| <<<Back 1 day (to 2015/08/17) | 20150818 |
jimgray | Does mupdf support other language search? except english. | 01:41.29 |
| I cannot input japaniese word in mupdf to search text. | 01:42.11 |
kens | chrisl I'm kind of lost in the UFST code. I can 'fix' the problem by reducing the downscale factor I'm using, which makes it quite likely that the problem is busting a value somewhere in the UFST code (it doesn't happen with teh FT code). Unfortunately since I can't debug it on Windows gdb is hampering me looking at it. | 07:22.34 |
| I'm inclined just to decrease the downscaqle from 1/1,000,000 to 1/10,000 whch works. Unles you want to look at it and figure out why ? | 07:23.05 |
chrisl | kens: I have the Windows build problems fixed, I think - but I'm not sure that's going to help debugging the UFST code! | 07:24.16 |
| Where is the downscale factor applied? | 07:24.35 |
kens | You mean by me ? Its in pdf_ops.ps where we set the Text Matrix | 07:24.59 |
| I can't figure out where that's getting applied in the UFST code.... | 07:26.02 |
chrisl | Oh, it's not a FAPI thing | 07:26.14 |
kens | I don't believe it is, no, its something UFST'ish because FreeType works just fie | 07:26.32 |
| fine* | 07:26.39 |
chrisl | I meant the downscale | 07:26.48 |
kens | Its to fix a 0 scale in PDF, I intercept it and replace it with a very small but non-zero scale factor (1/1,000,000) | 07:27.20 |
chrisl | Do you have a cut down test file? | 07:27.52 |
kens | Because I changed teh font code so that it throws an error on degenerate matrixces no matter whether TextAlphaBits is 0 or not. | 07:28.00 |
| I have a massively reduced file, one glyph | 07:28.12 |
| The original is one of those monster CAD drawings | 07:28.39 |
chrisl | Can you send me that? I'll have a poke at the UFST code | 07:28.51 |
kens | Sure, I'll mail it, and tell you where to change the scale factor to stop the problem as well. | 07:29.19 |
| One minute | 07:29.22 |
chrisl | I'll grab a coffee...... | 07:31.19 |
kens | OK its on its way | 07:32.51 |
| I'll get a coffee too, brb | 07:33.01 |
Robin_Watts | Sparky at work here. Expect me to fall off the web a lot. I may be working on my laptop tomorrow. | 09:47.19 |
tor8 | Robin_Watts: oh great. visual studio 2005 errors out on install. unknown error cannot install document services bla bla bla. | 10:19.12 |
Robin_Watts | tor8: hmm. I've never had it not install before :( | 11:23.56 |
chrisl | tor8: maybe "Run as Administrator..."? | 11:26.09 |
Robin_Watts | It installed fine for me on 7 and 8.1 | 11:33.21 |
| chrisl: I updated my 'build an android version of libgs.so' Makefile so that it works with the new directory structure. | 11:55.12 |
| and I hit a problem. | 11:55.32 |
| When you do make so, the last phase of the so-subtarget: is a call to make GSSOC_XE and GSSOX_XE | 11:56.22 |
| Neither of those build for me, so I've commented out those lines in the unix-dll.mak file. | 11:56.41 |
| This isn't a massive problem, but I thought I'd mention it in case there is a cunning trick I can do in the top level Makefile that will avoid me needing to make changes in unix-dll.mak | 11:57.42 |
chrisl | Hrm, okay. not sure why it would be a problem - I'll look at it | 11:58.10 |
Robin_Watts | The other thing... if I do: make -f Makefile.android so | 11:58.30 |
| it fails, because the recursive makes call Makefile, not Makefile.android | 11:58.52 |
| That confused me for a while :) | 11:59.07 |
chrisl | That's probably why all the instructions for the non-configure makefiles say "copy it to Makefile and then make......" | 12:00.23 |
| "make so" works fine for me | 12:02.49 |
| Robin_Watts: the recursive calls problem is unsolvable unless we drop support for non-gnu make tools | 12:25.16 |
| Robin_Watts: actually, I can solve it by detecting GNU make in configure, and setting (or not) the "-f" option for the recursive call based on that | 12:54.54 |
henrys | chrisl: did you intend the default pcl device to be bbox? | 12:55.26 |
chrisl | henrys: No, I thought it was x11alpha... | 12:56.12 |
kens | is eeing a lot of non-deterministic files in cluster runs recently | 12:58.53 |
henrys | chrisl: yea it should be one of the display things but your build changes have it as bbox | 12:59.06 |
chrisl | And yet, it's x11alpha for Ghostscript - how is that even possible :-( | 13:00.29 |
jogux | hm, has bugzilla gone down? | 13:02.16 |
henrys | chrisl: right same tree for me and gs is x11alpha and pcl bbox | 13:03.14 |
jogux | ah, no, seems okay now | 13:03.33 |
chrisl | henrys: it's because the pcl code completely ignores the graphics library's default device setting, and just selects the first device in the list, which happens to be the bbox device | 13:18.03 |
henrys | chrisl: that's the way it's always worked, gs used to be the same. | 13:18.56 |
| chrisl: I can fix it to conform. | 13:19.18 |
chrisl | henrys: I'll do it, as I'm already in there | 13:19.36 |
henrys | chrisl: okay thanks | 13:19.51 |
chrisl | henrys: FWIW, the addition of an explicit default device list dates back to 2007 | 13:21.32 |
henrys | chrisl: well initially I couldn't change it because I couldn't use the default device on windows, there was no display device. | 13:23.10 |
chrisl | henrys: the reason it's a default device *list* is to cope with situations like that | 13:23.49 |
henrys | but yes it should have been changed. I didn't really think about it.... | 13:23.53 |
| before kens did the display device I think pcl defaulted to some printer device on windows and x11alpha or x11 on unix. Rather torn between X11 and x11alpha - rops don't play nice with alpha but most files look better. | 13:28.21 |
kens | forgot I'd done that | 13:28.43 |
henrys | chrisl: which sort of raises the question of why the graphics library should decide the default, i.e. X11 might be best for PCL | 13:30.57 |
| chrisl: not important... | 13:31.23 |
chrisl | henrys: which is why I added an API to the graphics library to allow the calling code to set the default device list | 13:31.40 |
Robin_Watts | fredross-perry: GS was broken, in that it was producing duff rgb data. | 13:32.15 |
| Should be fixed now. | 13:32.22 |
| Updated libgs.so for android in the usual place. | 13:32.29 |
fredross-perry | thanks | 13:32.35 |
chrisl | henrys: at present, that API is only used by the GTK Ghostscript front end | 13:33.07 |
Robin_Watts | android app now behaving exactly as I would expect here. | 13:33.38 |
henrys | chrisl: okay sorry I didn't know about that. I thought you just the library select a default. | 13:34.04 |
Robin_Watts | I might try running some quick profiles on the gs gproof device, see if there are any instant hotspots. | 13:34.06 |
henrys | s/just/just had | 13:34.47 |
chrisl | henrys: there is now the *option* of the calling code changing the default device list - but there is still the graphics library default list created at build time | 13:35.41 |
| kens: I think this is what we need for the UFST problem: http://git.ghostscript.com/?p=user/chrisl/ghostpdl.git;a=commitdiff;h=2da90f4c | 13:42.43 |
kens | chrisl I'll have to take your word for it :-) | 13:43.14 |
chrisl | kens: it just looks like I had a senior moment, or misread the manual...... | 13:43.44 |
kens | Beats me, I can more or less see that its sclaing realted, but beyond that...... | 13:44.09 |
| Oh I see, that does look odd yes | 13:44.44 |
chrisl | Well, anyway, the problem file now works, and no other files show any differences | 13:45.05 |
kens | Sounds good to me! | 13:45.14 |
| I know the scaling factor is silly, but it stops problems with the matrix being 0 | 13:45.32 |
chrisl | Well, it could conceivably happen in the real file, hence I preferred to sort out the FAPI code | 13:46.09 |
kens | Techncally I guess its possible, though I'd have thought unlikely. And since I said that, tomorrow someone will submit a file that does exactly that....... | 13:46.46 |
| If you're happy with it, I am too | 13:47.04 |
| Please feel free to reassign the bug and close it with that fix | 13:47.18 |
chrisl | Well, I've run everything in comparefiles at 72, 75, 150 and 300dpi, and the only file that shows a difference is the one we're looking at so..... | 13:48.04 |
kens | Yeah that sounds good. Like I sad, I'm not really 100% sure about ths UFST any more, but the scaling by 256 does seem rather coarse jumps | 13:48.36 |
| At least I can understand why I was having trouble finding the source of the problem | 13:50.08 |
chrisl | I think it's actually scaling by 32 (since the scale is 1/8ths of a "pixel"), but still seems coarse. | 13:50.32 |
kens | Fair enough | 13:50.43 |
| presumably its 32 on each pass so that's 32, 1024, 32768.... | 13:51.35 |
chrisl | The hx/hy values are in 1/8 pixel scale | 13:52.50 |
| And then the world_scale is the magnitude of the hx/hy values - it's all a bit icky :-( | 13:53.41 |
kens | Yeah, like I said, I'm not surprised I couldn't figure out the problem | 13:54.07 |
| Which makes me feel a bit better anyway | 13:54.17 |
chrisl | The thing is, when the scaling gets small enough to trigger that code, it's effectively invisible anyway, so it's hard to care about trying to be accurate! | 13:55.28 |
kens | Yep. Of course, I wqanted it small so that it would be invisible...... | 13:55.49 |
henrys | how can fixed customer bugs be 12 in the bug report. We had 11 open customer bugs last time - got 1 more this week, if we fixed 12 we'd be at 0 | 14:11.21 |
kens | maybe they only come out as fixed after marcos verifies them ? | 14:11.50 |
| He did verify a load of them when he came back form holiday | 14:12.03 |
henrys | last week he had 6 fixed customer bugs though... | 14:13.05 |
kens | have to ask marcos really | 14:13.28 |
henrys | yeah something odd about that number | 14:13.38 |
chrisl | henrys: default device fix: http://git.ghostscript.com/?p=user/chrisl/ghostpdl.git;a=commitdiff;h=aa6434d | 14:20.41 |
henrys | chrisl: looks painless | 14:21.29 |
chrisl | It's (very slightly) simpler than before, which is nice | 14:21.56 |
henrys | so meeting time in 2 minutes | 14:28.55 |
Robin_Watts | I have an electrician here, so periodic power cuts. | 14:29.42 |
| I should be OK for the meetings, but if I drop off, accept my apologies in advance. | 14:29.59 |
rayjj | my coffee is full, so I'm ready. | 14:30.09 |
henrys | Robin_Watts: no problem | 14:30.30 |
Robin_Watts | Tomorrow, I expect to be out of power most of the day, so I'll be working on the laptop. | 14:30.32 |
henrys | Robin_Watts: what's going on? | 14:30.44 |
Robin_Watts | New Kitchen | 14:30.51 |
henrys | ah | 14:30.57 |
Robin_Watts | Which involves moving the massively complex fuseboard thing. | 14:31.16 |
henrys | so the 2 big items are release planning and gproof. | 14:31.18 |
Robin_Watts | I believe I've fixed the gproof problems that fred had run into. | 14:31.32 |
| So, he should be clear to keep making progress. | 14:31.48 |
henrys | Robin_Watts, mvrhel_laptop, fredross-perry : so how does this look for Chicago. | 14:32.02 |
| ? | 14:32.04 |
| schedule-wise | 14:32.13 |
mvrhel_laptop | henrys: I should have gsview working with it today assuming so that I can debug any color issues | 14:32.33 |
Robin_Watts | I think Michael had some issues with making it work under gsview, but that's not on the critical path for chicago, right? | 14:32.36 |
| mvrhel_laptop: I committed a fix earlier to the gs code that meant we weren't generating any rgb data. | 14:32.58 |
fredross-perry | Android is working. I need to see what happens in a multi-page document next. | 14:33.18 |
mvrhel_laptop | Robin_Watts: ok | 14:33.26 |
henrys | we do have a top ten issue for digital signatures, I'm wondering if paulgardiner would like a little time off from UI stuff to work on that or if we should find another victim? | 14:33.26 |
| marcosw_: can you explain what "fixed customer bugs" means in the report. | 14:34.22 |
| ? | 14:34.23 |
fredross-perry | : Robin_Watts: youâre going to look at proofing performance? | 14:34.28 |
paulgardiner | digital signature stuff is scary. What is it that we need done? | 14:34.33 |
Robin_Watts | I was going to have a quick look, yes. | 14:34.37 |
henrys | is there a gproof apk I can start playing with? | 14:34.51 |
| see how it looks on my device? | 14:35.12 |
Robin_Watts | henrys: I can upload you one. | 14:35.14 |
tor8 | paulgardiner: Have you looked at LibreSSL? The cleaned up fork of OpenSSL that OpenBSD released recently. | 14:35.44 |
paulgardiner | No I haven't | 14:36.02 |
fredross-perry | Robin_Watts: henrys: let me give Robin sightly newer UI code first | 14:36.20 |
| ⦠in ten minutes | 14:36.27 |
tor8 | paulgardiner: They say they've pruned lots of cruft; might be worth looking into using it. | 14:36.32 |
henrys | paulgardiner: 696132 the multiple sigs per document business | 14:36.55 |
mvrhel_laptop | Robin_Watts: so we still want to go ahead and add the profiles into the gproof format as well as handle them in the gs call in mupdf. I can try to get something together for that today | 14:37.08 |
| It would be nice to have a set of profiles on the android device to use/select from | 14:37.44 |
marcosw_ | henrys: I believe itâs the number of bugs tha were closed within the last week that have a non empty customer field. | 14:38.07 |
mvrhel_laptop | fredross-perry: so in addition to setting proofing resolution you will want to be able to set two different profiles | 14:38.15 |
henrys | chrisl: how are you feeling about that release? | 14:38.20 |
| s/that/the | 14:38.27 |
Robin_Watts | mvrhel_laptop: Yes, but it's demoable without that, I feel. | 14:38.38 |
chrisl | henrys: a little nervous..... | 14:38.39 |
fredross-perry | mvrhel_laptop: let me know the details | 14:38.51 |
henrys | pushing back a month is an option if you aren't fully okay with it. | 14:38.56 |
| perfectly reasonable given the changes this time. | 14:39.15 |
mvrhel_laptop | fredross-perry: lets chat later in the day and we can talk about the gsview ui also | 14:39.21 |
fredross-perry | yes ok | 14:39.31 |
marcosw_ | henrys: and arenât enhancements. | 14:39.31 |
chrisl | henrys: I think we'll be as ready as we can be, I'll try to get it done in Sept | 14:39.40 |
henrys | marcosw_: well you have 12 fixed customer bugs. How can that be? | 14:39.55 |
marcosw_ | good question. I will check. | 14:40.08 |
paulgardiner | I'm completely failing to find 696132 | 14:40.26 |
fredross-perry | Robin_Watts: android.zip in the usual place | 14:40.35 |
henrys | oops 696123 | 14:41.17 |
| http://bugs.ghostscript.com/show_bug.cgi?id=696123 | 14:41.45 |
Robin_Watts | henrys: http://ghostscript.com/~robin/gsproof-dev.apk | 14:41.48 |
henrys | thanks Robin_Watts | 14:42.02 |
chrisl | Robin_Watts: I'm going to add a target that just builds the .so library (and associated symlinks) - I think that makes sense | 14:43.11 |
paulgardiner | henrys: That one might not be too hard to sort out | 14:44.29 |
henrys | tor8: thinking about XFA - it would give us a leg up on linux, there is no way I see to do a tax form on linux with free solution AFAICT | 14:44.35 |
Robin_Watts | chrisl: Cool. | 14:45.02 |
| The libgs.so is named 'libgs.so.9' internally. | 14:45.24 |
| but I could change that by fiddling the flags in the Makefile. | 14:45.36 |
chrisl | Yes, that's the correct name | 14:45.44 |
Robin_Watts | chrisl: Not for android :) | 14:45.54 |
henrys | well a US tax form, not sure if XFA is used in the European forms commonly, it does seem to be getting used by the US govt. | 14:45.58 |
chrisl | Robin_Watts: Well, for *all* other Unix systems | 14:46.13 |
Robin_Watts | henrys: The gsproof-dev.apk I've uploaded there will load PDF files as normal. Hit the (v) symbol, then the proofing symbol (3 colored circles) and the PDF will reload as a 300 dpi proof. | 14:47.01 |
| There is a single funny icon in the top (red square). As you tap that it disables each separation in turn. | 14:47.28 |
jogux | henrys: not sure about other EU countries, but the primary filing method in the UK now is via a government provided web app, no pdfs involved. | 14:47.31 |
Robin_Watts | fred is working on a proper UI now. He can give you better builds as he goes. The main thing is that this shows the speed of rendering/zooming/changing separations and shows that we are behaving pretty much correctly. | 14:48.20 |
henrys | the only other thing I had was to ask folks to go through the workflowy before the upcoming meeting, the usual... | 14:48.23 |
tor8 | henrys: yeah. the forms support in our linux viewer is lacking ... it's what I've been working on fixing the past couple of days. | 14:49.00 |
henrys | Robin_Watts: yeah I should be able to look at it today. I have a samsung note for the demo and I wanted to make sure it worked well. | 14:49.11 |
tor8 | but now I ran into trouble with windows :/ | 14:49.12 |
| henrys: basically, we can't even do regular forms on linux a.t.m. | 14:49.35 |
| fixing that would be a first necessary step, I think | 14:49.42 |
chrisl | Robin_Watts: so, what does Android expect the shared lib to be called? | 14:50.00 |
Robin_Watts | My priorities for the next few days are: 1) profile gsproof (to see if there are obvious simple wins), 2) look at the transparency stuff on SOT. | 14:50.05 |
| chrisl: libgs.so | 14:50.12 |
| No .9 | 14:50.14 |
chrisl | Well, that makes no sense *at all* :-( | 14:50.26 |
| You're supposed to have the major version number to indicated ABI compatibility | 14:51.05 |
Robin_Watts | I think I saw a bug come in that looked like it should be mine. A crash in the downscaler... dunno how that measures up with priorities. | 14:51.07 |
henrys | tor8: okay, we do have the customer request for XFA though ... I imagine windows will be the important platform but I'm not sure about that. | 14:51.40 |
| Robin_Watts: as long as gsproof is on schedule I think it doesn't matter much what comes next. I also assigned some unicode business to you. I don't know if you saw that. | 14:52.46 |
rayjj | Robin_Watts: IMHO, bug 696151 is low priority | 14:52.51 |
fredross-perry | henrys: robin_watts: try thi one http://ghostscript.com/~fred/gsproof-dev.apk more better ui | 14:53.29 |
Robin_Watts | ok, | 14:53.47 |
rayjj | Robin_Watts: it's a user bug, with psd format. Unless it presents with other devices. But downscale is not widely used by our customers, iirc | 14:54.31 |
Robin_Watts | rayjj: No, but it's in a lot of our devices :) | 14:54.55 |
| It's low priority, but might be quick. One to look at if I get bored. | 14:55.18 |
henrys | I don't know that bug bothers me. | 14:55.26 |
Robin_Watts | henrys: gsproof-dev.apk installs as "MuPDF", btw. :) | 14:55.40 |
henrys | a crash is P1 | 14:55.43 |
| so the apk will replace my mupdf app? | 14:56.26 |
Robin_Watts | It will. | 14:56.35 |
henrys | 10/4 | 14:56.43 |
| other meeting stuff? | 14:57.25 |
Robin_Watts | fredross-perry: That's lovely. | 14:57.45 |
fredross-perry | thanks | 14:58.09 |
Robin_Watts | We should probably hide the link/reflow/search icons. | 14:58.14 |
fredross-perry | ok | 14:58.24 |
henrys | fredross-perry: can you let me know when you have new stuff. I'll be doing the demo in Chicago. | 14:58.31 |
| ? | 14:58.34 |
fredross-perry | I was also think of making the popup more modal, so you can select/deselect several seps at once, and then done. | 14:58.59 |
marcosw_ | It looks like casper is stable, so presumably apache using up all the system memory was the problem. I still wnat to figure out why it only started happening last week. | 14:59.04 |
Robin_Watts | fredross-perry: Being picky... why does the icon have CMYK split into a pie chart with M being twice the size of the others? | 14:59.09 |
fredross-perry | henrys: I will | 14:59.12 |
| Robin_Watts: because Iâm not a designer. ;-) | 14:59.38 |
| but point taken | 14:59.54 |
| I used an icon I had laying around and just colored it. I can make something better | 15:00.33 |
chrisl | marcosw_: there seemed, on occasion, to be a *lot* of processes called "view" and "edit" being spawned on casper - which I assume are twiki related? | 15:02.31 |
marcosw_ | chrisl: yes, that was the problem. but why starting last week? | 15:03.30 |
chrisl | marcosw_: possibly we were being targeted | 15:04.10 |
Robin_Watts | Did someone open a twiki session on a dodgy wifi connection? :) | 15:04.13 |
Robin_Watts | whistles gently. | 15:04.22 |
marcosw_ | chrisl: I did look briefly look at the apache logs to see if this was the case but didnât see anything obvious to indicate a ddos. | 15:05.11 |
chrisl | shrugs..... | 15:06.29 |
Robin_Watts | fredross-perry: I am intimately familiar with the "not a designer" status :) | 15:07.16 |
jogux | are we up to date with twiki security patches? | 15:07.27 |
Robin_Watts | he he. he. | 15:07.37 |
| jogux so funny. | 15:07.43 |
fredross-perry | he. | 15:08.04 |
jogux | if we're not it would probably be reasonable to assume someone evil now has shell access to casper | 15:08.12 |
marcosw_ | Can anyone create a twiki account and create/edit pages? | 15:08.18 |
jogux | (until proved otherwise) | 15:08.19 |
Robin_Watts | marcosw_: Anyone can create an account. Anyone can edit the public webs. | 15:08.51 |
| (IIRC) | 15:08.59 |
marcosw_ | Robin_Watts: is there a way to see if there were many accounts or new pages created last week? | 15:10.01 |
Robin_Watts | http://twiki.ghostscript.com/do/view/Main/UserList | 15:10.27 |
jogux | I just looked through the history pages for each of the webs and didn't see anything obviously untoward looking | 15:11.04 |
marcosw_ | and http://twiki.ghostscript.com/do/view/Sandbox/WebStatistics doesnât show abnormal activity for August | 15:11.49 |
| though why the page http://twiki.ghostscript.com/do/view/Sandbox/WebPreferences should have had 258 views is puzzling. | 15:12.54 |
| when Iâm back from vacation Iâll look at the apache logs more carefully, if it really only took 258 twiki page views in total to run out of memory on 3 or 4 occasions I would have not identified that has a ddos. | 15:15.01 |
jogux | marcosw: just sent you an email on this subject. | 15:15.04 |
| (I didn't want to share the link on the channel...) | 15:15.16 |
marcosw_ | jogux: okay, Iâll take a look at that. thx. | 15:15.53 |
jogux | it sounds bad :-( | 15:16.06 |
marcosw_ | jogux: I tried the test for vulnerability on the link you sent and did not get back Vulnerable!. otoh, I may have misunderstood how to run the test. | 15:19.37 |
jogux | marcosw_: you got further than me then :-) odd. | 15:20.07 |
Robin_Watts | I get an 'internal error' | 15:23.21 |
marcosw_ | Robin_Watts: thatâs what I see, so presumably not vulnerable. | 15:23.40 |
| I have to run, waterfalls and gorges await (also, more importantly, latte). I will look at the high number of customer bug closes tonight and the casper crashes on Friday when Iâm back home (the former should be easy). | 15:26.50 |
Robin_Watts | marcosw_: Where are you? | 15:27.09 |
| I see: /home/robin/crap/TWiki-6.0.0.tgz | 15:29.56 |
| Which makes me think it was v6 that I installed :) | 15:30.04 |
jogux | 6.0.0 is what I see in the twiki configure page | 15:31.17 |
marcosw_ | Robin_Watts: portland | 15:34.50 |
| we came home for Scotland on Thursday and left for Portland on Saturday. | 15:35.43 |
| (waiting for daughter to get dressed, so may be here a while yet). | 15:36.04 |
Robin_Watts | oregan? Never been to oregan. | 15:37.04 |
marcosw_ | Robin_Watts: I was here for a day once with Artifex. Scott, Miles, and I visited Tektronix (I think I can say that in public, it was 15+ years ago and nothing became of it). | 15:38.13 |
fredross-perry | Robin_Watts: Iâm not getting good results with multi-page docs. compare http://ghostscript.com/~fred/cmyk.pdf with http://ghostscript.com/~fred/cmyk2.pdf | 15:39.18 |
tor8 | Robin_Watts: fatal error C1001: internal error in the compiler in MSVC 2005 link.exe when linking mupdf in release mode ... anything familiar to you? | 15:41.42 |
jogux | tor8: that randomly triggers an answer of something like do you have VS2005 SP2 installed. | 15:46.43 |
| (iirc microsoft update is stupid and doesn't provide SP2, you have to go find it manually) | 15:47.01 |
rayjj | marcosw: I also went to Tek twice with Scott and Miles. Somewhat more recently (last time was > 6 yrs ago). Same results | 15:47.20 |
fredross-perry | henrys: robin_watts: i updated http://ghostscript.com/~fred/gsproof-dev.apk. it now presents a popup where you can check/uncheck multiple separations, and they say âdoneâ. And it sports a more symmetric icon. ;-) | 16:46.14 |
| it works with single-page docs, like tiger. Not multipage docs yet. | 16:47.27 |
mvrhel_laptop | oh so mupdf windows is still broken? | 17:34.32 |
Robin_Watts | mvrhel_laptop: Not if you review the commits on robin/master :) | 17:35.15 |
mvrhel_laptop | I did | 17:35.21 |
Robin_Watts | Sorry, I missed that. You were happy? | 17:35.38 |
mvrhel_laptop | did you not read the logs :) | 17:35.38 |
| I was happy | 17:35.44 |
Robin_Watts | mvrhel_laptop: Been a zoo here today, sorry. | 17:35.50 |
| Pushed. | 17:36.03 |
mvrhel_laptop | no problem is was yesterday ;) | 17:36.05 |
Robin_Watts | It should be fixed now. Thanks. | 17:36.07 |
mvrhel_laptop | thanks | 17:36.16 |
| Robin_Watts: by the way, I wrote a script to generate the winrt project files in gsview now | 17:37.19 |
| so you no longer have to go and do all the wacky setting manually | 17:37.33 |
Robin_Watts | Ah, sensational. | 17:37.57 |
fredross-perry | mvrhel_laptop: are you available for a UI chat? | 18:12.16 |
mvrhel_laptop | fredross-perry: yes. let me grab my head set if you want to just talk on skype | 18:48.22 |
fredross-perry | yes, skype | 18:48.31 |
Robin_Watts | is here if you need non GSview specific gproof discussion. | 18:49.07 |
| Ok. So I can speed the gproof rendering up by a factor of 4 by using Z_BEST_SPEED rather than Z_BEST_COMPRESSION. | 18:51.44 |
| Obviously that's more disc space, but I don't think we care. | 18:51.54 |
fredross-perry | Robin_Watts: that sounds good. | 19:04.53 |
Robin_Watts | New libgs.so in the usual place. | 19:05.34 |
fredross-perry | let me know if youâve checked that in; I am going to work on gsview proofing also so I need to guild gs for mac | 19:06.05 |
henrys | fredross-perry, mvrhel_laptop I think you know but the android stuff has a hard deadline gsview does not. | 19:06.58 |
fredross-perry | yes. Do you have an idea what might be amiss with multipage docs in Android? | 19:07.48 |
Robin_Watts | fredross-perry: Yes, checked in. | 19:08.15 |
| No, what's up with multi-page docs? | 19:08.26 |
fredross-perry | When I open a doc thatâs got two or more pages, the separation operations seem to do nothing. | 19:08.59 |
| compare http://ghostscript.com/~fred/cmyk.pdf with http://ghostscript.com/~fred/cmyk2.pdf | 19:09.19 |
Robin_Watts | fredross-perry: Leave that with me, I'll have a look. Will probably be tomorrow before I get any results though. | 19:09.41 |
fredross-perry | Thatâs fine, thanks. | 19:09.57 |
Robin_Watts | fredross-perry: I can confirm that I see no changes with the separation stuff for multipage docs. | 19:21.17 |
fredross-perry | when I doit I only see changes on page 2 of a 2-page doc. | 19:23.30 |
| OK, so maybe this is just confusion about whatâs the current page. | 19:25.06 |
Robin_Watts | yeah. | 19:25.29 |
fredross-perry | After I create the .gproof file and view it, I am looking at page 2 (which I did not notice). | 19:25.51 |
| So maybe the thing to do is to remember the current page, and then go there after creating the gproof file. | 19:26.47 |
Robin_Watts | Certainly we should jump to the same page that we were on when we started the proofing process. | 19:27.17 |
| That's something my quick first version ignored. | 19:27.30 |
| But I don't think that explains why changes aren't appearing when you toggle separations. | 19:27.58 |
fredross-perry | also I am seeing that maybe controlSepOnPage() applies separately to different pages? | 19:28.29 |
Robin_Watts | Do you have an up to date android.zip I can steal your source from? | 19:28.35 |
| fredross-perry: Yes. | 19:28.44 |
| In a PDF file every page can have different separations. | 19:28.53 |
fredross-perry | So I have to account for that in the UI, no big. | 19:28.58 |
Robin_Watts | yeah. I may need to do stuff in the core for that too. | 19:29.11 |
| cos we only keep 3 pages loaded at once. | 19:29.22 |
| The moment we move out of that 3 page range, the fz_page is discarded, and hence the selected separations are reset. | 19:29.42 |
| Possibly we should just say "separations are reset when you change page". | 19:29.58 |
| Gotta go, boss is calling me for dinner. | 19:30.21 |
fredross-perry | oh right. So if I could have isSepEnabledOnPage(index) or some such, that would help. Then the UI doesnât really casre. | 19:30.44 |
| lunch! | 19:30.50 |
mvrhel_laptop | lunch | 20:26.02 |
Robin_Watts | fredross-perry: I was thinking that the existing functions would always be working on the current page. | 22:04.53 |
fredross-perry | Robin_Watts: look at the bottom of ReaderView.onLayout2(). Itâs handling the current page, and then the one that follows. When we call controlSepOnPage() at some point after that, I think itâs affecting th last page rendered, which is page 2, not page 1. Thoughts? | 22:30.16 |
| I think the simplest thing to do is to pass a page number to getNumSepsOnPage(), controlSepOnPage(), and getSep(). Assuming that fz_* is holding onto that stuff. | 22:47.46 |
| Forward 1 day (to 2015/08/19)>>> | |