| <<<Back 1 day (to 2014/07/01) | 2014/07/02 |
pedro_mac | morning folks | 07:42.32 |
jogux | good morning all | 08:07.08 |
kens | Morning | 08:07.14 |
shelly_ | morning all, hope you are all having fun! | 08:59.50 |
kens | Morning shelly_ | 09:00.09 |
pedro_mac | hi shelly_ - absolutely :) | 09:01.14 |
shelly_ | kens: I am trying to look into the latest bug from Ray 695343 but failing to find a checkout that works | 09:02.41 |
kens | err, let me look at that bug | 09:02.57 |
shelly_ | kens: is there a list of old regression runs that I can check to see when this started? | 09:03.11 |
kens | I've no reason to believe it ever worked.... | 09:03.32 |
shelly_ | kens: ah ha, we have reached the same point | 09:04.11 |
kens | Its not a deg fault, so it may never have worked, and nobody ever noticed | 09:04.14 |
| THat happens a lot for me with pdfwrite, I fix something and a whole bunch of test suite fiels come up as diffs, because they never worked previously | 09:04.48 |
shelly_ | ok, so this is a fix from scratch rather than repair something that is now broken | 09:05.24 |
kens | I would say so, yes. Of course I could be wrong, but I don't see anythgin there that indicates this ever worked | 09:05.50 |
| I can quickly try versions of GS back to 8.71 if that helps | 09:06.16 |
shelly_ | I tried back to just before the openjpeg 1.5 merge in but no luck | 09:06.47 |
kens | Well, its jbig2, so I wouldn't expect openjpeg to make any difference | 09:07.19 |
| let me see if I have the tets file here | 09:07.34 |
shelly_ | sorry I meant the last big jbig2 merge in, been working on openjpeg too long! | 09:07.58 |
kens | Hmm, I don't have the test file here | 09:08.06 |
| I'll get it off casper | 09:08.22 |
shelly_ | don't worry, let me dig a little further back myself, catch you guys another time | 09:08.57 |
kens | OK | 09:09.03 |
| Oh shelly went off, it turns out that 8.71 does work. Can someone let him know if he comes back please | 09:11.05 |
| Hmm, actually 8.71 doesn't work either, its just a different message. | 09:12.25 |
| Heading off to ride, back in a few hours. | 09:18.02 |
sebras | mattchz: good morning! | 09:18.16 |
| mattchz: there's some info for you in the logs. | 09:18.49 |
mattchz | sebras: coo, thanks | 09:21.00 |
sebras | mattchz: I'm curious about your results though, so keep me posted. (I'm at work right now, so it might take some time before I answer) | 09:23.15 |
mattchz | sebras: cool. I actually forgot to bring my HTC desire in today after all that. Our 2.3 device here is running a samsung version of android, which doesnât have the Downloads app. | 09:25.24 |
| pedro: do we still have the emobix HTC desire in the office? | 09:25.43 |
pedro_mac | mattchz: donât think so - or at least I havenât been able to locate it when we looked last time | 09:28.04 |
pedro_mac | only found the hero and wildfires | 09:28.22 |
mattchz | I read a bit more about the new behaviour in L this morning | 09:28.35 |
| pedro> odd. Iâm sure I bought it back in after I borrowed it. | 09:28.49 |
pedro_mac | mattchz: I had it at home about a year or so ago for something, but I didnât find it | 09:30.43 |
mattchz | Iâll try and actually remember to bring mine in tomorrow. | 09:31.16 |
sebras | mattchz: I remembered just now that my employers phone also runs android 2.3.x. I verified the old behaviour with mupdf 1.2 just now. let me upgrade to 1.5 and try again. | 09:40.28 |
| mattchz: this is android 2.3.6 btw. | 09:40.34 |
mattchz | sebras: I presume all parts of the bug are still the same? | 09:40.34 |
| ok | 09:40.44 |
sebras | mattchz: ok, the behaviour is identical to what I described in the bug. | 09:43.19 |
mattchz | cool. | 09:43.27 |
| what device is your employers phone? | 09:43.39 |
sebras | mattchz: it's a samsung device which has a download app though. | 09:43.42 |
mattchz | oh, interesting. perhaps I missed it on this device. | 09:43.55 |
| oh, itâs there after all :) | 09:44.12 |
| Iâm just stupid. | 09:44.17 |
sebras | mattchz: ;-) | 09:44.27 |
Robin_Watts | sebras: specifically you're getting the 'showing the wrong document' problem? | 09:44.28 |
sebras | Robin_Watts: I am. | 09:44.35 |
mattchz | silly me to assume that the apps would all be in in alphabetical order. | 09:44.36 |
sebras | mattchz: mmm, I think that is a samsung feature. | 09:44.51 |
| mattchz: HTC does the same nowadays, and you can have subcategories of apps and so on... | 09:45.16 |
mattchz | ah, you can move stuff around in the launcher. | 09:45.26 |
pedro_mac | you can change that in the options | 09:45.32 |
mattchz | I use stock android, so I donât see that on my phone. | 09:45.45 |
| Samsung have basically copied the iphone behaviour :) | 09:46.12 |
| ok, I can reproduce it now. | 09:50.30 |
Robin_Watts | pedro_mac, mattchz, jogux: I got a mail from miles last night, asking about the picsel.com website. | 09:51.16 |
jogux | mmhmm? | 09:51.32 |
Robin_Watts | If you go to the picsel.com website, and click on "view" and then on "Visual Explorer", you get a screenshot with lots of files as thumbnails. | 09:51.55 |
| How do you actually get to that in the app? | 09:52.06 |
| Has that been replaced by the timeline thing? | 09:52.14 |
jogux | looks | 09:53.51 |
| robin_watts: finally back to style stuff btw. problem with shapes showing as rectangles is that the NPH group type has been blatted over where the shape type was stored :-S | 09:54.29 |
Robin_Watts | jogux: Ah. | 09:54.56 |
jogux | I think that was my exact response. | 09:55.13 |
Robin_Watts | Can we put 2 things on a single group? | 09:57.07 |
| Or do we need to nest groups? | 09:57.19 |
jogux | not both as the type; not sure there's another field we could store one of them into | 09:57.48 |
pedro_mac | UE2FILEVIEWER_VE_TIMELINE_VIEW is set for dispman_ue2fileviewer, so thatâs probably what switches the explorer type | 09:58.48 |
sebras | mattchz: ok, great. | 09:58.49 |
Robin_Watts | pedro_mac: ah. | 09:59.31 |
jogux | I certainly can't find a way to get to it in the appstore build on iphone or our latest build on ipad. | 10:01.11 |
mattchz | has not clue, sorry. | 10:01.30 |
pedro_mac | its enabled on some builds and explicitly removed in others I have a test build going to check | 10:02.00 |
mattchz | reads the Tasks & Back Stack page again, to see if I can actually understand it properly this time. | 10:14.50 |
jogux | robin_watts : we can have multiple "styles" (equivalent to the html class="foo" attribute), so I think I'll try changing NPH/PH to be stored as styles | 10:24.32 |
sebras | mattchz: http://bugs.ghostscript.com/show_bug.cgi?id=693344 isn't this one already solved..? | 10:24.57 |
| mattchz: for android. | 10:25.14 |
mattchz | looks like it. Iâl l close it. | 10:25.59 |
sebras | mattchz: I'm unsure about ios that tor8 mentioned though... | 10:30.46 |
mattchz | oh, hmm. | 10:30.59 |
| looks like we donât in ios. | 10:31.27 |
| Iâll reopen. Iâm not sure itâs very iOSy to put the version in the title tbh. | 10:31.55 |
| It perhaps should be just hidden away in a preference entry for the app (along with any license agreements etc) | 10:32.19 |
jogux | mattchz: might not be terrible to have a 'about' button somewhere in the app (I'm thinking that might be a good place to stick a 'hey, did you know you can license mupdf for inclusion in your own apps' type thing) | 10:34.08 |
mattchz | mm. | 10:35.42 |
jogux | perhaps a hardcoded 'about mupdf' pdf. hmm. dunno. :) | 10:36.46 |
mattchz | I donât mind really. Let me know and Iâll implement it. | 10:37.13 |
jogux | I'm just brain dumping, tor8 would be the one to decide perhaps | 10:38.45 |
Robin_Watts | Having an embedded PDF would be nice, as we could put simple instructions in there. | 10:41.08 |
| but I'm not sure how we do that in the interface for the android version. | 10:41.29 |
| (I have no opinion on ios) | 10:41.47 |
jogux | that said, an embedded pdf maybe is'nt the best place to put the version number. | 10:45.03 |
Robin_Watts | It would be hard to change the embedded PDF without changing the whole app, so it seems 'safe' at least. | 10:46.38 |
paulgardiner | pedro_mac: amongst your commits are some changes to the Android build which look like they aren't Good specific. Not seeing anything odd about them, but I'm wondering why I haven't needed them to perform builds of SO2 here. | 10:58.56 |
pedro_mac | paulgardiner: which one(s) in particular? | 10:59.29 |
jogux | spins the ATS wheel of fortune. | 11:00.27 |
paulgardiner | pedro_mac: I think there's 3 or 4. The very first is one of them. | 11:00.47 |
Robin_Watts | morning tor8. | 11:04.05 |
pedro_mac | its possible that has been fixed on master since that was done; i was getting linker failures due to the response file being rejected | 11:04.16 |
Robin_Watts | There has been some discussion of the need for an about box on ios mupdf. | 11:04.28 |
pedro_mac | or possibly by sdk tools update | 11:04.46 |
Robin_Watts | How do you feel about that? Or an embedded PDF with the version in (and possibly some instructions) | 11:04.49 |
tor8 | Robin_Watts: we used to have an instruction file embedded in it. | 11:06.07 |
| an about box or About.pdf file would be nice, I agree. I guess an about box will be easier to update with the version number automatically. | 11:06.47 |
paulgardiner | pedro_mac: there's a change in the location of AndroidManifest.xml in "Add android Good project to solution build for VisualGDB debugging" | 11:07.43 |
mattchz | sebras/jogux/tor8/paul_gardiner/anyone else: so what do you think the correct behaviour of ths should be? | 11:12.04 |
| (this applies to ios too, because when using open in, we currently stack documents too). | 11:12.17 |
tor8 | mattchz: I think we should embed an About or README file with instructions | 11:12.28 |
| at least for iOS | 11:12.33 |
mattchz | sorry, I was talking about the issue sebras reported (the switching to the wrong document one). | 11:13.01 |
| I see there are a few options: | 11:13.26 |
| 1) Only ever allow one document to be open at once | 11:13.35 |
tor8 | mattchz: if you're asking about the 'open in...' stuff, I would personally prefer if we just switched to the running app, or start it at the top level, and not live "inside" another app's context | 11:13.35 |
| but please list the options :) | 11:13.56 |
mattchz | ok | 11:14.03 |
| 1) Only ever allow one document to be open at once. A newly opened document always replaces an existing one. If the existing one has changes, we would need to ask the user to save or discard them. | 11:14.47 |
pedro_mac | paulgardiner: yep; that was (iirc) necessary to get the manifestmerge working correctly (merging in the Good library/assets). Iâll make a note to check again whether its mandatory | 11:15.15 |
mattchz | On Android, this would probably involve everything run in the âMuPDFâ task, rather than as part of another appâs task | 11:15.34 |
| 2) Allow documents to be stacked on top of each other. This could be done either within the launchers task (e.g. Downloads), or everything could open in the single MuPDF task. | 11:16.29 |
| If the user opened A,B, then A again, Iâm not sure what could happen. | 11:16.46 |
| I donât think we can reorder activities on android. | 11:16.54 |
| We could close B, popping back to A. | 11:17.15 |
| or we could have a second instance of A on top of B (i.e. A B A) | 11:17.25 |
tor8 | mattchz: is it possible on android to always have the 'back' button go back to the file selection view? | 11:17.55 |
| though I guess that'd mess up more than it'd help | 11:18.10 |
mattchz | it should be, if we run everything in the MuPDF task. Maybe. | 11:18.21 |
Robin_Watts | I think I like the idea of other apps being able to invoke us as part of themselves. | 11:18.40 |
mattchz | actually, Iâm not sure if that would only work if the MuPDF had already been started manually. | 11:18.41 |
| Robin: me too. and that seems to work quite well on 4.x | 11:19.44 |
tor8 | my preference would be to only allow one document open at a time per task (so back would always go back to the previous view) and each time you open we start a new task | 11:19.46 |
mattchz | On 2.3, itâs bizarre. | 11:19.49 |
Robin_Watts | I suspect we possibly can say "this is just 2.3, who cares?" now ? | 11:20.07 |
| It's not like it's a crash etc. | 11:20.15 |
tor8 | so you don't end up with "open in..." in one app, and then back will take you back to the previous document you were viewing in mupdf.app | 11:20.21 |
| so you can see I've already changed my mind :) | 11:20.56 |
mattchz | tor8: that wonât be an issue if we launch as part of the other appâs task (as we do now) | 11:20.58 |
| Imagine the sequence | 11:21.16 |
| 1) Launch download app | 11:21.19 |
| 2) Pick a.pdf | 11:21.22 |
| 3) Switch to home screen (or another app) | 11:21.31 |
paulgardiner | Option 1 sounds good to me, for memory usage if nothing else | 11:22.04 |
mattchz | In 4.x, youâll see one entry in the task manager âDownloadâ. Selecting Downloads will take you back to a.pdf. Pressing back will take you back to the download list. Thatâs fine, imho. | 11:22.25 |
| In 2.3.x, it appears that selecting âDownloadsâ from them task manager takes you back to the Download list. | 11:24.07 |
paulgardiner | mattchz: so are you saying that one sensible option is leave it as it is? | 11:24.42 |
mattchz | If you then launch b.pdf, then b.pdf appears on top of a.pdf | 11:24.53 |
| pressing back takes you back to a.pdf, then back again takes you to the download list. | 11:25.50 |
| paulgardiner: yep, if we donât care about 2.x | 11:26.05 |
Robin_Watts | mattchz: That behaviour sounds perfect to me. (Except for 2.x obviously) | 11:26.28 |
mattchz | tbh, I think Iâd be tempted to leave it as it is, then experiment with the new L stuff and see if we can make it better there. | 11:26.47 |
| the memory thing, is a concern, I guess. | 11:27.04 |
| I think in L, we can have a separate task manager entry for each document. | 11:28.00 |
| That is far nicer, imho. | 11:28.03 |
| lunch beckons. | 11:28.55 |
paulgardiner | pedro_mac: ah, so tjhat part isn't really part of setting up VisualGDB? It's part of the commit two back from there? If so I can do some git rebase -i stuff to put it back there. | 11:29.05 |
pedro_mac | paulgardiner: yeah, it shoudlnât have been in that commit. :( | 11:29.54 |
paulgardiner | Hey no worries. Easy to reorder. I usually go through the same process with my branches | 11:31.40 |
pedro_mac | paulgardiner: I think the remaining oddity on the branch is an assert in the alien during a call to Pal_Thread_blockReschedule() following an onResume() of the PicselViewer activity (switching to homescreen and back to SOG). The assert is that weâre not trying to set block to the value it is already set to | 11:34.47 |
paulgardiner | Right. I'll have a look in a bit | 11:36.20 |
| Still going through getting to understand things. | 11:37.18 |
| The AndroidManifest.xml moving down into a bin subdir: is that a file we generate, or do we copy an original or something? | 11:38.21 |
| It's just that, in addition to seeing a change in the declaration of where it is, I was expecting to also see it being moved if it was part of our source, or it's being generated in that new location. | 11:39.58 |
pedro_mac | paulgardiner: we copy AndroidManifest_Template.xml from the source dir to the ue2fileviewer dir in genroot then perform unifdef and some token replacement. The bin version gets generated as part of the build iirc | 11:40.07 |
paulgardiner | Hmmm. So was the declaration without "bin/" simply wrong? | 11:40.52 |
pedro_mac | due to the manifestmerge | 11:41.02 |
| no, just that weâve never used manifestmerge before | 11:41.19 |
paulgardiner | And using manifextmerge changes it's location? | 11:41.49 |
| or the bin version is the result of the merge? | 11:42.25 |
pedro_mac | yep, it takes the project manifest and augments it with the merged activity declarations etc | 11:42.30 |
| stores the merged version in bin/ ready for aapt to build the apk | 11:42.46 |
paulgardiner | So (do we need to)/(are we) now using manifestmerge for non-Good builds? | 11:44.27 |
pedro_mac | no, we only specify manifestmerge for good builds (just running a vanilla dispman_ue2fileviewer build again on the branch so that I can answer your next question :) ) | 11:45.54 |
pedro_mac | lunches | 11:46.17 |
sebras | mattchz: tor8: Robin_Watts: paulgardiner: since it appears to work on 4.x then I'm tempted to agree with mattchz that we could just blame 2.3.x and then close the bug. | 11:49.16 |
henrys | jogux, paulgardiner : Ron is going to show up 9:00 am pacific, so 4 pm Greenwich - with CMS questions. | 11:50.25 |
jogux | ok | 11:50.41 |
henrys | paulgardiner, pedro_mac : in talking to miles we decided just to wait for Matt Sturges to return, plenty of bugs to work on until then. | 11:52.35 |
paulgardiner | henrys: right. sounds like a plan | 11:57.18 |
| henrys: and in the case of Android, there's issues we're looking at that aren't blocked on waiting for their feedback | 12:00.12 |
mattchz | sebras: cool. Shall I close it then? | 12:48.11 |
| I think itâs definitely worth opening an enhancement request for Android L though. | 12:48.32 |
kens | Hmm, that's an interesting series of errirs on the cluster | 12:53.50 |
| a bunch of 'connection interrupted' and 5 conection timed out, alarm' | 12:54.15 |
| So it looks like hubbles is trying to run all the tests. THis might take some time.... | 12:54.40 |
Robin_Watts | pedro, jogux, paulgardiner: http://intranet.picsel.com/~ats/cgi-bin/results-tgvath.pl?resultid=1237 | 13:08.47 |
| Those are the results of my space reworking commits. | 13:09.16 |
| just a single commit now, actually. See robin/master | 13:09.27 |
henrys | kens: hubbles grinds on | 13:09.28 |
kens | Yep, I've no idea how to stop it and restart the test..... | 13:09.47 |
| Come back next week..... | 13:10.00 |
Robin_Watts | kens: I can try... | 13:10.01 |
kens | Robin_Watts : it woiuld be worth trying | 13:10.13 |
| I did a test beforehand so I'm confident of the code | 13:10.24 |
paulgardiner | Robin_Watts: it looks to have changed the selection boxes a bit, as though the the defined area of the characters are a tiny box | 13:12.13 |
Robin_Watts | paulgardiner: Indeed, the ascender/descender of the spaces are now 0. | 13:12.35 |
| I can change that, but it means that more files experience slight vertical shifts. | 13:13.30 |
| kens: test should be rerunning. | 13:17.49 |
Robin_Watts | pops out to get a typhoid jab. | 13:17.56 |
kens | Seems to be thanks Robin | 13:18.01 |
jogux | robin_watts : I'm slightly surprised by search-restore-pdf-1.03.bmp where the selection box no longer seems to cover the top of the S. | 13:18.42 |
Robin_Watts | jogux: Yeah. I can't explain that. | 13:19.45 |
| Unless the ascender/descender for that font is broken? and it was only working before because of the defaults used in the spaces fonts? | 13:20.30 |
jogux | yes, or some other similar existing brokeness that your change has uncovered | 13:20.56 |
mattchz | has anyone seen a crash when opening a pdf from gmail into mupdf before? | 13:21.38 |
Robin_Watts | mattchz: I think there used to be a bug about that, but it was sorted. | 13:21.55 |
| If people could find the time to look at my patch and feedback, I'd be grateful. Back in a bit. | 13:22.19 |
mattchz | E/AndroidRuntime(12438): java.lang.RuntimeException: Unable to start activity ComponentInfo{com.artifex.mupdfdemo/com.artifex.mupdfdemo.MuPDFActivity}: java.lang.SecurityException: Permission Denial: reading com.android.email.provider.EmailProvider uri content://com.android.email.provider/attachment/31 from pid=12438, uid=10043 requires com.android.email.permission.ACCESS_PROVIDER | 13:22.22 |
| Actually, itâs not âGmailâ, itâs the Email app | 13:23.39 |
kens | Hmm, cluster just died again | 13:25.41 |
| leaving hubbles, fermis and peeves running | 13:25.58 |
jogux | robin_watts : so my fix of changing the type -> a style/class has worked, but there's still issues. Various odd layouts + a crash due to processing styling that we'd previously ignored and apparently don't understand. | 13:43.32 |
Robin_Watts | mattchz: OK, so that's a change to the android manifest required or something? | 13:44.52 |
mattchz | I donât think so. | 13:45.03 |
Robin_Watts | jogux: OK... | 13:45.21 |
mattchz | It should have permission, because when I open the intent in Intent Analyzer, itâs been granted special permission for the URI. | 13:45.22 |
| Just trying to figure out why it doesnât have permission... | 13:45.32 |
| (other apps can read it) | 13:46.03 |
tor8 | chrisl: argh. the microsoft fonts don't always have metrics that match the AFM files from Adobe's Base 14 fonts | 13:46.57 |
Robin_Watts | tor8: indeed, no. | 13:47.11 |
| For SOT the fonts match MS, and we then 'force' PDF metrics onto them at runtime for the PDF agent. | 13:47.31 |
tor8 | macron: 333 vs 500, plusminus: 564 vs 549, etc | 13:47.34 |
Robin_Watts | The question is, do PDF and PCL agree ? | 13:48.11 |
tor8 | I'll have to ask chrisl to rerun his script but with the full character set rather than just the ones that don't exist in the MS fonts | 13:48.56 |
| Robin_Watts: the other question is: what metrics do the PDF generators that assume arial == built-in helvetica use? | 13:49.58 |
jogux | decides to stop and fix the ATS script viewer, as it's getting frigging annoying that it doesn't work. | 13:51.13 |
Robin_Watts | tor8: They must follow the PDF metrics, surely. | 13:51.54 |
chrisl | tor8: the only slight worry is that as I mentioned, these aren't metrics read directly from the font, but have been through the scaling via the rip | 13:51.57 |
henrys | tor8, Robin_Watts : in the ufst you are to use the same metrics but the documentation prescribes an unusual PCL rounding algorithm. | 13:52.33 |
Robin_Watts | henrys: right, but do the standard PCL fonts have the same metrics as the MS ones? or as the PDF ones? or as something else entirely? | 13:53.08 |
henrys | same in pcl and pdf that is. | 13:53.11 |
mattchz | robin_watts: oh, I think itâs something to do with the code added to deal with the transformer primeâs file manager, maybe. | 13:53.33 |
henrys | Robin_Watts: I thought the URW ttâs were windows compatible. | 13:54.08 |
Robin_Watts | henrys: Well, that implies they are PDF-incompatible. | 13:55.02 |
| PS/PDF-incompatible. | 13:55.11 |
| If we were hoping to get fonts from urw that would be PS/PDF/PCL/Windows compatible, we're out of luck. | 13:55.49 |
henrys | does mac word use windows tt fonts? | 13:56.56 |
Robin_Watts | What do you mean by 'windows tt fonts' ? | 13:57.17 |
| tt fonts are tt fonts. | 13:57.29 |
| windows and mac use slightly different tables from inside them I think, but basically the font structure and metrics are the same. | 13:58.01 |
chrisl | I suspect henrys meant does Word on Mac ship the same TTF files it has access to on Windows | 13:59.01 |
Robin_Watts | then I would assume the answer is yes. | 13:59.14 |
| Anyone here have office on a mac? | 13:59.32 |
henrys | chrisl: yes that is what I meant - Iâm checking now. | 13:59.56 |
pedro_mac | Robin_Watts: jogux has office365 | 14:00.13 |
jogux | I have an Artial.ttf in my mac's /Libraries/Fonts that has lots of (C) microsoft in it. does that answer the question? | 14:00.22 |
| Arial.ttf, even. | 14:00.46 |
jogux | can't find a mac that I haven't install office on to check if that's part of the system or came with office. | 14:02.05 |
chrisl | I can boot one if necessary - although it's and older OS X version | 14:02.58 |
pedro_mac | jogux: I can check that | 14:03.11 |
Robin_Watts | Arial apparently has shipped with Macs since at least 10.0 | 14:03.29 |
| http://support.apple.com/kb/ht5154 | 14:04.09 |
tor8 | Robin_Watts: Macs have always had the core web fonts, IIRC | 14:04.16 |
pedro_mac | no Arial for me | 14:04.56 |
| (mavericks) | 14:05.13 |
henrys | jogux: my verson of word for mac uses a âWindow Office Compatibleâ collection which I assume MS installed in Library/FontCollections | 14:05.17 |
| tor8: yeah that sounds right | 14:05.30 |
Robin_Watts | pedro_mac: Even in /Library/Fonts ? | 14:05.35 |
pedro_mac | nope, thatâs where Iâm looking | 14:05.46 |
jogux | pedro_mac : try typing arial in spotlight? | 14:06.07 |
Robin_Watts | pedro_mac: That would seem to contradict what the link above says then :( | 14:06.19 |
jogux | henrys: hm. they're not fonts though. I'm not sure what they are. | 14:06.21 |
pedro_mac | ok, thereâs no arial font in /libraries/Fonts, but thereâs an Arial font in FontBook | 14:07.19 |
henrys | jogux: ugh - Iâm looking at open files and ports for the word app hoping to see the font file | 14:07.28 |
chrisl | jogux: mac font files can be "augmented" - I can't remember the details | 14:07.37 |
jogux | :) | 14:07.40 |
| it's a binary plist, but I can't figure out what it's really saying | 14:08.00 |
chrisl | IIRC, there's an offset into the file to find the actual font data, which should then be basic TTF | 14:09.05 |
tor8 | there's also the msttcorefonts package on linux, which downloads the core web fonts | 14:09.57 |
| which I assume are the most common denominator version we could use | 14:10.08 |
pedro_mac | confesses - theyâre in /System/Library/Fonts... | 14:10.27 |
| but only when I search in spotlight | 14:11.19 |
| not when browsed via finder | 14:11.26 |
henrys | ah Library/Fonts/Microsoft/ | 14:13.25 |
jogux | pedro_mac : Apple in their wisdom hide all kinds of stuff in the finder to prevent users messing things up | 14:13.42 |
pedro_mac | strike that last comment - getting confused by the fact that all the /Library/Fonts fonts are also in /System/Library/Fonts | 14:13.45 |
jogux | henrys: ahh, nice. a second, different, copy of Arial.ttf. | 14:14.36 |
henrys | jogux: on my system the Arial is 0 bytes | 14:14.51 |
| with no ttf extension | 14:15.28 |
jogux | http://pastebin.com/Q9hrgqcq is what I have | 14:15.36 |
jogux | has no idea what the 0 byte ones are about | 14:16.19 |
henrys | jogux: theyâre all acl for me. | 14:18.50 |
jogux | acl? | 14:19.56 |
tor8 | interesting. the middot character in the mscorettfonts times.ttf matches PDF but not on the newer Win7 times.ttf | 14:20.35 |
| so I think I know which one to use :) | 14:20.41 |
chrisl | Oh, OS X has the option for font data to be stored in the data fork - I'm not sure that's what I was meaning, though | 14:22.43 |
tor8 | if it's a ".dfont" on OSX it's an old format mac TTF but with the resource fork stored in the data fork | 14:23.22 |
| and the 0 byte ones are probably ones with all the data still in the resource fork | 14:23.50 |
henrys | jogux: the extended attributes the @ in the listing. But that doesnât seem to lead to anything enlightening | 14:25.54 |
jogux | ah right | 14:26.02 |
pedro_mac | has this result for Arial in /library/Fonts and /System/Library/Fonts: http://pastebin.com/yRqVZB5u | 14:26.31 |
| thatâs a vanilla macbook with no M$ stuff installed | 14:27.05 |
henrys | when I open a 0 byte font say arial in Word I get a resource fork: /Library/Fonts/Microsoft/Arial/..namedfork/rsrc | 14:36.37 |
mattchz | paulgardiner/robin_watts: do you still have an old Transformer Prime to test the file manager stuff on? | 14:39.46 |
| Ours seems to have 4.x, which iirc works differently :( | 14:39.57 |
paulgardiner | Robin has one, but probably updated | 14:41.03 |
Robin_Watts | My transformer prime has 4.x on too, I think. | 14:41.10 |
mattchz | ah :( | 14:41.13 |
| Iâm not actually convinced the current code is right, anyway :( | 14:41.25 |
| or necessary, perhaps. | 14:41.31 |
henrys | tor8, chrisl: so we have 3 different metrics, I assume ufst, windows and pdf? | 14:41.39 |
mattchz | Iâm struggling to find out if we really should be looking for the _data field in the content resolver results, or if really we should just always just call getContentResolver().openInputStream(uri) instead. | 14:42.30 |
| doing a query on the URI, is causing the security exception (on 2.x) when opening an attatchment from âEmail'. | 14:43.12 |
| on 4.x, itâs succeeding, but not able to open the path in the _data field. | 14:43.30 |
| I think we should probably be just calling getContentResolver().openInputStream(uri) in all cases. | 14:43.52 |
henrys | asked a simple question to support about my wireless adaptor and I was simply sent a shipping box. Maybe this is the future of support, talking costs too much. | 14:43.52 |
Robin_Watts | mattchz: How about trying what you think is the proper way, and if that fails, fall back to the old way ? | 14:43.54 |
mattchz | I could do, yeah. | 14:44.10 |
| Either way, it would be good to test this. | 14:44.18 |
| Iâm just trying to find out where the _data thing came from ⦠was that just reverse engineered, or was it an officially suggested thing to try? | 14:44.54 |
Robin_Watts | mattchz: That was a paulgardiner-ism. Possibly from stackoverflow ? | 14:45.16 |
jogux | mattchz: I'm going to put money on "c&ped from stackoverflow" | 14:45.19 |
mattchz | facepalms | 14:45.24 |
| not that I donât do the same :) | 14:45.37 |
henrys | Robin_Watts, paulgardiner: should we worry about MS metrics for SOT will you incorporate any of these fonts in SOT or can we narrow the discussiong to PDLâs? | 14:46.27 |
| the urw fonts we are ordering? | 14:46.41 |
paulgardiner | Absolutley. Is there another way to approach these things? :-) | 14:47.20 |
Robin_Watts | henrys: For SOT the existing 'plan' is that all fonts are windows compatible, and we 'force' PDF metrics on the base PDF ones. | 14:47.23 |
| but I would ignore the SOT needs for now and concentrate on PDLs. | 14:47.51 |
paulgardiner | I've thought for some time thought we'd be better copying to a local file and opening | 14:48.43 |
tor8 | henrys: make that 4: adobe, ufst, microsoft core web fonts, microsoft windows 7 | 14:49.24 |
| I was surprised that the metrics have changed in the microsoft fonts depending on which version you use | 14:49.45 |
henrys | tor8: I donât understand the issue with ms - ufst and adobe not matching is a problem. | 14:50.33 |
tor8 | chrisl: could you get me the full metrics set from ufst? | 14:51.06 |
kens | My cluster test finished :-) | 14:51.41 |
chrisl | tor8: yeh, I'll need to do a little rewriting, shouldn't take long..... is the csv form okay | 14:52.09 |
tor8 | henrys: yeah, I'm not inclined to worry over much about ms fonts not matching adobe. we'll just match adobe where we have the data and use the ms where adobe metrics aren't available | 14:52.22 |
| chrisl: yeah, that form is perfect | 14:52.30 |
| I'm using csv myself :) | 14:52.43 |
Robin_Watts | I think MS is a complete red herring here for the PDL use. | 14:52.57 |
tor8 | Robin_Watts: the problem is, lots of pdf generators assume the MS fonts are available and don't embed them, and then proceed to use glyphs that aren't in the core adobe set | 14:53.31 |
Robin_Watts | The worry is that we wanted to get a set of fonts that we could use for PS/PDF/PCL, and it now looks like that's screwed, as PS/PDF and PCL disagree on metrics. | 14:53.50 |
tor8 | and all assume that Arial == Helvetica | 14:53.55 |
henrys | Robin_Watts: indeed I was sort of curious how MS was handling because I though MS and Mac metrics were different and I wanted to see how they handled that. | 14:54.03 |
tor8 | if I get the data from chris I can compare and see if the UFST and Adobe actually mismatch | 14:54.27 |
Robin_Watts | so either we need 2 sets of fonts (PS/PDF and PCL), or we need to be able to 'force' the metrics on the fonts after the fact. | 14:54.28 |
henrys | Windows and Mac metrics different that is | 14:54.35 |
Robin_Watts | tor8: That sounds sensible. | 14:54.37 |
tor8 | if it's only rounding error differences from getting scaled metrics out of ufst, I'm not inclined to worry | 14:55.13 |
| but differences for things like 'plusminus' where it's 549 in ms and 564 in adobe are more disconcerting | 14:55.50 |
| thankfully it's only a rare handful of glyphs that differ so | 14:56.02 |
henrys | tor8: for pxl almost everything is xyshow so escapement is not much an issue. | 14:56.05 |
tor8 | and I'd be happy to just go with the adobe metrics for that | 14:56.40 |
jogux | robin_watts : other major problems with this styling change is it seems to have broken editting in very odd ways. like you change the level of a bullet point, and 3 other buillet points vanish. joy. (plus also a very minor alignment issue.) | 15:04.29 |
Robin_Watts | jogux: urgh. | 15:07.01 |
jogux | so that's 3 more issues I need to look into before we can think about commiting this | 15:07.24 |
| but in good news, I found 3 other documents in ATS it seems to fix. | 15:07.34 |
| and in other good news, the ATH script viewer in ATS now works :-) | 15:07.46 |
Robin_Watts | well, that's a good sign, I guess. | 15:07.48 |
henrys | marcosw2: can you make an admin bugzilla change, under Admin -> Field Values -> Status -> Add âWaiting for reviewâ ⦠I assume nobody will object to that⦠| 15:21.43 |
chrisl | tor8: metrics info on it's way - slightly different set of fonts (based on PS2 + PCL rather than PS3 + PCL) - I *think* it should contain all the ones you need, if not, let me know | 15:42.19 |
kens | henrys did you see the gs-devel email from Johannes Meixner ? | 15:55.00 |
henrys | kens: canât we shut that thing down? just kidding, Iâll look | 15:55.43 |
chrisl | kens: I was tempted to reply with "update or sod off!" | 15:56.08 |
kens | I'm not sure how to reply to him, the obvious answer is 'no we don't have a public test file repository'. Chrisl I feel somewhat the same, if they patch GS randomly its hteir lookout | 15:56.26 |
| But he did ask nicely | 15:56.39 |
chrisl | We do have a test archive of non-private files | 15:56.54 |
kens | Yeah, but its not actually anywhere public I htink | 15:57.08 |
chrisl | No, I think it used to be accessible via svn, but not any more | 15:57.32 |
kens | I'm tempted to ask him what other applications do in this situation, I can't believe they supply patches for archaic versions of their software | 15:57.36 |
tor8 | chrisl: that's what I needed, thanks! | 15:57.50 |
| chrisl: henrys: but ... bad news | 15:57.59 |
henrys | chrisl: so what do the ufst do when we build a pcl/ps/pdf rip? | 15:58.35 |
chrisl | henrys: huh? | 15:58.58 |
henrys | tor8: tor implied the metrics are different | 15:59.13 |
tor8 | ufst metrics match *neither* the adobe nor the microsoft fonts | 15:59.14 |
| for 'mu', adobe=500, ufst=250, ms=576 | 15:59.51 |
chrisl | Which font? | 16:00.07 |
tor8 | for 'periodcentered', adobe=250, ms=333, ufst=350 | 16:00.20 |
| Times-Roman | 16:00.25 |
henrys | chrisl: surely your favorite customer is using the ufst for all the languages. | 16:00.25 |
chrisl | henrys: 532? | 16:00.59 |
tor8 | dcaron: 688, 608, 646 | 16:01.02 |
henrys | kens: Iâve always wanted to have a public directory but it never seems we can get a very large collection of files becuase everything we get is private by default | 16:01.29 |
| chrisl: yes | 16:01.36 |
kens | henrys so someone should just reply and say 'no' ? | 16:02.18 |
chrisl | henrys: we can only use the metrics we get. | 16:02.22 |
henrys | the ufst might have a table of metric separate for postscript | 16:04.35 |
chrisl | No, it doesn't | 16:04.56 |
jogux | hi Ron | 16:05.20 |
Ron_Licht | Good morning. | 16:05.32 |
henrys | hello Ron_Licht | 16:05.45 |
chrisl | tor8: I think there's a cock-up - I get a width of 500 for mu from UFST | 16:06.06 |
Ron_Licht | Henry suggested I enlist your help on some problems I'm having with the Picsel CMS - Expression Engine.... | 16:06.17 |
| There are a series of User Guides but I cannot find where they are controlled, and how they can be edited. | 16:07.11 |
henrys | Ron_Licht: we are also interested in getting the forum up as soon as possibly we are being sent user email and not responding to it. | 16:07.17 |
chrisl | tor8: and my table has mu at 500 for Times-Roman - there is a glyph called mu1 which is 250 | 16:08.03 |
Ron_Licht | Forum is there, I just have to clean it and populate it with some initial content. | 16:08.05 |
jogux | ron: where / how can I log in to have a look at things? | 16:08.10 |
tor8 | chrisl: ufst has both mu and mu1 at the same char, 500 and 250 respectively | 16:08.22 |
Ron_Licht | jogux - is it ok to post URL, login and password here? | 16:08.44 |
jogux | no logins - can you put them in a text file on casper? | 16:08.58 |
Ron_Licht | I can put them in a text file, but what is / where is casper? | 16:09.21 |
jogux | ron: ghostscript.com - maybe you don't have an ssh login for that though | 16:09.38 |
chrisl | tor8: yes - I assume mu1 is some HP compatibility thing, but using the same glyph | 16:09.40 |
tor8 | chrisl: yeah, and since the 'mu' matched the adobe I didn't emit an exception for that instance of it | 16:10.02 |
Ron_Licht | I don't have access. | 16:10.06 |
jogux | ron: you don't. how about you email to joseph@emobix.co.uk | 16:10.09 |
Ron_Licht | jogux - stand by ... look for an email from ron@rlicht.com | 16:10.44 |
tor8 | chrisl: and for U+00B7 I also have two instances, periodcentered=350 and middot=250 | 16:10.59 |
| and middot matches the adobe | 16:11.16 |
chrisl | tor8: well, I'm working primarily with glyph names rather than codepoints | 16:11.37 |
tor8 | chrisl: yeah. I'm trying to preserve both, but the ms fonts work with codepoints | 16:11.58 |
| U+010F dcaron is still worrying | 16:12.31 |
chrisl | tor8: note that there is also Dcaron which isn't listed in AGL | 16:12.55 |
| but that's worse! | 16:13.24 |
tor8 | you mean Dmacron? | 16:13.39 |
chrisl | No I mean Dcaron | 16:14.05 |
tor8 | Dcaron is in the AGL | 16:14.11 |
henrys | kens,chrisl: we could put what we have in git and make it public, or do you want to just say no. I donât feel strongly either way. It seems there are no shortage of pdf and even ps files on the web that anyone could cobble together a regression test suite. | 16:14.25 |
Robin_Watts | henrys: git would be a bad choice. | 16:14.54 |
| our old test suite is still available by svn. | 16:15.04 |
chrisl | henrys: note that James Cloos has just replied...... | 16:15.07 |
Robin_Watts | as a poster on the list has just said. | 16:15.10 |
jogux | ron: I'll keep an eye out, but nothing yet | 16:15.26 |
chrisl | tor8: sorry, dcaron is not in the AGL, Dcaron is - their widths are different | 16:15.37 |
kens | henrys, we *could* put something together, but it would take time and would not be very comprehensive, we couldn't put the QL tests in there for instance. While it would be nice to have 'something' better than the old svn set (which is very small) I can't see it being worth our time | 16:15.38 |
chrisl | It's got Ghent, and a fair number of other PDF files | 16:16.32 |
kens | The only thing 'special' about our set, beyond the moderatly comprehensive nature of the QL tests, is that they have once caused problems in GS | 16:16.43 |
chrisl | We could trawl bugzilla for non-private attachments to closed bugs | 16:17.07 |
Ron_Licht | OK jorgux - email is on the way with a brief recap. | 16:17.10 |
kens | What might be a bigger problem woudl be if we kept it up to date, then they would get new files which cause GS a problem, are fixed in newer versions and defintiely not fixed in older ones...... | 16:17.18 |
jogux | ron: got it | 16:17.29 |
kens | chrisl I did consier that,yes | 16:17.30 |
tor8 | chrisl: I don't quite follow you. dcaron and Dcaron are both in the AGL. UFST Times-Roman metrics matches Adobe for Dcaron but not dcaron. | 16:17.37 |
kens | But like I said, it would take time, and not really be any more comprehensive than doing the same on the web | 16:17.56 |
henrys | kens, chrisl: alex put together a bizillion files and parked them somewhere. We could put that up there but I just donât want to fool with it myself. | 16:18.03 |
jogux | ron: do you have an ssh login or similar for the hosting account? | 16:18.16 |
Ron_Licht | Henry: is there any content I should populate the forum with to start, or do you want to just begin with a clean slate? | 16:18.26 |
kens | Alex trawled them off public sources henrys ? | 16:18.28 |
henrys | kens: yes | 16:18.38 |
kens | Well we could make them available, but..... | 16:18.52 |
chrisl | tor8: Sorry, I'm trying to do too many things at once, and not making sense on any of them..... TBH, I'd just go with the Adobe metrics - out primary requirement is for PDF and PS, after all | 16:18.59 |
kens | Its no better than them doign this themszelves, and I don't think we've actually tested them to see if they work with *any* version of Ghostscript.... | 16:19.21 |
tor8 | chrisl: okay. then I'll go with Adobe in preference where they differ, and use UFST or MS where Adobe is missing metrics | 16:19.29 |
Ron_Licht | jogux - i sent you the only login/pw info I have | 16:19.31 |
Robin_Watts | chrisl: Is it? I thought at least part of the driving force for this was that henrys wanted a complete pcl set too ? | 16:19.34 |
jogux | ron: my initial thoughts are that this content isn't being managed by expression engine | 16:20.28 |
mattchz | robin_watts: is there any difference between placing code in an fz_always() and placing it after the entire try/always/catch sequence? | 16:20.43 |
Ron_Licht | jogux: that was my conclusion too | 16:20.48 |
chrisl | Robin_Watts: ideally, yes, but having one common set of fonts that's broken for one or other application isn't going to work | 16:20.49 |
henrys | chrisl: on the business side are requirement is ufst compatibility so I donât think we can do that. | 16:20.49 |
| s/are/our | 16:20.56 |
Robin_Watts | mattchz: Yes. | 16:20.58 |
chrisl | henrys: but we're not talking about UFST compatibility - that was deemed too expensive | 16:21.25 |
mattchz | assuming you donât re-raise an exception, | 16:21.30 |
Robin_Watts | In that case, broadly no. | 16:21.40 |
Ron_Licht | jogux: but since I am so unfamiliar with Expression Engine (this is my first encounter), I thought it was because I wasn't looking in the right place. | 16:21.48 |
henrys | chrisl: no miles is still trying to get that customer to finance the remaining fonts. | 16:21.55 |
Robin_Watts | You mustn't return from an fz_always. | 16:21.59 |
chrisl | henrys: well, at least we can stop worrying about compatibility with Adobe, then | 16:22.41 |
Robin_Watts | but other than that, I can't think of any real issues, no. | 16:22.47 |
mattchz | indeed, but: fz_try(ctx) { .. .} fz_catch(ctx) { ⦠} fz_always { free(x); } | 16:22.52 |
Robin_Watts | eh? | 16:23.02 |
mattchz | is equivalent to fz_try(ctx) { .. .} fz_catch(ctx) { ⦠} free(x) | 16:23.09 |
Robin_Watts | try/always/catch | 16:23.12 |
| not try/catch/always | 16:23.17 |
mattchz | oh, sorry. | 16:23.19 |
henrys | chrisl: thatâs obviously no good either. Iâm surprised theyâre appreciably different | 16:23.35 |
Robin_Watts | try/catch/always - god knows what that'll do :) | 16:23.40 |
| Sadly I can't see an obvious way to catch that error in the macros. | 16:23.59 |
chrisl | henrys: they're probably rarely use glyphs that are generally positioned "manually" | 16:24.04 |
mattchz | robin_watts: thatâs what I mean, I guess. Iâm thinking of java where itâs try - catch - finally | 16:24.21 |
| I hadnât noticed the different order | 16:24.30 |
Robin_Watts | mattchz: mmm, that's why it's always not finally :) | 16:25.06 |
mattchz | :) | 16:25.12 |
| same semantics though? | 16:25.17 |
chrisl | tor8: does that sound right? The ones that differ significantly are more rarely used glyphs? | 16:25.32 |
Robin_Watts | mattchz: should be. | 16:25.48 |
henrys | chrisl, tor8 : is it just the advance or are their design differences? | 16:25.50 |
mattchz | cool. | 16:25.51 |
jogux | ron: looking at the fasthosts control panel I think we can get ssh access, just trying to setup a login for that. | 16:26.02 |
Robin_Watts | mattchz: Ah, wait, I thought of something else | 16:26.13 |
chrisl | henrys: I haven't looked at the outlines - at least, not during this exercise | 16:26.19 |
Robin_Watts | Ideally nothing in the fz_always should throw. | 16:26.21 |
| I don't think it's fatal if it does, it just jumps to the catch, but it's nasty to reason about. | 16:26.53 |
tor8 | chrisl: henrys: I'm only looking at the advance. and yes, the ones that differ are all pretty rare. | 16:26.57 |
mattchz | robin: nod | 16:27.10 |
Ron_Licht | jogux: I can add that to my to do list. | 16:27.40 |
jogux | ron: already done ;) | 16:28.00 |
henrys | chrisl, tor8 : iâm willing to go with adobe if the designs arenât different | 16:28.02 |
Ron_Licht | jogux: I love crossing things off my to do list! :-) thanks | 16:28.48 |
chrisl | henrys: the ones I've looked at in the past, the outlines are similar. | 16:29.35 |
jogux | ron: it's a custom system picsel wrote using the 'Smarty' template language | 16:29.56 |
| ron: I've set you the username/password for ssh.picsel.com | 16:30.46 |
| I've *sent* | 16:30.54 |
Ron_Licht | jogux: is the content accessable? | 16:31.34 |
jogux | ron: yes :-) | 16:34.02 |
| ron: how to actually make a change a get it to show up is the bigger question ;-) | 16:34.15 |
Ron_Licht | jogux: that would be nice too :-) | 16:35.39 |
| jogux: I'm sorry but I have to ask a dumb question. how to I get to the ssh login prompt? | 16:36.16 |
jogux | ron: are you using windows? | 16:36.27 |
Ron_Licht | yes. windows 7 pro | 16:36.46 |
jogux | ron : you can download putty | 16:37.03 |
| ron : are you familiar with the unix command line? | 16:37.16 |
Ron_Licht | I've been exposed to it, but I am nowhere close to familiar to it. | 16:38.38 |
| *with* it | 16:38.53 |
jogux | http://www.picsel.com is still hosted by fasthosts.... right? | 16:40.20 |
Ron_Licht | jogux: yes | 16:40.40 |
jogux | well I've made changes, and they're not appearing, and I don't understand why (I deleted the caches) | 16:41.08 |
Ron_Licht | welcome to my world! :-) | 16:42.36 |
tor8 | chrisl: ufst advance widths differ significantly in the basic greek alphabet (and ufst doesn't have cyrillic at all) | 16:42.45 |
jogux | ron: there's probably someone I can ask for help with this | 16:43.16 |
pedro_mac | jogux: does it require you to publish each change in the front end? | 16:43.18 |
jogux | pedro_mac : "front end"? | 16:43.26 |
kens | Robin_Watts : tor8 SO question: | 16:43.29 |
| http://stackoverflow.com/questions/24532030/mupdf-get-every-image-coordinates-from-pdf-page | 16:43.30 |
jogux | pedro_mac: I'll looking at some MalcolmS php ;-) | 16:43.40 |
| I'm | 16:43.46 |
Robin_Watts | kens: I think he asked on here yesterday. | 16:43.50 |
chrisl | tor8: I think we established that before - can you give me a common cyrillic glyph name? | 16:43.58 |
pedro_mac | so are they not using expression engine then? | 16:43.58 |
Robin_Watts | I'll write an answer. | 16:44.02 |
kens | Robin_Watts : I thought that too, but that question is new, seems to post-date his questions here | 16:44.11 |
jogux | ron: I need to head off now - can you leave this with me for a day? perhaps you can sort out a new clean forum on whatever hostname was picked in the meantime :-) | 16:44.37 |
tor8 | U+0410, afii10017,CYRILLIC CAPITAL LETTER A, | 16:44.51 |
henrys | bbiab | 16:45.12 |
Ron_Licht | jogux: of course! | 16:45.13 |
chrisl | tor8: nope, that doesn't exist in Times-Roman..... | 16:45.29 |
Ron_Licht | jogux: I'll work on the forum in the meantime | 16:45.36 |
tor8 | chrisl: yeah, it only exists in Univers and CGTimes :( | 16:45.41 |
| and all the greek advance widths differ. do we care? | 16:45.55 |
chrisl | Differ by much? | 16:46.42 |
tor8 | chrisl: significantly | 16:47.14 |
| 03B1,alpha,524,GREEK SMALL LETTER ALPHA,web win | 16:47.25 |
| 03B1,alpha,611,GREEK SMALL LETTER ALPHA,ufst | 16:47.25 |
| 03B2,beta,509,GREEK SMALL LETTER BETA,web win | 16:47.25 |
| 03B2,beta,556,GREEK SMALL LETTER BETA,ufst | 16:47.25 |
| 03B3,gamma,442,GREEK SMALL LETTER GAMMA,web win | 16:47.25 |
| 03B3,gamma,556,GREEK SMALL LETTER GAMMA,ufst | 16:47.25 |
chrisl | I wonder how those compare to the PCL font set | 16:50.19 |
| tor8: admittedly picking a font at random, the PCL TTF I looked at seems close to UFST for alpha - 620 | 16:54.22 |
| I can't immediately find AFMs for the Adobe fonts..... | 16:56.04 |
tor8 | chrisl: tor/core35.git | 16:56.43 |
| has the core14 AFM files from Adobe | 16:56.57 |
| (and core35) | 16:57.06 |
chrisl | Ah, and you're using those, I assume? | 16:57.15 |
tor8 | yes, those are my source for the adobe metrics | 16:57.27 |
| the core14 AFMs are newer than the core35 | 16:58.52 |
chrisl | Hmm, and the Adobe font files I have don't have an alpha glyph in them! | 16:59.04 |
tor8 | chrisl: yeah. | 16:59.18 |
chrisl | Except Symbol | 16:59.36 |
tor8 | so my suggestion is: use adobe metrics for latin, microsoft metrics for greek and cyrillic, and ufst for the odds and ends that only UFST has | 16:59.46 |
| the adobe and microsoft metrics are close enough for latin, except for a few strange glyphs | 17:00.24 |
| does that sound reasonable? | 17:01.03 |
| or should we just let them do whatever for greek and cyrillic and not worry about it? | 17:01.14 |
| I'm sure they already have some finished greek/cyrillic charsets they're just going to paste into the fonts anyway... | 17:01.35 |
chrisl | Well, I'm happy with either option. I have a *strong* suspicion that a lot of the really strange glyphs in the UFST output are just there "because they can" | 17:01.55 |
| FWIW, I keep coming back to: embedded the *freakin'* font, then you don't have these worries | 17:02.30 |
tor8 | chrisl: do we need those really strange glyphs at all then? | 17:02.45 |
chrisl | tor8: only for this fetish about UFST compatibility | 17:03.04 |
tor8 | it would be nice (and simple) to just say match microsoft for the WGL4 charset | 17:03.18 |
| but then there are the odd differences with adobe | 17:03.25 |
| which probably matter for PS/PDF | 17:03.43 |
chrisl | I'm intrigued about the differences between Adobe and MS, since on Windows, Acrobat seems to use the MS TTF system fonts...... | 17:04.29 |
Robin_Watts | chrisl: It probably does, but it probably forces the metrics. | 17:04.52 |
chrisl | Seems bonkers | 17:05.18 |
tor8 | I think almost everyone nowadays use the MS TTF system fonts | 17:05.58 |
| which is an argument to just clone them as they are | 17:06.08 |
Robin_Watts | tor8: sure. As long as we then force the metrics. | 17:06.35 |
chrisl | That's not really practical for Postscript fonts in a Postscript rip - possible, but a bit of a 'mare | 17:07.30 |
| And would probably hurt performance | 17:08.01 |
tor8 | Robin_Watts: if the PDF has a /Widths then either way we do the font will work | 17:08.14 |
| the problem is when it relies on the default built-in and doesn't specify a /Widths | 17:08.27 |
chrisl | And PS doesn't have Widths arrays | 17:08.49 |
Robin_Watts | chrisl: It can be done, but it needs help at the font engine level to be efficient. | 17:09.29 |
chrisl | Robin_Watts: it shouldn't *need* done | 17:09.56 |
kens | I'mwith chrisl, we need PostScript fonts that work in PostScript | 17:10.13 |
Robin_Watts | chrisl: If you don't want to do it, then the fonts need to have the correct metrics to start with. | 17:10.20 |
tor8 | well, if we ignore the UFST specific chars, the differences between MS and Adobe are limited to 12 characters | 17:10.42 |
Robin_Watts | and getting the fonts with the correct PS/PDF metrics in seems a fine way to go. | 17:10.48 |
tor8 | and they're the same ones in both Times and Helvetica | 17:10.51 |
Robin_Watts | but they will be wrong for pcl. | 17:11.02 |
chrisl | Robin_Watts: I'm not sure it's possible, either - what if the font is reencoded, or the glyph names are modified? | 17:11.14 |
Robin_Watts | chrisl: If the font is reencoded, then you reencode the font with the modified widths. | 17:11.33 |
chrisl | Robin_Watts: how do we know? | 17:11.48 |
tor8 | macron, plusminus, mu, divide, dcaron, Lcaron, lcaron, tcaron, partialdiff, summation, radical and lozenge | 17:11.50 |
Robin_Watts | chrisl: I don't follow. | 17:12.26 |
chrisl | Robin_Watts: how do we know the font has been reencoded and how do we know what the reencoding is? | 17:12.53 |
Robin_Watts | I really wish I'd just kept my mouth shut on this. | 17:14.17 |
| I have to try and concentrate on this damn SOT thing, sorry. | 17:14.42 |
chrisl | Robin_Watts: it's just all the crazy stuff that you can do with PS because it's a programming language | 17:15.12 |
Robin_Watts | Right, so get the fonts as an exact match to adobe, and force them if we ever want to use them with PCL. | 17:16.08 |
chrisl | Exactly, I was just about to write that | 17:16.27 |
Robin_Watts | I suspect that all PCL users really care about is "are all the glyphs there". | 17:16.43 |
chrisl | With PCL we have a much more "closed" environment for fonts | 17:16.55 |
kens | night all | 17:17.30 |
pedro_mac | night ken | 17:17.41 |
chrisl | tor8: So, the Adobe/MS differences are only the 12 glyphs - ignoring the glyphs only in UFST, how many major differences are there in there? | 17:17.53 |
pedro_mac | disappears too | 17:17.57 |
mattchz | could someone review the last 3 commits here: http://git.ghostscript.com/?p=user/matt/mupdf.git;a=summary please? | 17:18.03 |
chrisl | tor8: I mean Adobe/MS/UFST? | 17:18.17 |
tor8 | chrisl: in the WGL, Adobe/MS vs UFST results in 188 differences | 17:20.45 |
chrisl | I was hoping for less :-( | 17:21.17 |
tor8 | chrisl: it's all the greek + box drawing and arrows | 17:22.13 |
| (and the 12 that already differ between adobe and ms) | 17:23.04 |
chrisl | tor8: personally, I would be tempted to take your suggestion above, and worry about the other differences if anyone notices them. | 17:23.39 |
tor8 | 0132,IJ,735,LATIN CAPITAL LIGATURE IJ,web | 17:23.45 |
| 0132,IJ,773,LATIN CAPITAL LIGATURE IJ,ufst | 17:23.46 |
| is the only major difference in the core that doesn't already have ambiguous answers | 17:24.10 |
| s/core/latin/ | 17:24.18 |
chrisl | That's a smaller difference than I would worry about.... | 17:25.06 |
Robin_Watts | mattchz: All seem fine to me. | 17:25.36 |
tor8 | chrisl: so, WGL4 character set and match Adobe with preference, fallback to Microsoft if adobe has no metrics? | 17:25.47 |
mattchz | thanks, Iâll push then. | 17:25.47 |
| robin_watts: so many cases to think about when opening a file :( | 17:25.59 |
chrisl | tor8: yes, and UFST for any glyphs only in UFST | 17:26.18 |
tor8 | chrisl: that'd be nobreakspace, softhyphen, divisionslash and bulletoperator in the WGL4 charset | 17:27.24 |
| I can add your original list of characters as well, with UFST metrics, if you want | 17:27.42 |
| but that's nothing that should be needed by PS/PDF | 17:27.49 |
chrisl | Indeed. The worry is if we want to claim UFST compatibility........ someone will check it | 17:28.39 |
| tor8: FWIW, when I reviewed the URW vs Adobe widths a while back, my cut-off was about five per cent difference, so the "LATIN CAPITAL LIGATURE" above would have been right on my limit for "significant" | 17:29.10 |
tor8 | chrisl: right. | 17:29.34 |
| chrisl: well, it only adds 53 characters most of which will be copy-pastable between fonts being mathematical symbols | 17:31.00 |
chrisl | tor8: That's good - in all honesty, from what I've seen, my feeling is that a fair number of glyphs in the UFST/MT output are "unstyled" - i.e. exactly the same outline across all the fonts (except maybe Symbol/Zapf etc) | 17:32.44 |
tor8 | chrisl: yeah. | 17:35.36 |
chrisl | tor8: I hope we're not expected to eyeball all the glyphs to make sure the outlines match MS/Adobe/UFST!!! | 17:38.02 |
tor8 | chrisl: I don't intend to! | 17:38.11 |
| I'm only going to eyeball the basic greek/cyrillic to see if they're at least serif and sans-serif :) | 17:38.31 |
| the odds and ends I don't care about | 17:38.43 |
chrisl | I doubt anybody cares, except the bean counters and overly anal QA folks - who we, unfortunately, often have to satisfy :-( | 17:41.23 |
tor8 | chrisl: okay, so I have a set of csv files for each font with the desired metrics for all 726 glyphs we want | 17:42.59 |
chrisl | tor8: sounds good | 17:44.07 |
mattchz | hometime, nn all | 17:45.52 |
tor8 | chrisl: I've mailed them to you. if you can double check if they look reasonable I'd appreciate it, and if all looks good we'll forward them to henrys | 17:46.19 |
| chrisl: unicode,glyphname,width,description,source-of-metrics | 17:47.04 |
chrisl | tor8: the only other thing I could do would be massage my Postcript job, and run it on cust 532's simulator, and check the results in that - but I'm fairly sure when I did it before (for a small subset of glyphs, admittedly), they matched within the 5% I used before | 17:47.43 |
tor8 | chrisl: http://ghostscript.com/~tor/all-metrics.zip the source metrics for the adobe and microsoft fonts in CSV format | 17:49.59 |
| msweb are the core web font archive, mswin are from my windows7 machine | 17:50.58 |
chrisl | tor8: I won't have time to check these properly tonight - I'll do it in the morning | 17:51.12 |
| tor8: one thing is that the code points for agrave and the 8 following it have come out as 0 | 17:53.26 |
tor8 | chrisl: they might be glyphs that are only used as composite characters? | 18:01.12 |
| as part of | 18:01.18 |
| chrisl: no rush, I'm about done for the day | 18:01.23 |
chrisl | I think they all their own entries in the AGL: agrave aacute acircumflex atilde adieresis aring ae ccedilla egrave eacute | 18:04.18 |
| It's 00E0-00E9 that are set to 0 instead | 18:06.23 |
tor8 | chrisl: what are you looking at? the glyph-metrics.csv file I got from you has code points for all of those | 18:07.23 |
chrisl | tor8: I'm looking at the csv files you sent me | 18:07.54 |
tor8 | chrisl: or is the CSV import messing up on the import? | 18:07.57 |
chrisl | Oh, that's possible | 18:08.08 |
tor8 | run sed -i -e 's/^/U+/' *.csv | 18:08.10 |
chrisl | Yes, the csv import is screwing up - in two different spreadsheet apps, ffs! | 18:09.21 |
| Sorry..... strange | 18:09.53 |
tor8 | better add that U+ prefix before we send the files out | 18:10.11 |
chrisl | Like I said, I'll look properly tomorrow | 18:11.07 |
tor8 | chrisl: good night then :) | 18:11.23 |
chrisl | tor8: you too! | 18:11.50 |
halabund | I have two EPS figures of precisely the same width. Is there a way I can create a sinlge figure out of them, by stacking them on top of each other, leaving zero space inbetween? | 19:18.10 |
Robin_Watts | halabund: emacs ? :) | 19:30.29 |
henrys | halabund: you can import each figure into latex, illustrator, or some other application. | 19:36.50 |
| or printer, scissors and tape ;-) | 19:37.39 |
halabund | henrys: Editors are not an option, unfortunately they tend to mangle this particular figure (especially the fonts). Iâll try LaTeX and hope that the EPS -> PDF -> EPS conversion wonât mangle anything. | 19:49.40 |
shelly_ | evening all | 20:53.41 |
| henrys: thanks for the email update | 20:54.24 |
| henrys: wrt closed bugs, some are closed due to my patches and I marked the duplicates accordingly but I believe some are due to the OpenJPEG 2.1 updates | 20:57.03 |
henrys | shelly_: okay thanks | 20:57.38 |
signnow | Hi everyone, | 23:42.16 |
| Can someone help me setup MuPDF on a personal Android project? | 23:42.36 |
| How do I get to use MuPDF inside my personal Android project? | 23:42.49 |
Robin_Watts | signnow: I don't follow the question. | 23:44.04 |
signnow | Hi Robin, | 23:44.11 |
Robin_Watts | MuPDF is licensed in 2 ways. Firstly, it's under the GNU GPL. | 23:44.27 |
signnow | I want to use the latest MuPDF to integrate into my Android app. | 23:44.30 |
| I am trying to follow this thread from Stackoverflow: | 23:45.03 |
| http://stackoverflow.com/questions/8500530/integrate-mupdf-reader-in-an-app | 23:45.03 |
| and had no luck. | 23:45.07 |
Robin_Watts | If you are prepared to comply with the terms of the GNU GPL (which include, but are not limited to, giving all the source for your app away) then you can just use it. | 23:45.12 |
| If you aren't prepared to abide by those terms then you need to get a commercial license. | 23:45.29 |
signnow | Is there some resource that can give step by step instruction for a beginner? | 23:45.32 |
| I see. | 23:45.34 |
Robin_Watts | Depending on the commercial license you get you might get some support bundled with it. | 23:45.55 |
signnow | I am prepared to comply & abide by the GNU GPL | 23:46.32 |
Robin_Watts | Otherwise, if you use the GNU GPL version, you just get to come on here and ask questions, and (if we're not too busy) we'll try to answer them. | 23:46.33 |
signnow | I see. | 23:46.43 |
Robin_Watts | That stackoverflow page is laughably out of date. | 23:47.45 |
| One answer talks about 0.9, the other talks about 1.2 | 23:47.57 |
| We are at 1.5 now. | 23:48.01 |
signnow | yeah | 23:48.02 |
| I am completed misguided there :) | 23:48.08 |
| Is there any resource that can help me understand how I can integrate? | 23:48.27 |
Robin_Watts | So, your best bet is to get the MuPDF source code from git, and then to follow the instructions in ReadMe.txt to be able to build it. | 23:48.41 |
signnow | yes, well I can build it | 23:48.51 |
| However, I am not sure how to integrate it within another Android app | 23:49.11 |
Robin_Watts | Most of the time people just call to it using the standard Intent mechanism. | 23:49.32 |
signnow | ok | 23:49.38 |
| Quick question as a Android newbie: | 23:50.02 |
Robin_Watts | very quick, cos it's almost 1am here and I am off to bed :) | 23:50.35 |
signnow | How do I have my own Activity with a view for MuPDF, instead of launching MuPDF as a separate activity? | 23:50.39 |
| ok | 23:50.41 |
| np | 23:50.42 |
| go ahead, I will come back tomorrow | 23:50.47 |
| I am heading home from work now, thank you Robin! | 23:50.57 |
| :) | 23:50.58 |
Robin_Watts | signnow: I believe that is possible using the standard intent mechanism. | 23:51.00 |
signnow | ok | 23:51.04 |
| I will try it out | 23:51.15 |
Robin_Watts | i.e. you get that for free. | 23:51.17 |
| You *can* bundle MuPDF into your app if you want so you only have to install one thing, but that might require a bit more fiddling. | 23:51.52 |
| come back during the day UK time, and ask again :) | 23:52.05 |
signnow | ok | 23:52.05 |
| ok, I will come back. Have a good night! | 23:52.20 |
Robin_Watts | you too. | 23:52.25 |
| Forward 1 day (to 2014/07/03)>>> | |