.38 |
pedro_mac | henrys: :) | 12:31.53 |
| it would be funny, apart from the fact its costing us a lot of time&money | 12:32.19 |
henrys | pedro_mac: very unfortunate we would probably be much further along on the bugs if you and paulgardiner werenât wrapped up in this. Weâre going to try and bill them for the time but that wonât make up for it. | 12:34.11 |
kens | rayjj my email is currently in a black hole (forwarded from GMail into Spamcop whcih has eaten it), so no I'm not trplying to anything right now | 12:34.20 |
rayjj | kens: sorry to hear that | 12:35.20 |
kens | Yeah me too | 12:35.28 |
pedro_mac | henrys: indeed | 12:35.36 |
chrisl | kens: warned you spamcop was spiralling :-( | 12:35.52 |
kens | CESMail claim a 'catastrophic server failure' and 'the needed parts are not available in Atlanta' | 12:36.15 |
| chrisl and I agreed, but I can't get GMail to talk to Eudora any more | 12:36.32 |
chrisl | Odd, I'm okay with tbird..... | 12:36.51 |
kens | Some kind of password/authentication screw up that I can't diagnose | 12:37.15 |
rayjj | kens: I use tbird | 12:37.17 |
kens | I tried Tbird, it wouldn't import my Eudora mail archive | 12:37.42 |
chrisl | I seem to remember having a major faff with gmail when I reset my password | 12:37.44 |
kens | chrisl its something like that yes | 12:37.57 |
rayjj | kens: I recently upgraded to 24.6 (not much difference, thankfully0 | 12:38.06 |
kens | Actually, I could get newly arrived mail sent to another account, but that won't retrieve the stuff that's been blackholed. I'm assuming its somewhere in CESMail unabel to escape | 12:38.54 |
paulgardiner | henrys: no reply to my email to Matt Sturges so far. :-( | 12:39.06 |
rayjj | kens: Tbird's folder format for mail is quite simple. It should be simple enough to write a converter to allow import. ISTR, Eudora was also not too bad about storing emails in an understandable form | 12:40.00 |
kens | rayjj I just used the tbird import and it failed catastrophically (killed tbird) | 12:40.29 |
| Which kind of dented my confidence before I started | 12:40.39 |
rayjj | kens: understandable | 12:40.52 |
chrisl | When I converted Eudora -> tbird, it imported all the mails, but lost the attachments | 12:41.09 |
kens | I wouldn't care about the attachments, but it failed importing the mail | 12:41.32 |
chrisl | Well, my mail archive was probably rather smaller than yours is now! | 12:41.59 |
kens | 15 years worth of records, yes..... | 12:42.14 |
| OK I can forward elsewhere (found the setting in GMail at last), let me set that up and test it | 12:42.55 |
chrisl | kens: you could try using Eudora with gmail again - possibly the password changes will have filtered onto the pop server by now | 12:44.14 |
kens | I may give that a bash | 12:44.29 |
| lets try this first, its probably easier | 12:44.36 |
| Great, now I need to wait for the confirmation email.... | 12:45.27 |
| OK rayjj can you forward me a copy of that mail please ? | 12:46.07 |
mattchz | sebras: excellent, thanks. | 12:48.03 |
| is your HTC Sensation running 2.3 or 4.x? | 12:48.55 |
sebras | mattchz: 2.3 I think.. | 12:49.15 |
| mattchz: probably 2.3.5 IIRc. | 12:49.21 |
mattchz | nods. | 12:49.26 |
| Iâve tested on 4.4 but I couldnât reproduce it. Iâll bring ing my 2.2 HTC device tomorrow. | 12:49.50 |
| Assuming there isnât a problem on HTC Sensation, I think we can close that bug? | 12:50.14 |
| The behaviour is consistent with some other Android apps | 12:51.36 |
| (The lack of a separate task). | 12:52.27 |
sebras | mattchz: as longs as we are happy with the behaviour then we can of course close it. :) | 12:52.50 |
mattchz | If the other parts of the bug are still happening (wrong file opening etc), then Iâd say that it is a bug we need to fix. | 12:52.50 |
Robin_Watts | mattchz: I don't think we care about the lack of separate task (I don't, certainly). Displaying the wrong file is the problem. | 12:53.00 |
mattchz | Robin_Watts: cool. I couldnât get it to display the wrong file in 4.4 (it wasnât possible to open the downloads list without explicitly backing out of the MuPDF view) | 12:53.57 |
| So, I guess we just need to test that bit of thje bug on the Sensation. | 12:54.18 |
kens | chrisl still getting authentication failures with GMail/Eudora | 12:54.20 |
Robin_Watts | mattchz: Hmm. Does MuPDF hide the top bar then? | 12:54.32 |
mattchz | the MuPDF toolbar? nope. | 12:55.22 |
| should it? | 12:55.25 |
Robin_Watts | mattchz: In that case, set a download running. | 12:56.01 |
| then start mupdf and open a PDF. | 12:56.07 |
| Then when the download completes you'll get a notification in the top bar. | 12:56.20 |
chrisl | kens: odd, does it seem to be failing on the p/word or encryption? | 12:56.34 |
Robin_Watts | Drag that down and click on the notification, and it should open mupdf with the PDF file in. | 12:56.38 |
| (the download should be a PDF file, obviously!) | 12:56.46 |
kens | chrisl I'm assuming password since it says its an authentciation failure | 12:57.01 |
Robin_Watts | That should start a second mupdf without you explicitly having backed out of the first. | 12:57.13 |
chrisl | kens: of course, I'm using imap, so the settings may not be the same | 12:57.32 |
Robin_Watts | Then when you hit 'backup' from that, you should return to the original file in mupdf. | 12:57.39 |
kens | I tried IMAP and that failed also | 12:57.42 |
chrisl | security ssl/tls and "normal password" authentication | 12:58.22 |
kens | What's hte mailbox prefix ? | 12:58.38 |
mattchz | Robin> oh, possibly. That shouldnât be a problem though, should it? | 12:58.52 |
jogux | IMAP authentication is a PITA :( | 12:59.04 |
kens | Well POP is failing so.... | 12:59.15 |
| THough I have no cinfidence in IMAP eikther | 12:59.25 |
chrisl | I don't know where the prefix is set | 12:59.32 |
kens | :-) | 12:59.36 |
| I'#ll google it | 12:59.41 |
Robin_Watts | mattchz: I would hope it should work. It's the closest thing I can think of to do to be able to replicate the conditions of the original problem. | 12:59.41 |
sebras | Robin_Watts: I'll try that on my phone. | 12:59.55 |
| mattchz: Robin_Watts: ok so I opened a pdf in mupdf, then switched task to the browser and downloaded a file. when it was finished I opted to open it in mupdf, and then it showed the downloaded pdf. then I click back and it showed the first pdf that I opened. | 13:02.15 |
| mattchz: Robin_Watts: not the browser as you might have expected. | 13:02.26 |
| but as I mentioned before I don't know how android stuff is supposed to work in this area. :) | 13:02.46 |
mattchz | sebras: nobody does :) | 13:02.53 |
Robin_Watts | sebras: I think that's how android works. | 13:02.56 |
| certainly I am not aware of us having any input into what tasks swaps to what :) | 13:03.10 |
mattchz | robin: there are lots of complicated flags that interact with each other in confusing ways. | 13:03.35 |
Robin_Watts | is now forced to imagine mattchz as kim jong il. | 13:03.36 |
mattchz | ? | 13:03.47 |
Robin_Watts | "It'll be like 9/11 times ...." | 13:04.15 |
| Team America? Never mind :) | 13:04.24 |
mattchz | what did I say that made you think of that ? ;) | 13:04.45 |
| Good film | 13:04.53 |
| sebras: were you in the browser when you decided to open the downlaoded file? | 13:05.25 |
Robin_Watts | mattchz: There is a running joke through the film, where they say "it'll be like 9/11 times 100", "9/11 times 100? That'd be..." "yes, 91100" | 13:06.19 |
mattchz | yes, in fact the same happens for me. | 13:06.26 |
| sebras: Iâm surpried it didnât return to the browser too. | 13:06.35 |
chrisl | rayjj: I'm not sure it's a good idea to commit to doing JBIG2 in TIFF output..... we certainly can't leverage libtiff for it | 13:06.42 |
mattchz | sebras: but Iâm not sure if itâs âwrongâ or not :( | 13:06.58 |
| ah right :) | 13:07.08 |
Robin_Watts | and it culminates with KJI saying "It'll be 9/11 times 1234" "that'll be... I don't even know what that'll be". and KJI says "nobody does!" | 13:07.08 |
mattchz | ah :) | 13:07.17 |
henrys | chrisl: he left I was just thinking that | 13:07.20 |
| bbiab | 13:07.35 |
mattchz | did you see this btw? | 13:07.51 |
| https://uk.yahoo.com/movies/five-films-besides-the-interview-that-north-korea-has-90067784542.html | 13:07.52 |
| robin_watts/sebras: Iâve read this document many times: http://developer.android.com/guide/components/tasks-and-back-stack.html | 13:10.20 |
| and every time Iâve come away utterly confused. | 13:10.27 |
| how we are launched also depends on how the âsourceâ app constructs the Intent. | 13:11.15 |
jogux | mattchz: so does Android L properly fix this? | 13:11.58 |
mattchz | itâs possibly that the downloads app is constructing the intent differently to the browser app, and hence why in the downloads case we appear as part of the downloads task, but in the browser we appear as part of the MuPDF task. | 13:12.00 |
| jogux> fiik :) | 13:12.07 |
jogux | did mupdf work okay on Android L btw? | 13:12.10 |
mattchz | yes. although I didnât check this task stack behaviour. I think we need to use special APIs to do the new Android L multiple docs in task switcher behaviour. | 13:12.54 |
jogux | ah :-S | 13:13.14 |
mattchz | I only expect it to be even mroe confusing ;) | 13:13.27 |
jogux | possibly that should go into bugzilla as an enhancement-ish thing if user's will expect the new behaviour :-S | 13:13.30 |
mattchz | yeh | 13:13.44 |
| +a | 13:13.46 |
pedro_mac | suspects users are too confused to know whatâs going on | 13:13.52 |
mattchz | well, if we canât understand how itâs supposed to work, I suspect users have no chance ;) | 13:14.31 |
jogux | pedro_mac : I was, perhaps mistakenly, presuming the new Android L behaviour might be clearer :-) | 13:14.55 |
mattchz | the new behaviour in L though makes these multiple tasks visible to the user, so they would probably expect it to work. | 13:15.07 |
| I might be wrong about the new API. | 13:15.16 |
rayjj | chrisl: I mentioned that we already have JBIG2, and the we could do JBIG for TIFF (tag 34661) | 13:17.23 |
kens | rayjj looks like the WiPro mail may be referring to Bug #693594, sinceit has exactly the same title: | 13:17.24 |
| http://bugs.ghostscript.com/show_bug.cgi?id=693594 | 13:17.24 |
rayjj | since that's built in libtiff :-) | 13:17.31 |
chrisl | rayjj: I'd worry that was too fine a distinction for these folks | 13:18.33 |
rayjj | kens: I didn't look at our tracker. Thanks. But in any case, it sounds like somebody has a bug in another tracker, which implies a product. | 13:19.20 |
| chrisl: from the bug, it looks like you (also) confused JBIG2 (which isn't in libtiff) and JBIG (which is all that fax supports) | 13:20.05 |
chrisl | rayjj: yep, looks like I did..... | 13:21.03 |
rayjj | what isn't clear from the bug (Avadhut) is whether or not this is for HP's commercial product, which I didn't know sent faxes | 13:21.43 |
chrisl | There's a at least one name in the CC list of that mail that looks very like one from cust 130 - may not mean anything, though | 13:22.26 |
rayjj | but if Wipro gets back to us and says they already are a commercial customer, we can clarify that it is HP's product | 13:22.33 |
omni_ | hi everyone. i was brought here because i visited and used mupdf. it was stated in the website the developers are hanging out here. i just want to know if its possible to load a NSData instead of the file path of the PDF in Mudoc? btw, mupdf is really great. | 13:22.47 |
chrisl | rayjj: I must admit, I thought JBIG died years ago! | 13:23.13 |
mattchz | omni_: the core library supports loading from a data array, Iâm not sure if the iOS layer does, but if not it should be easy to add | 13:23.43 |
rayjj | omni_: paulgardiner is the iOS expert here (afaik, NSData is an Apple thingy) | 13:24.02 |
Robin_Watts | omni_: The core library is all written in terms of fz_stream's. | 13:24.09 |
| That's a generic class that can be implemented on top of pretty much anything. | 13:24.27 |
| so, while I am not familiar with what an NSData is, the answer is almost certainly, "yes, if you're prepared to do a bit of coding" | 13:25.01 |
jogux | an NSData trivally converts to a char * or similar + a length. | 13:25.29 |
Robin_Watts | jogux: Ah, well, then we already have a fz_stream that works from a memory buffer, so that should be easy. | 13:26.38 |
mattchz | there is code which does a similar thing for Android, except it takes a java array of bytes. | 13:26.47 |
omni_ | Thanks everyone for the reply. | 13:27.05 |
| i think im on the right track when you mentioned about the fz_stream | 13:27.23 |
Robin_Watts | lunches, back in a bit. | 13:27.30 |
rayjj | jogux: Are you also an apple developer ? (for mupdf, paulgardiner was stuck with the app) | 13:27.49 |
omni_ | i've been playing with it since morning and trying to figure out the correct way to convert NSData to unsigned char | 13:28.00 |
| I was just hoping there is a method already which i can use and just pass the actual NSdata | 13:28.45 |
mattchz | omni_: [NSData bytes] should do it. | 13:28.49 |
jogux | rayjj: I'm primarily an iOS developer, but not for mupdf. mattchz is doing iOS + Android mupdf. | 13:28.52 |
rayjj | that's why I like plain ol' C -- there's lots of ways to 'convert' (just pass pointers around :-) ) | 13:29.02 |
mattchz | If you have: | 13:29.14 |
tor8 | omni_: [NSData bytes] and [NSData length] | 13:29.15 |
mattchz | NSData *myData = â¦.. ; | 13:29.23 |
| then you can do | 13:29.29 |
| const unsigned char *bytes = [myData bytes] | 13:29.44 |
omni_ | im also trying this (Byte *)[dd bytes] | 13:29.46 |
tor8 | stm = fz_open_memory(ctx, [NSData bytes], [NSData length]); | 13:30.02 |
omni_ | thanks mattchz and tor8 | 13:30.22 |
rayjj | wow! help from every corner :-) | 13:30.23 |
tor8 | doc = fz_open_document_with_stream(ctx, "pdf", stm) | 13:30.40 |
mattchz | omni_: np | 13:31.10 |
paulgardiner | tor8: typo? did you mean myData not NSData? | 13:31.16 |
tor8 | paulgardiner: yeah, make that myData not NSData | 13:31.25 |
omni_ | thanks guy! i will implement all those things. this is a big help. | 13:31.28 |
chrisl | henrys: it's just dawned on me that it's tuesday, and I have an appointment at 3pm, so I'll miss the IRC meeting..... | 13:32.51 |
mattchz | So, what process do we need to go through to release MuPDF for iOS? | 13:37.26 |
| is there a testing process? | 13:37.36 |
henrys | chrisl: okay main thing was to double check whatever tor8 comes up with for the fonts or at least nod. | 13:39.12 |
chrisl | henrys: tor8 and I spoke earlier - I'm getting some extra info together for him (should be done for tomorrow), then we can review where we stand. | 13:40.08 |
henrys | chrisl: great | 13:40.23 |
chrisl | henrys: tor8 might want some feedback from you about the character set(s) we want to support, though | 13:40.48 |
henrys | chrisl: right, Iâd like to be careful we cover everything in the PCL set since I might use these new fonts in pcl - they are providing t1 and tt | 13:41.52 |
chrisl | henrys: I'm not sure sure if the PCL set even crossed tor8's radar.... | 13:42.38 |
henrys | Iâm sure it crossed nobodyâs radar but mine ;-) | 13:43.07 |
| which is probably best | 13:43.17 |
tor8 | chrisl: henrys: is there a list somewhere? | 13:43.19 |
| all I need is a a list of the glyphs (by ps glyph name or unicode value) and preferably advance widths | 13:44.01 |
henrys | tor8: okay Iâll prepare it today. | 13:44.23 |
chrisl | tor8: actually, if we get all the glyphs from my list from the UFST/MT fonts, we should be good for PCL | 13:44.26 |
henrys | chrisl: if you built in a shared language configuration with the UFST | 13:45.16 |
chrisl | henrys: I was using the pcl5ps3 FCO | 13:45.37 |
henrys | chrisl, tor8 : okay so we are good | 13:45.57 |
chrisl | But that doesn't actually matter in terms of glyph coverage, only font coverage | 13:46.00 |
henrys | chrisl: I also think the ufst will synthysize pcl fonts - unstyled characters like boxes and such and you may not have enumerated those glyphs. | 13:47.19 |
jogux | mattchz : I believe testing is essentially all adhoc by as many of us on this channel as you can persuade to try it :-) | 13:52.17 |
mattchz | fair enough :) | 13:52.37 |
| please try it ;-) | 13:52.42 |
paulgardiner | which usually means by no one! | 13:53.11 |
mattchz | ship it :) | 13:53.20 |
jogux | paulgardiner: who actually makes the "ship it" decision? | 13:53.28 |
| paulgardiner: I wonder if we should do a release of iOS SO+ too (maybe a week or two after SO2, in case there are issues?) | 13:54.30 |
paulgardiner | Well Miles has said something like "the worse that could happen is it's a disaster and we have to revert back to the old version", which I took as a sort of go ahead | 13:54.38 |
| So far the few people who have tested versions have found problems which we've felt were serious enough to need adressing. | 13:56.07 |
jogux | paulgardiner: cool | 13:56.19 |
| paulgardiner: so. SO2 bug, testflight version, iOS 8, ipad: open getting started, save->open pdf in, transfer to ibooks or mupdf, the artifex logo on the last page gets pixelated to hell | 13:57.03 |
paulgardiner | Okay thanks. Probably not worth holding the release for. Presumably it can't be a regression because the currently-on-sale version would have no images in pdf conversions. | 13:59.40 |
jogux | paulgardiner: urm. I tried it in the appstore build and the picsel logo appeared fine | 14:01.05 |
paulgardiner | Really! I thought Robin_Watts added image support to pdf out | 14:01.30 |
| Confused. | 14:02.01 |
jogux | did Robin do rotated images maybe | 14:02.23 |
| ? | 14:02.24 |
| I'm not sure if it's a bug or not, but when you use "open in mupdf" in iOS, you can end up with multiple documents opens, stacked in the navigation stack. | 14:05.43 |
paulgardiner | Maybe we don't use pdfexport on iOS | 14:05.46 |
jogux | paulgardiner: I'm sure we do. I don't know how else we'd do it :) | 14:06.29 |
| paulgardiner: ah, it copes with ImageURLs before, now it copes with Images too. Which document formates/features use Images rather than ImageURLs I'm not sure. | 14:08.09 |
henrys | kens:feel free to miss the meeting and work on your email - I donât think there is much gs stuff to talk about. | 14:08.46 |
jogux | I thought one of their fixes for the patent issue was to stop using 'Image' but maybe I misremember. | 14:08.48 |
| is anything new in mupdf 1.5 that we should explicitly check? | 14:09.34 |
tor8 | mattchz: the file type registration for CBZ doesn't seem to work on iOS. XPS works though. want me to open a bug report for it? | 14:11.31 |
jogux | curious. I have one particular pdf that iOS mupdf won't let me draw annotations on. | 14:12.41 |
| are there types of pdf mupdf can't annotate? | 14:12.57 |
kens2 | henrys my email is fine at the moment, its forwarding to another account, its only the overnight mail which is AWOL, I just have to hope it'll show up eventually | 14:13.08 |
tor8 | jogux: it might have the permissions set to disallow annotations, I have a vague recollection of paul adding something to respect that flag | 14:13.14 |
paulgardiner | I have no such recollection, but I guess I may have. | 14:14.24 |
jogux | :-) | 14:14.29 |
| bah. it crashes when I try to delete a file :( | 14:14.36 |
rayjj | henrys: should we (I) do a binary for Sriram of HCL ? (for customer 534). It's easy enough, but I'd sooner push back and tell them how to build it themselves | 14:14.58 |
jogux | wonders if we should stick crashlytics into iOS mupdf | 14:15.00 |
rayjj | henrys: I was just about to send the email to them telling how to build it themselves, but thought I should ask if that seems reasonable | 14:15.43 |
henrys | tor8, mattchz : I see lots of stuff is going back to mupdf core? Have we scanned all the bugs for reassignment. | 14:15.52 |
Robin_Watts | rayjj: So a customer is getting someone else to do work on gs for them, and that company is not capable of doing builds? | 14:16.13 |
| what exactly is that company doing then? | 14:16.24 |
rayjj | Robin_Watts: typical of consultants | 14:16.45 |
henrys | rayjj: I like to leave that with marcosw | 14:17.05 |
paulgardiner | jogux: testing on Android, I get the same result as you from "getting started", but with another file I created and added an image to, it seemed to work fine. | 14:17.30 |
rayjj | henrys: it's been since 6/23 and marcosw hasn't replied yet. | 14:17.39 |
jogux | paulgardiner: :-S | 14:17.49 |
Robin_Watts | paulgardiner: open it as a bug, I'll look into it. | 14:17.50 |
kens2 | marcosw may have missed it, it came in while he was on vacation ? | 14:17.52 |
rayjj | I noticed it because the associated Bug 695298 is still open and I'd like to close it | 14:18.02 |
paulgardiner | Not a blocker on release thought, right? | 14:18.23 |
| s/thought/though/ | 14:18.35 |
rayjj | henrys: marcosw was back on 6/20 AFAIK | 14:18.50 |
Robin_Watts | paulgardiner: no. | 14:18.53 |
paulgardiner | On the other hand, it shows that we do need to test. | 14:18.54 |
kens2 | rayjj I'm happy to tell them *how* to build GS, I'd rather we didn't do a 'one off' build for them, I'msure you feel the same :-) | 14:19.01 |
Robin_Watts | paulgardiner: before my work it would have been exported as missing, so it's clearly an improvement. | 14:19.15 |
henrys | rayjj, kens2: assing the bug to marcosw | 14:19.23 |
kens2 | ROFL :-) | 14:19.34 |
rayjj | kens2: yes, that's the email I have drafted | 14:19.37 |
paulgardiner | Robin_Watts: jogux claimed that it worked for him on the app store version. | 14:19.49 |
rayjj | henrys: I'll just forward my reply to marcos / support for him then | 14:20.06 |
Robin_Watts | oh, well, that's odd then. I didn't change anything... | 14:20.11 |
jogux | henrys : mattchz is just making tea btw, he'll be back in a few minutes | 14:20.19 |
| robin_watts : the changes is to getting-started.pdf | 14:20.26 |
Robin_Watts | The app store version presumably didn't have an artifex logo? | 14:20.31 |
jogux | it now have an artifex logo instead of a picsel one | 14:20.33 |
kens2 | chat logs finally caught up to where my network hiccuped...... | 14:20.34 |
jogux | yes | 14:20.34 |
Robin_Watts | right. | 14:20.38 |
jogux | I'm presuming the new pptx breaks in the old build, but got distracted by mupdf crashing :) | 14:20.58 |
kens2 | mvrhel_laptop : you there ? | 14:21.03 |
henrys | rayjj: did you see that crash gigs reported, I assigned it to you, looked worrisome | 14:21.20 |
kens2 | saw it but didn't look at it yet | 14:21.38 |
rayjj | henrys: when was that ? | 14:22.04 |
paulgardiner | yeah, that's all we need to confirm - that the new "getting-started" breaks in the old build. | 14:22.31 |
kens2 | bug #695333 ? | 14:22.38 |
paulgardiner | seems likely | 14:22.41 |
kens2 | rayjj: | 14:22.59 |
| http://bugs.ghostscript.com/show_bug.cgi?id=695333 | 14:22.59 |
rayjj | kens2: thanks. | 14:23.04 |
| henrys: yes I did see that, but haven't dug into it | 14:23.18 |
mvrhel_laptop | kens2: yes I am | 14:23.28 |
henrys | rayjj: youâre still in STL right? | 14:23.53 |
kens2 | mvrhel_laptop : I had a first pass at the V2 ICC profile, and it crashed, can we talk it through after the meetings ? | 14:24.07 |
mvrhel_laptop | kens2: yes | 14:24.20 |
kens2 | OK thanks | 14:24.25 |
rayjj | kens2: BTW, just mentioning Bug ###### is enough for me. Chatzilla knows to open bugs.ghostscript.com with just that and takes me right to the bug | 14:24.32 |
kens2 | rayjj yes, but URL is nice too | 14:24.48 |
henrys | coffees before meeting in 5 minutes | 14:25.16 |
rayjj | henrys: yes, I'm officially on vacation until 7/7 (next Monday) | 14:25.17 |
kens2 | thinks rayjj shouldn't attend meetings whil eon vacation | 14:25.32 |
rayjj | henrys: but I wake up earlier than my kids, so spend time doing something I enjoy | 14:25.50 |
kens2 | Which reminds me. henrys I'll miss the next 2 meetings | 14:25.56 |
rayjj | not that doing stuff with the family isn't also enjoyable, but watching them sleep isn't that entertaining | 14:26.37 |
henrys | rayjj: I guess your kids have done the arch trip. | 14:28.41 |
rayjj | kens2: yes, we did the arch and the City Museum | 14:29.50 |
kens2 | thinks that was henrys you meant rayjj :-) | 14:30.05 |
rayjj | henrys: kens2: yes, oops | 14:30.26 |
henrys | we are getting ready to order fonts from URW, if anyone wants to weigh in, see the glyph list and make suggestions let me know. | 14:32.01 |
mattchz | jogux> multiple docs things, hrm, donât know. | 14:32.37 |
| I guess if it doesnât crash, it seems âreasonableâ, if perhaps confuing for the user. | 14:32.56 |
| tor8: yes please, that would be good (CBZ bug) | 14:33.11 |
henrys | kens2, mvrhel_laptop :any gs stuff? | 14:33.42 |
mattchz | jogux> annotations etc. | 14:33.50 |
jogux | mattchz: it perhaps has the potential to eat memory | 14:33.51 |
kens2 | Not really, except that I need to talk to Michael re the V2 ICC profile code | 14:33.59 |
henrys | more profile problems I noticed | 14:34.05 |
mvrhel_laptop | not really other than that I have the icc stuff with kens2 | 14:34.09 |
mattchz | You canât annotate encrypted documents anymore (because we canât save them). | 14:34.11 |
| You canât annotate non-PDF files either. | 14:34.16 |
mvrhel_laptop | and I have a question from a customer I see in support | 14:34.30 |
jogux | I don't think it's encrypted, or at least I don't need a password to open it. | 14:34.33 |
kens2 | henrys you mean the bug reports ? I just fixed that one, and it wasn't really profiles, it was the new colour code | 14:34.36 |
| I'm glad Marcos is back, the support list has been much quieter this last week | 14:34.58 |
henrys | kens2: no I just saw you said here it crashed | 14:35.01 |
kens2 | henrys oh yes, but this is all new stuff, first time I've had a chance to try it out. Likely I'm doing something silly | 14:35.25 |
henrys | kens2: Iâm sure heâs talking to the customers | 14:35.27 |
| ;-) | 14:35.33 |
mattchz | jogux> mm | 14:35.40 |
kens2 | I only just finished the code changes a few minutes back | 14:35.46 |
mvrhel_laptop | have any windows users been using gsview? | 14:36.25 |
kens2 | I have, but not extensively | 14:36.39 |
mvrhel_laptop | I have it set up as my default viewer for pdf, ps, xps | 14:36.45 |
kens2 | I need to use Acrobat a lot for comparisons | 14:36.51 |
mvrhel_laptop | just to excercise it | 14:36.57 |
| maybe set it on your wife's computer :) | 14:37.18 |
henrys | mattchz: I was asking earlier how much mupdf stuff is getting reassigned back to core - it might make sense to go through all the bugs at once looking for core problems | 14:37.27 |
kens2 | doens't thnk so..... | 14:37.31 |
mvrhel_laptop | even I debated that one and decided no | 14:37.54 |
henrys | mvrhel_laptop: I need to reinstall the later gsview Iâll do it right after the meeting. | 14:38.16 |
kens2 | Domestic Hell desk duties are bad enough as-is..... | 14:38.19 |
Robin_Watts | henrys: it might make sense to unleash mattchz on some of the core ones. | 14:38.36 |
marcosw1 | mvrhel_laptop: did you see the email from Zoltan asking about Type 103 images and transparency? It may have been put in your spam folder. | 14:38.42 |
Robin_Watts | Or, if not that, then to let him loose on the JNI stuff. | 14:38.55 |
kens2 | It went to my spam folder, don't know why | 14:39.05 |
mvrhel_laptop | marcosw1: yes I see it. I will respond to him today. | 14:39.26 |
mattchz | henrys: yes, I went through them all and bounced stuff back to core | 14:39.38 |
| I think all the remaining ones are android/ios specific | 14:39.47 |
| did I miss anyoneâs comments btw? I think that was everything (reading the scrollback) | 14:40.31 |
henrys | Robin_Watts: I was more leaning towards tor8 jumping on those and mattchz doing SOT - Miles really has his hopes up for September ⦠| 14:40.32 |
| mattchz: okay great | 14:40.59 |
marcosw1 | mvrhel_laptop: thx | 14:41.19 |
mattchz | jogux> crashlytics. That sounds a great idea, but are there GPL issues? | 14:41.25 |
Robin_Watts | mattchz: Our appstore apps are not GPL licensed :) | 14:42.01 |
mattchz | ah, right. | 14:42.17 |
henrys | marcosw1: we wanted to disable CLUSTER_UNTESTED and any other keyword that are similar | 14:42.20 |
mattchz | I thought they were AGPL? | 14:42.32 |
marcosw1 | support received an email with the subject: FIRST BANK OF NIGERIA PLC. Are they not even trying to fool us anymore? | 14:42.32 |
jogux | robin_watts : paulgardiner : I've confirmed that the new artifex getting started is broken in the current appstore build. so sort of not a regression. | 14:42.50 |
Robin_Watts | jogux: cool. so it should be opened as a bug for me to look at. | 14:43.12 |
paulgardiner | jogux: good | 14:43.13 |
marcosw1 | henrys: not sure what you mean. CLUSTER_UNTESTED is something you can put in a commit message. disabling it would involve sending out an email to everyone telling them to stop using. | 14:43.30 |
| it | 14:43.35 |
Robin_Watts | mattchz: We own our own apps. We release them under the GNU APGL, and also under a commercial license. | 14:43.47 |
rayjj | henrys: I've been using gsview. | 14:43.50 |
henrys | marcosw1: we donât want it to be recognized | 14:43.52 |
rayjj | Is it ready to start opening bugs, or should I wait ? | 14:44.13 |
mattchz | robin_watts: cool. Can we go with the crashlytics thing then? itâs proved to be very helpful in other projects | 14:44.19 |
Robin_Watts | Other people need to stick to the AGPL. We don't need to :) | 14:44.30 |
mattchz | :-) | 14:44.36 |
Robin_Watts | mattchz: I don't really know what's involved, but I can't see why not. | 14:44.48 |
marcosw1 | henrys: done. | 14:45.16 |
mattchz | I guess it might cause issues, as weâd need to build from a slightly different tree that we wouldnât want to make public (probably) | 14:45.20 |
jogux | henrys : do you want to keep mattchz beyond the mupdf mobile stuff being sorted then? It wasn't too clear how long you wanted him for :) | 14:45.47 |
Robin_Watts | mattchz: yeah, I was just about to ask if it would require the build to be bent at all. | 14:46.02 |
rayjj | will adding crashlytics bloat it much ? | 14:46.20 |
Robin_Watts | If it's just a build option, then that's fine. | 14:46.41 |
mvrhel_laptop | so rayjj: is Zoltan thinking that he won't have to deal with softmask images? I guess I am a bit confused by his question | 14:46.45 |
mattchz | robin> you need an additional framework, some hooks in the postlink stage and a single-liner to initialise Crashlytics with our API key. | 14:47.02 |
| rayjj: itâs quite lean | 14:47.09 |
henrys | jogux: stay with mobile for now. I think weâd reprioritize tor8âs list before we put mattchz on core mupdf stuff | 14:47.16 |
rayjj | mvrhel_laptop: I haven | 14:47.21 |
| mvrhel_laptop: I haven't read the email in detail. And I certainly have no idea what he's thinking :-) | 14:47.58 |
Robin_Watts | mattchz: The stuff that needs to be added to the git tree is relatively small, right? | 14:48.10 |
mvrhel_laptop | ok. well I will reply the best I can | 14:48.10 |
henrys | anything else for gs or mupdf? | 14:48.19 |
Robin_Watts | Can we have it as a build option? | 14:48.21 |
jogux | henrys : sure. would still be useful to know the "how long" anwer though as we have other customers asking us to plan in work for them :-) | 14:48.23 |
mattchz | robin_watts: yes. not sure if itâs stuff we can make public or not. | 14:48.25 |
rayjj | mvrhel_laptop: should I start opening bug reports against gsview, or do you want me to hold off ? | 14:48.54 |
mvrhel_laptop | rayjj: that would be great | 14:49.05 |
Robin_Watts | mattchz: If we can have a build option, that controls an include, and all the non-public stuff is in that include file, we should be OK. | 14:49.06 |
mvrhel_laptop | do we have it as an option in bugzilla? | 14:49.19 |
Robin_Watts | Then we can check in a 'blank' include file. | 14:49.21 |
mattchz | some of the stuff will be in the project file. | 14:49.24 |
mvrhel_laptop | marcows: can you add it? | 14:49.34 |
mattchz | so, possibly not. | 14:49.35 |
henrys | jogux: if mobile finishes he can have SOT hours. I donât have any sense how much mobile is left to do. | 14:49.35 |
Robin_Watts | mattchz: Ah. | 14:49.38 |
mvrhel_laptop | that is gsview as a option in bugzilla | 14:49.44 |
| marcosw^^ | 14:49.51 |
| to many conversations... | 14:49.58 |
Robin_Watts | henrys, jogux: It sounds like we'd like to commit to keeping mattchz busy until at least september? | 14:50.22 |
mattchz | http://try.crashlytics.com/sdk/ | 14:50.22 |
| "On average, Crashlytics adds less than the size of a single image to the weight of your application." | 14:50.29 |
henrys | jogux: yes we can commit to september at current hours. Does that help? | 14:50.59 |
jogux | henrys : that helps immensenly, thanks :) | 14:51.17 |
mattchz | it doesnât say if that image is a RAW 50Mp image though :) | 14:51.24 |
| henrys> cool, thanks! | 14:51.34 |
henrys | jogux: assuming heâs available to do SOT - | 14:51.36 |
| some folks donât seem to like that stuff ;-) | 14:52.00 |
jogux | henrys : yeah, we'll just have to up our order of psychotropics | 14:52.08 |
pedro_mac | looks around | 14:52.10 |
mattchz | Iâll do anything for love, but I wonât do that | 14:52.20 |
| :) | 14:52.21 |
| No, just kidding, itâs fine | 14:52.25 |
paulgardiner | :-) | 14:52.31 |
henrys | pedro_mac, paulgardiner : I was expecting matt to kick engineering. I didnât necessarily expect a response. | 14:53.13 |
paulgardiner | It would have been a much shorter song if it had been about working on SOT | 14:53.16 |
jogux | henrys : I think there was a question with mupdf if any of the enhancement stuff should be tackled (there's some general enhancements, plus stuff to work with new features in android l / ios8 ) | 14:53.35 |
mattchz | if thereâs a choice, I would of course rather do MuPDF ;) | 14:53.52 |
Robin_Watts | mattchz: I think that's a song on meatloafs *next* album, | 14:54.22 |
pedro_mac | henrys: to be honest, any kind of kick will help. It really feels like a few hours with some knowledgeable engineer(s) would get us through all the issues we currently have between us and releasing | 14:54.25 |
| henrys: have we had anything back on Veracode ? | 14:54.38 |
henrys | pedro_mac: nothing ugh | 14:54.57 |
| pedro_mac: but I thought you notice a little activity after the email to matt or you think it was unrelated | 14:55.49 |
pedro_mac | henrys: Iâve done all the testing I can on android preparing for veracode (whcih is mostly handling policy changes, revoking, and the supported services like file-transfer) and they seemed ok (in between crashes swithcing in and out of the app that isâ¦) | 14:56.09 |
henrys | jogux: I would like to do the enhancements but I feel it would be going against what miles wants generally. | 14:57.01 |
pedro_mac | henrys: it might be related, and good to get at least some response | 14:57.06 |
paulgardiner | henrys: Hmmm. MattS may have assumed the questions I sent to him had already been sent to their support department. | 14:58.57 |
| Well, hopefully he'll send them on to an engineer. | 14:59.27 |
henrys | pedro_mac: so maybe paulgardiner should follow up with a response request and ask what became of his original Iâll wait until end of business, if I donât see anything Miles and I will get him on the phone. | 14:59.48 |
jogux | henrys : ah, right, ok. | 14:59.51 |
mvrhel_laptop | henrys: I think I am good as far as SOT right now. jogux and Robin_Watts talked me through a rough idea of what I need to do. So far the code as made some sense. In some respects the EDR/display list of SOT may be easier to work with than the clist in gs. Of course I am not going to be doing anything dealing with editing files with what I am doing. | 14:59.53 |
| and I am working in a very compartmental part of the code (charts) | 15:00.32 |
pedro_mac | henrys: cool, thanks. | 15:00.56 |
henrys | mvrhel_laptop, jogux : one of the things I was hoping to accomplish is getting anything mvrhel_laptop trips over documented. How does a new person dive into this stuff type of thing. | 15:00.57 |
mvrhel_laptop | and working with ooxml makes things easier too | 15:01.04 |
| henrys: I will keep some notes and we can update the twiki with what I have found to be asier | 15:01.38 |
henrys | but Iâm not sure if that is practical, it may just be a matter of jumping in and asking questions | 15:01.40 |
pedro_mac | It may be useful for us to just start with a âcookbookâ of issues/resolutions/comments on the wiki first then sanitise it later | 15:02.01 |
paulgardiner | henrys: sorry didn't follow. Who am I trying to ilicit a response from? | 15:02.07 |
henrys | paulgardiner: matt - a followup to your last email requesting a response to it and veracode feedback. | 15:02.38 |
paulgardiner | Ah. Okay got you | 15:02.54 |
Robin_Watts | mvrhel_laptop: possibly we should make a list of every question you had on a twiki page? | 15:03.03 |
| and then we can put the answers in there too? | 15:03.16 |
| Maybe that will turn into something useful? | 15:03.24 |
henrys | paulgardiner: if nothing comes back by this afternoon miles and I will call him | 15:03.36 |
| (my time) | 15:03.44 |
pedro_mac | it should certainly give us a list of areas where the existing docs need expanded | 15:03.45 |
Robin_Watts | I can't think of any questions/answers so far that that would likely help with... | 15:03.51 |
mvrhel_laptop | Robin_Watts: perhaps. I think just a few notes/comments about what I see in the code and what I did not know starting out and what I now know after digging through the code can be useful | 15:04.31 |
jogux | is mupdf meant to require ios 6.1? | 15:04.47 |
| hm, guess so, current appstore version does | 15:05.22 |
Robin_Watts | Last time I tried, it worked on my ipod touch (5.1.1) | 15:05.22 |
henrys | jogux: what was the release policy at picsel? I guess you know we schedule mupdf and gs 2x a year, we havenât completely settled on an SOT policy yet. | 15:05.25 |
mvrhel_laptop | more like use this VS tool for looking at ooxml files. I did not know that the output text dump had the EDR and the display list information in it | 15:05.36 |
| simple things like that | 15:05.47 |
Robin_Watts | henrys: I think the policy at picsel was to release to customers as and when required. | 15:06.11 |
henrys | other SOT issues, anything I should convey to Miles? | 15:06.12 |
Robin_Watts | mvrhel_laptop: That ooxml tool is VERY useful. | 15:06.24 |
jogux | henrys : I'm not sure there was a consistent policy for the appstore releases. nominally their policy was the appstore build was a few months behind in terms of features vs the OEM customers. | 15:06.53 |
mvrhel_laptop | Robin_Watts: yes I would still be struggling if I did not stumble upon that tool | 15:08.09 |
henrys | jogux: makes sense. | 15:08.19 |
jogux | henrys : but there are some oddities, eg. the expensive £20 quid SO+ is much older than SO2 at the moment, for no obvious reason. | 15:08.49 |
Robin_Watts | jogux: maturity. | 15:09.10 |
jogux | robin_watts : I suspect "low sales" or "got forgotten". | 15:09.28 |
henrys | like wine ;-) | 15:09.31 |
Robin_Watts | jogux: not low sales, "exclusivity". | 15:09.49 |
jogux | :) | 15:09.52 |
pedro_mac | henrys: Iâm away from saturday for 2 weeks, but Iâll probably do between 3-4 hours a day before the lazy ones get up. I guess some of that time may be worth spending on SOT bugfixing?. While we wait on Good responses paulgardiner has offered to put another pair of eyes on the android activity swithcing issues to see if Iâm missing something | 15:09.57 |
pedro_mac | clarifies that the âlazy onesâ are the womenfolk :) | 15:10.51 |
henrys | pedro_mac: yeah the bug fixing would be good. Iâm also wondering if we should get marcosw to report more bugs. | 15:11.00 |
jogux | robin_watts: the 1.4 build on the appstore definitely won't install on your ios5 ipod... | 15:11.25 |
| robin_watts: I guess/hope someone deliberately changed that | 15:11.33 |
Robin_Watts | henrys: I keep hoping that the styling fixes that jogux is working on will solve a lot of problems. | 15:12.14 |
pedro_mac | henrys: yeah, there are a lot of ooxml issues we could report which arenât on the existing list. Styling should be a big help | 15:12.25 |
Robin_Watts | so if marcosw reports bugs now, we might just get to close them again. | 15:12.36 |
| but if he can spot things that clearly aren't styling related, they would be good to have. | 15:12.56 |
jogux | robin_watts : it doesn't seem to fix that much in ATS iirc. | 15:13.07 |
Robin_Watts | jogux: aww... | 15:13.24 |
pedro_mac | there are also various things we donât support like center-aligned tabs, footnotes, watermarks, smartart etc; bigger chunks of work but Iâd wager that things like smartart would be seen by a lot of customers as a core feature | 15:13.55 |
henrys | marcosw: may have a feel for what is there and can come up with some orthogonal issues. Iâm not sure. | 15:14.03 |
Robin_Watts | smartart is a scary amount of work. | 15:14.19 |
pedro_mac | yup | 15:14.24 |
paulgardiner | henrys: I'd like to book a week off in August (not sure exactly when yet). Would that be a problem? | 15:16.21 |
henrys | another path is to have an âexpertâ look at xml and make more educated guesses at what to report as a bug. | 15:16.23 |
pedro_mac | I had been wondering about subsets - there may be some commonly used elements we could support without doing the whole thingâ¦.*if* weâre wanting to cover âcommonly usedâ features initially- like the â3dâ button backgrounds in PPT | 15:16.34 |
henrys | paulgardiner: nope, itâs summer | 15:16.45 |
paulgardiner | Ta | 15:16.50 |
| It's even summer here, which is very unusual. | 15:17.19 |
rayjj | marcosw: I think you are confusing issue #2 and #3. AFAIK, all they downloaded was the later pdf_info.ps which doesn't require rebuilding | 15:17.43 |
Robin_Watts | pedro_mac: The problem is that even the 'subsets' of smart art that you pick would rely on the core framework. | 15:17.46 |
| Unless you do it very crappily and special case it all. | 15:18.02 |
jogux | henrys : both mupdf / smartoffice have some relatively nasty bugs interacting with the OS in the iOS 8 beta. I reckon they're unstability on the OS side, but may be worth keeping an eye on. | 15:18.12 |
pedro_mac | yeah, I havenât looked at what the split would be | 15:18.25 |
jogux | henrys : also, smartoffice crashes a lot on the new Android L, that's probably fairly important to look at. | 15:18.28 |
henrys | paulgardiner: a guy Iâm training with (a brit) always complains about the UK weather - we did a 100 mile bike ride in 35 C he was ready to go home⦠| 15:19.07 |
marcosw2 | rayjj: I just send customer 532 a patch for gsicc_create.c (the email subject is fts_19-1904.pdf). You might check to make sure the patch works on the simulator. | 15:19.38 |
paulgardiner | henrys: :-) Like it | 15:19.56 |
henrys | any other SOT stuff or shall we call it? | 15:19.56 |
rayjj | marcosw: OK. will do. | 15:20.08 |
jogux | robin_watts : http://bugs.ghostscript.com/show_bug.cgi?id=695347 for that PDF issue | 15:20.47 |
Robin_Watts | jogux: Ta. | 15:20.55 |
paulgardiner | jogux: are the Android L and iOS 8 problems likely to be worse than one would see from the current app store versions? | 15:21.01 |
jogux | henrys : I guess, where are we with forums? That's the other piece we need before releasing new app store. | 15:21.07 |
| paulgardiner: no, I don't believe they're regressions. | 15:21.17 |
henrys | jogux: oh yes thank you for reminding me. | 15:21.24 |
jogux | paulgardiner: the Android L one we're certain to need to fix before September, we're 99% certain it's on our side. | 15:21.46 |
| paulgarinder: iOS 8 stuff is just keep an eye on, and panic if it still seems broken in the next but one beta :) | 15:22.09 |
Robin_Watts | henrys: Also, I worry that when september rolls around, we're going to have a massive problem when we try to do builds for all the customers at once. | 15:22.15 |
henrys | jogux: the forums and Iâve asked ron to share his problems with the CMS picsel was using⦠heâs just banging his head against the wall. I was hoping you guys might be able to help | 15:22.34 |
Robin_Watts | I'd really rather arrive at september having done at least one build for every customer. | 15:22.47 |
jogux | robin_watts: SOT? by then it'll be a button push in ATS ;-) | 15:22.47 |
| henrys : I recall everyone hating that's touched it hating that CMS :-) | 15:23.21 |
Robin_Watts | Partly to prove that we have all their toolchains working, and partly that we are supplying exactly what they expect. | 15:23.34 |
jogux | robin_watts : historically SOT upgrades haven't really been pushed to customers at once - it was more if a customer actually *asked* for a new build with new features. | 15:24.27 |
Robin_Watts | jogux: Right, but come september, we are going to have to push to all customers at once. | 15:24.48 |
henrys | jogux: we make all crashes P1 | 15:24.49 |
jogux | robin_watts : oh right. I didn't know that. | 15:25.22 |
Robin_Watts | which I suspect will mean that we won't actually get everything out to all customers until at least october :) | 15:25.28 |
jogux | do the customers know that? :-) | 15:25.29 |
pedro_mac | has to pop out for half an hour - back at 5 | 15:25.37 |
Robin_Watts | jogux: I think miles has been promising "a september release" to customers as he signs them up. | 15:25.52 |
jogux | robin_watts : right, ok. | 15:26.08 |
henrys | Robin_Watts: I think that is something worth talking miles out of. We can say we made the improvement and they can have them if they want them. Iâm betting many wouldnât upgrade - embedded upgrrade == pain | 15:26.30 |
jogux | I don't think we even have a list of all the builds for all the customers right now :-( | 15:26.39 |
| (which would actually be useful to know what we should put into ATS) | 15:26.51 |
Robin_Watts | jogux: Which is part of the issue I am trying to raise. | 15:27.08 |
jogux | robin_watts : yeah, sorry, I'm stepping on your toes a bit :) | 15:27.26 |
Robin_Watts | jogux: Not at all. | 15:27.34 |
| It would be good to get a list of all the customers that are expecting a september release in advance, together with an up to date contact at each company. | 15:28.12 |
paulgardiner | is so impressed at people not so overwhelmed by present problems that they can think ahead towards avoiding future ones. | 15:28.48 |
Robin_Watts | That way we can do a test build to confirm that a) it works, b) it's what they expect etc. and it gives us something to put in ATS as an autobuild so we can be sure we don't break it. | 15:28.51 |
mattchz | The L thing can almost certainly be reproduced on KitKat (4.4) by enabling the preview of the new Android runtime. | 15:30.05 |
jogux | paulgardiner : so I've beaten about your testflight build of iOS SO2 and feel relatively happy with it. | 15:30.07 |
mvrhel_laptop | henrys: I have to step out for bit | 15:30.31 |
| bbs | 15:30.40 |
paulgardiner | mattchz: KitKat 4.4. Damn! I'll need a new phone. | 15:30.55 |
jogux | henrys : the CMS stuff doesn't need to hold up the SO appstore releases, correct? | 15:31.04 |
henrys | jogux: no | 15:31.18 |
paulgardiner | jogux: thanks. That's good | 15:31.54 |
mattchz | paul> :) | 15:31.58 |
rayjj | marcosw: It looks like the patch uses check_range vs. check_internal_range. There is no check_range in their simulator code, although there is a declaration in base/gxcie.h | 15:32.05 |
mattchz | I can look at the Android L/KitKat ART thing if necessary | 15:32.20 |
jogux | henrys : but we are awaiting forums, is there an ETA on them? | 15:32.41 |
| (I mean, we're waiting for the forums before we release) | 15:32.55 |
paulgardiner | The CMS thing. I don't understand. So is it that the web pages for the help documentation are autogenerated by the CMS from some simpler form? | 15:33.42 |
jogux | paulgardiner: I think so. | 15:34.24 |
henrys | jogux, paulgardiner Iâve only spoken to Ron a few times I have no idea | 15:34.26 |
paulgardiner | So could we just save the result as html/css etc? | 15:34.43 |
jogux | henrys : we decided on discourse, didn't we? | 15:34.46 |
| did someone tell Ron that? :-) | 15:34.56 |
henrys | paulgardiner: correct about autogenerating and he canât find where the âpicselâ logo lives | 15:35.11 |
jogux | paulgardiner: Yes. But I think we generate mutiple forms (android, ios, phone, tablet, etc), so that's likely replacing one problems with a much worse one. | 15:35.35 |
henrys | jogux: I did tell Ron we decided on discourse and the hostname we agreed to. | 15:35.45 |
paulgardiner | Oh I see | 15:35.57 |
Robin_Watts | henrys: Dunno what CMS it is, but I know with the artifex one, it required a specific browser (IE) to be used in order to see all the content. | 15:36.24 |
paulgardiner | I wouldn't mind taking a look if someone can direct me to where. | 15:36.32 |
Robin_Watts | chrome didn't work at least. | 15:36.34 |
jogux | paulgardiner: we may be able to ask someone that knows :-) | 15:37.04 |
paulgardiner | right :-) | 15:37.12 |
| What's Neil G doing these days? | 15:37.27 |
jogux | paulgardiner: working for us | 15:37.32 |
| us == emobix in this case :-) | 15:37.37 |
paulgardiner | Oh right. | 15:37.40 |
henrys | it uses expressengine | 15:37.40 |
paulgardiner | That could be handy. | 15:37.54 |
rayjj | marcosw: I sent a corrected version to Len (cust 532) | 15:37.55 |
jogux | paulgardiner: he's already been fixing zoomset code ;-) | 15:38.12 |
henrys | paulgardiner, jogux I just need to follow up with him. | 15:38.13 |
paulgardiner | jogux: exactly where I thought he might be handy, not that I'd put much restriction on where he might be handy | 15:38.43 |
| jogux: he might be able to confirm my interpretation of that ruler problem | 15:39.14 |
mattchz | Please could someone review the last two commits here please: | 15:41.00 |
| http://git.ghostscript.com/?p=user/matt/mupdf.git;a=summary | 15:41.00 |
| (from an IPR point of view, I think itâs ok, the OP suggested we just create a new patch âinspiredâ by his which I did) | 15:41.52 |
kens2 | Looks like my missing email is coming in now | 15:44.59 |
paulgardiner | mattchz: look good, but a comment near isInEditMode might be good because I can imagine staring at it wondering how it was related to editing documents. | 15:45.07 |
mattchz | yeah, good point. | 15:45.23 |
| paulgardiner: howâs that: http://git.ghostscript.com/?p=user/matt/mupdf.git;a=commitdiff;h=a2ed894eaf6782c7616f7d13fdc8a6002271a039 | 15:49.08 |
paulgardiner | Fine... other than it looks like the word "do" has gone missing. | 15:50.53 |
mattchz | Dâoh. Not having a good day today :( | 15:51.15 |
paulgardiner | henrys: did you see the automated response to my email to MattS? :-( | 15:51.39 |
chrisl | henrys: *all* microtype fonts are essentially "synthesised" and as far as I am aware, all available glyphs are available via glyph names | 15:51.59 |
mattchz | paulgardiner: http://git.ghostscript.com/?p=user/matt/mupdf.git;a=commitdiff;h=bae042e93be9e516d950b317349f206cf1921aea | 15:52.31 |
paulgardiner | mattchz: laaavely | 15:52.58 |
mattchz | ta | 15:53.03 |
henrys | paulgardiner: no auto responses usually just go to the original sender. Can you forward? | 15:53.03 |
jogux | paulgardiner: http://git.ghostscript.com/?p=mupdf.git;a=commitdiff;h=d4a69b3f changed the iOS build to need iOS 6.1, do you think that was intentional? :-) | 15:53.13 |
mattchz | If that *is* intentional, we could perhaps enable ARC. | 15:53.43 |
marcosw | rayjj: thx. | 15:54.43 |
paulgardiner | jogux: it was. It is necessary to using their latest SDK, and so presumably will be a condition on all GD apps. | 15:55.51 |
jogux | paulgardiner: that's a mupdf commit :-) | 15:56.01 |
paulgardiner | :-) oh! I didn't look because I remember the SOT commit. :-( | 15:56.29 |
jogux | hehe | 15:56.36 |
| oh, that reminds me, we should try that new SOT on an old iOS version | 15:57.10 |
| it claimes it'll work on iOS 4.3 :) | 15:57.48 |
paulgardiner | jogux: strange concerning MuPDF - I have a recollection of having to change the target to 6.1 for some reason, but not to do with that commit. | 16:00.01 |
henrys | chrisl: I used to think that as well but I wonder if that didnât change at sum point. Just eyeballing this it doesnât look synthesized to me: http://www.w3.org/Submission/2008/SUBM-MTX-20080305/ but that might be something separate from the microtype fonts we are using. | 16:01.31 |
jogux | paulgardiner: http://git.ghostscript.com/?p=mupdf.git;a=commitdiff;h=08ca7372 set it in more places :-) | 16:01.39 |
henrys | s/sum/some | 16:01.41 |
jogux | paulgardiner : albeit, I guess the root question is, you believe it *should* be iOS 6.1 and upwards only? | 16:02.21 |
paulgardiner | jogux: well that may be down to missing knowledge of iOS on my part. | 16:02.54 |
mattchz | Iâm glad Iâm not that hte only one that struggles to remember why I changed bits of code ;) | 16:04.00 |
paulgardiner | It might have been a mistaken belief that that was necessary for submission to the app store, whereas I guess that concerns the max version | 16:04.05 |
Robin_Watts | jogux: A quick google suggests that ios 4 and 5 account for less than 4% overall. | 16:04.10 |
paulgardiner | My current thoughts are the lower we can support the better | 16:04.30 |
Robin_Watts | paulgardiner: Sure, but if it's going to require us to do extra work, I think we can safely drop 4%. | 16:04.53 |
chrisl | henrys: that's not microtype, that's some method of coding/compressing TTF fonts - I think the use of the "Microtype" name is marketing nonsense, and misleading at that | 16:05.04 |
jogux | robin_watts: actually it's 2% [for devices currently active on the appstore] | 16:05.26 |
paulgardiner | I have to go, but I'll be in an out. | 16:05.35 |
Robin_Watts | jogux: OK, I think that reinforces my point :) | 16:05.58 |
jogux | I wonder what we tell paying customers we support. | 16:06.27 |
henrys | chrisl: that makes sense | 16:06.28 |
jogux | robin_watts: in fact I wonder that for SO2 too. | 16:06.46 |
paulgardiner | The most recent change I made was to add AirPrint support. Could that effect this? | 16:07.09 |
jogux | paulgardiner: don't think so. | 16:07.16 |
henrys | chrisl, tor8 so letâs just go with whatever you guys have now and not worry about pcl | 16:07.20 |
chrisl | henrys: the fact it mentions loca table, CVT table, hmdx etc - it's probably an extension of their "ACT" fonts "format" | 16:07.52 |
paulgardiner | The lowest value for which it builds without extra work sounds good. | 16:08.08 |
jogux | paulgardiner: we would have to actually test it you know ;-) | 16:08.22 |
mattchz | I have a 3.1.3 ipod on ebay right now if anyone wants to buy it btw :) | 16:08.50 |
paulgardiner | jogux: yeah I just realised the testing issue as I walked out to the kitchen | 16:09.30 |
jogux | paulgardiner: I'm inclined to agree with Robin that we should forget it, though I am curious as to what we tell paying customers we support for both mupdf & SO2 | 16:10.34 |
paulgardiner | and if there's anyone who likes mattchz's 3.1.3 iPod so much that they want a second one... | 16:10.50 |
jogux | we are not supporting iOS 3 :-) | 16:11.11 |
| In fact there's some iOS 2 code I need to delete. | 16:11.16 |
paulgardiner | Is no support old kit better than potentially broken support? Just thinking do we really have to test | 16:13.07 |
| ? | 16:13.10 |
jogux | paulgardiner: yes - if it's listed as not supported, people get to install/keep an older known working (hopefully) version | 16:13.36 |
| unless we want to get them to test for us :-) | 16:14.04 |
paulgardiner | yeah good point | 16:14.09 |
Robin_Watts | So, I need to pick a few unicode chars to look at to guess the ideographic space width from. | 16:14.33 |
| Anyone want to suggest a good set of 'prepresentative' ideographs? | 16:14.51 |
| or representative if you prefer. | 16:15.00 |
jogux | robin_watts : I don't really know :-S | 16:30.37 |
kens2 | Goodnight all | 16:32.24 |
pedro_mac | bye kens2 | 16:32.46 |
Robin_Watts | jogux: I picked a random set. | 16:44.20 |
mattchz | tor8: do you mind reviewing this please? (fixes CBZ on iOS) http://git.ghostscript.com/?p=user/matt/mupdf.git;a=commit;h=e7d36663c7b8385476d0d51fdab65213f98050c9 | 16:50.45 |
jogux | robin_watts: cool :-) | 16:58.13 |
| mattchz : I take it you did test that and it now works in some particular circumstance? :-) | 16:58.59 |
mattchz | it works on my iPad, and it didnât before :) | 16:59.19 |
| I generally do test my code before review, yes ;-) | 16:59.32 |
jogux | well, I mean: with an email attachment, when viewing from safari, when viewing from dropbox? | 16:59.35 |
mattchz | I tried with an email attachment. | 16:59.48 |
| I can try with dropbox if you think thatâll work differently. | 17:00.02 |
| It donât think it should, as it should be based on the file extension. | 17:00.16 |
jogux | and possibly it could be different on ios 6 vs 7 (I have a recollection of Paul deliberately adding .zip) | 17:00.18 |
| mattchz: safari would work off the mimetype presumably | 17:00.34 |
| although maybe it maps that to fileextension | 17:00.46 |
mattchz | the only case that I imagine wouldnât work is if another app exported a UTI for public.cbz-archive, without specifying CBZ as an extension | 17:00.49 |
| jogux> very possibly; we donât specify mimetypes for any our types | 17:01.01 |
jogux | you should get a choice of apps if there's multiple ones? | 17:01.04 |
mattchz | yeah, I got a choice of apps from email. | 17:01.14 |
| the mimetype in the email is actually application/octet-strema | 17:01.51 |
jogux | git blame says I'm wrong about Paul having added .zip :-) | 17:02.45 |
mattchz | yeah, it was tor8 :) | 17:02.52 |
| seems to work in Safari | 17:03.20 |
jogux | cool | 17:03.26 |
mattchz | this website uses the mimetype âtext/plain" | 17:03.58 |
jogux | nice | 17:04.45 |
mattchz | perhaps we should add mimetypes though. | 17:04.58 |
jogux | I never 100% understood exactly how this stuff worked :) | 17:05.10 |
mattchz | I have seen mention of âapplication/x-cbz' | 17:05.18 |
| jogux> yeah, me neither. Itâs confusing. | 17:05.29 |
pedro_mac | finds a few lines in his copious android logcat output which suggest some class lookup failure in the SOT alien⦠might be good to sort that before we do anything else⦠| 17:06.51 |
jogux | btw, why aren't there binary releases of mupdf libraries for Android? wouldn't that save people that want to incorporate it into their app a whole bunch of pain fiddling with ndk, etc? (I feel there must be an obvious reason that I missed) | 17:11.55 |
pedro_mac | too hot here - its around 26 degrees C. Why am I going to Turkey for an insane 40 degrees? | 17:19.37 |
pedro_mac | gets the thermometer and its 29. I must be in the wrong place | 17:30.29 |
Robin_Watts | jogux: We could do such releases. | 17:31.54 |
| but then the published API for mupdf is a C level one. | 17:32.07 |
| so people would need to do their own JNI classes. | 17:32.16 |
tor8 | mattchz: application/x-cbz is the common mimetype for CBZ files | 17:32.22 |
mattchz | yeah. | 17:32.39 |
jogux | robin_watts : I presume (perhaps incorrectly) that people would try to take the JNI files that are already there | 17:32.47 |
tor8 | that's becoming a standard format now I think, a lot of the comic book ebooks are distributed in that format | 17:32.52 |
Robin_Watts | and IME most android devs don't understand JNI. | 17:32.55 |
tor8 | mattchz: have you tested the change? | 17:33.13 |
Robin_Watts | jogux: The JNI files there do not reflect the underlying C api. | 17:33.14 |
| They offer a much smaller set of APIs packaged up in a way that makes sense for our app to sit on top of. | 17:33.40 |
tor8 | private type identifiers are supposed to be domain based (like the com.microsoft.xps and com.adobe.pdf) | 17:33.56 |
Robin_Watts | If people try to just extend them willy nilly they will hit all sorts of problems. | 17:34.00 |
mattchz | tor8: yep | 17:34.05 |
tor8 | but I don't think there's an "owner" for cbz ... com.example.cbz? ;) | 17:34.12 |
mattchz | Iâm currently testing a new patch, which adds the mimetypes: | 17:34.13 |
| http://git.ghostscript.com/?p=user/matt/mupdf.git;a=commit;h=2fc6ce674a746af2a0b20731f302cac7ea120f0f | 17:34.14 |
jogux | robin_watts : Hm. I am tempted to sum that up as "most android developers shouldn't try and use mupdf" but I suspect that wasn't what you meant :-) | 17:34.18 |
Robin_Watts | One of the jobs on our list is to do a proper JNI reflection of the C API and to make the example app use that. | 17:34.34 |
jogux | but I guess this is what your new jni... mm | 17:34.40 |
tor8 | mattchz: yeah, that commit looks more complete | 17:34.55 |
Robin_Watts | jogux: No, most android devs want to take the existing app and just fiddle at the java level. | 17:34.56 |
mattchz | tor8: https://github.com/arauchfuss/Simple-Comic/issues/49 | 17:34.59 |
| I found that earlier | 17:35.02 |
pedro_mac | I hate our SO jni layer too (where my head is just now) | 17:35.05 |
mattchz | tor8: and https://github.com/arauchfuss/Simple-Comic/blob/master/Info.plist | 17:35.19 |
jogux | robin_watts : if they want to do that, isn't that an argument for binary releases of the C bit? | 17:35.28 |
tor8 | mattchz: might make sense to look at what CloudReaders has registered | 17:35.40 |
| https://itunes.apple.com/gb/app/cloudreaders-pdf-cbz-cbr/id363484920?mt=8 | 17:35.46 |
mattchz | public.cbz-archive and public.archive.cbz looked the most standard | 17:35.52 |
Robin_Watts | jogux: for paying customers we'd build it for them :) | 17:35.54 |
mattchz | tor8: ok. | 17:36.00 |
jogux | robin_watts: :) | 17:36.06 |
tor8 | mattchz: ok. I'll take your word for it :) | 17:36.06 |
Robin_Watts | for free users, if they aren't capable of building it, I don't want 'em hacking on it :) | 17:36.12 |
jogux | Hm. :-) | 17:36.20 |
mattchz | tbh, Iâm not sure the UTI name matters that much. | 17:36.27 |
| So long as it doesnât clash with another that registers a different file extension | 17:36.40 |
tor8 | mattchz: might cause collisions if more than one app registers the same file extension with a different UTI | 17:36.52 |
mattchz | which I presume was the problem with the ZIP one | 17:36.54 |
tor8 | yeah. well, go with the mest common one (public.cbz-archive I guess) | 17:37.19 |
pedro_mac | Robin_Watts: TRUE, BUT THEYâLL JUST END UP ON HERE ASKING HOW TO BUILD⦠;) | 17:37.19 |
| oops - capslock failure | 17:37.29 |
tor8 | pedro_mac: I thought you were emulating the users who end up here asking how to build ;) | 17:37.44 |
pedro_mac | curse my stubby fingers | 17:37.45 |
mattchz | tor8: does that matter though? I guess it depends where the extension -> UTI mapping occurs. | 17:37.52 |
pedro_mac | tor8: :) | 17:37.52 |
jogux | haha | 17:38.04 |
mattchz | multiple defintions of the same UTI would be bad, I think | 17:38.22 |
tor8 | mattchz: I'd say let's worry about it if someone reports the bug. as long as I can open my application/x-cbz downloaded comics in mupdf I'm happy :) | 17:38.25 |
mattchz | :-D | 17:38.29 |
tor8 | I used to get by with using the file transfer to the app directly in iTunes ... but I don't have iTunes installed on any machine anymore which complicates matters | 17:38.57 |
mattchz | Iâm just testing with box.net and safari. it seems to work fine from email. I also need to test oxps now Iâve split that into a separate UTI. | 17:39.06 |
| (Iâd not heard of OXPS and CBZ until today!) | 17:39.30 |
Robin_Watts | mattchz: Few people have ever heard of OXPS. Even fewer use it. | 17:47.33 |
mattchz | are they only available in win 8? In Win 7, I can output to XPS, but no idea how to do OXPS. | 17:48.46 |
Robin_Watts | I'm not sure MS supports OXPS. | 17:51.17 |
| OXPS is just the "open" version of XPS. Essentially it's XPS with a very slightly different header, I think. | 17:51.51 |
| like a different id in the XML or something. | 17:52.01 |
mattchz | do you have any OXPS test files? | 17:52.48 |
Robin_Watts | not me, but there may be some in our repo on casper. | 17:53.10 |
| failing that tor or mvrhel may have some. | 17:53.18 |
mattchz | ta. whereâs the repo on casper? | 17:56.02 |
Robin_Watts | It's an SVN repo. | 17:57.45 |
| so it's not terribly convenient to check stuff out of. | 17:57.59 |
jogux | so it turns out we don't have any iOS 4.3 devices to test on (the ipod I thought had 4.3 only has 4.2 as latest available), so I think we should change SO2 to be minimum 5.0 at least. | 17:58.02 |
Robin_Watts | Unless you want the whole lot. | 17:58.14 |
| jogux: That sounds sensible. | 17:58.23 |
mattchz | perhaps I donât need to actually test OXPS in practice ... | 17:58.36 |
Robin_Watts | mattchz: I wouldn't bother. | 17:58.43 |
jogux | I'd be tempted to suggest higher, but would really want to know that customers weren't insisting on ios5 support first :) | 17:59.20 |
Robin_Watts | jogux: I'd be tempted to change to ios6, and then if customers complain we can rethink it. | 18:01.03 |
jogux | that would save on testing | 18:01.22 |
Robin_Watts | it's not like by changing to 6 now, we are making it harder to change back to 5 later on, right? | 18:01.29 |
| Just don't go culling the ios5 paths from the code yet :) | 18:01.40 |
jogux | that sounds about right. | 18:01.54 |
Robin_Watts | Anything that saves developer time has got to be good. | 18:02.09 |
jogux | we should probably explicitly tell customers iOS 5 is no longer supported when we send out releases | 18:02.11 |
Robin_Watts | jogux: No, we explicitly tell them that this release supports ios 6 and above :) | 18:02.32 |
jogux | yes, that :-) | 18:02.44 |
| paulgardiner: ^^^ that ;-) | 18:03.02 |
mattchz | tor8: is that latest patch ok? | 18:05.27 |
| I tested with CBZ/PDF/XPS from Email, DropBox and Safari, on iOS 7 and iOS 6.1 | 18:05.45 |
tor8 | mattchz: yes, LGTM | 18:10.43 |
mattchz | excellent ta. | 18:11.17 |
| oh, I also found a CBZ which crashed the app, but Iâll look into that more tomorrow | 18:11.41 |
| âopen document '/var/mobile/Applications/7C960E53-9417-48BC-8210-456D5631A395/Documents/Elf Receiver Radio-Craft August 1936.cbz' | 18:11.53 |
| error: cannot find end of central directory | 18:11.54 |
| uncaught exception: cannot find end of central directory | 18:11.55 |
| " | 18:11.56 |
| if that gives any clues... | 18:11.57 |
jogux | sounds like a corrupt zip | 18:12.21 |
mattchz | yeah, was thinking that. | 18:12.36 |
| ok, nn all. | 18:13.45 |
pedro_mac | cya mattchz | 18:13.56 |
Robin_Watts | rats, missed mattchz. | 18:14.12 |
| That sounds like we're cocking up the exception handling somehow. | 18:14.29 |
jogux | hometime! goodnight all. | 18:27.50 |
Robin_Watts | night. | 18:27.57 |
pedro_mac | goes in search of food | 18:29.55 |
sebras | mattchz: (for the logs) my HTC Sensation runs Android 2.3.3. I just compiled Mupdf 1.5 (b2f096d) for my device. | 23:42.46 |
| mattchz: I just verified that the behaviour is still the same. | 23:45.11 |
| mattchz: if you manage to see the same with your Android 2.3.x device then I guess we can conclude that this is a behaviour change in Android itself somewhere between 2.3.x and 4.3 that I used on my HTC One. | 23:46.17 |
| (no, I haven't upgraded yet) | 23:46.26 |
sebras | sleepz. | 23:47.37 |
| Forward 1 day (to 2014/07/02)>>> | |