| <<<Back 1 day (to 2014/06/15) | 2014/06/16 |
kens | chrisl ping | 07:40.14 |
chrisl | kens: pong | 07:41.27 |
kens | chris I'm seeing a significant percformance penalty for the binaries I build myself releative to the ones you build, can you tell me what command line you use to build GS ? (I'm using the IDE atm) | 07:42.08 |
chrisl | nmake -f psi\msvc.mak DEVSTUDIO= BUILD_SYSTEM=64 (WIN64=1) | 07:42.56 |
| And nmake -f psi\msvc.mak DEVSTUDIO= BUILD_SYSTEM=64 (WIN64=1) nsis - for the installer build | 07:43.15 |
kens | Hmm, I was using the 32-bit lis | 07:43.20 |
| 32-bit binaries... | 07:43.25 |
| I'll rebuild using nmake and see what happens | 07:43.35 |
chrisl | kens: what version of VS are you using? | 07:44.52 |
kens | 2008 | 07:44.56 |
chrisl | Hmm, I use 2005 for the release binaries | 07:45.11 |
Robin_Watts | 2005 FTW. | 07:45.14 |
kens | As an example, the 9.14 release, run 1000 times, executing 'quit' averages 0.18 seconds start to finish. THe libraries I build average 0.24 | 07:45.59 |
| aha, and built using hte nmake command line, takes 0.17 | 07:47.38 |
| Interesting.... | 07:47.47 |
| Anyway, thanks for that, I'll go rerun the tests on master now :-) | 07:48.08 |
chrisl | That's, um, odd..... | 07:48.36 |
kens | It could be I have somethgin peculiar in my VS setup | 07:48.52 |
| Since I don't do releases I don;t really bother to keep it prstine | 07:49.12 |
chrisl | You're not using the project from git? | 07:49.44 |
kens | Just switching back to master form teh 9.14 tag, so I can check the result there (I was bisecting to find the slowdown) | 07:49.51 |
| chrisl no | 07:49.56 |
chrisl | need some shopping - back in 20 mins....... | 08:09.24 |
pra | hii.......... can anyone help me to do copy a annotation and create a new annotation using existing one? | 09:10.28 |
| hiii anyone there? | 09:12.53 |
kens | Yes. | 09:13.00 |
pra | can anyone help me to do copy a annotation and create a new annotation using existing one? | 09:13.16 |
kens | You haven't said which application. | 09:13.31 |
pra | in mupdf....... windows version | 09:13.53 |
kens | THen I can';t help,you will have to wait until one of teh MuPDF developers comes online | 09:14.12 |
pra | who is MuPDF developers? | 09:15.00 |
kens | There are several, why do you need to know ? | 09:15.21 |
pra | I am working on mupdf improvement | 09:15.53 |
kens | Then you are in the right place | 09:16.14 |
pra | I am already write a function which can create an annotaion....... | 09:16.45 |
kens | There's really no point in talking to me about it, I can't help you | 09:17.04 |
pra | So, who is main MuPDF developer? | 09:17.49 |
kens | Why do you need to know ? You are in the correct channel, this channel is logged and the developers read the logs. When they come online they will read any questions and respond here. If you aren't online 24/7 or logging this channel yourself you should check back in our logs when you return to see if there were any replies | 09:19.39 |
pra | ok..... | 09:20.14 |
kens | chrisl my timings indicate that the current master code is no slower on startup than 9.05 (within the timer resolution they are the same) and the performance is as good on all the files the customer used, in some cases better than 9.05. | 09:44.26 |
chrisl | kens: cool! My testing suggested the current code was still *slightly* slower, but hopefully close enough...... | 09:45.16 |
kens | For me they are the same (using nmake to build) and the processing is slightly faster overall. I'm attempting to do a bisect from curiosity to see what caused the (large) performance hit on some files with 9.09, and also the (modest) startup cost, again introduced in 9.09 | 09:46.29 |
| I'm not getting anywhere though, Git keeps on muttering about merges, so I'm going to give up. | 09:48.07 |
chrisl | merges? odd..... | 09:48.40 |
kens | Yeah it keeps saying somethign about 'merge needs testing' or something | 09:49.03 |
chrisl | Branch root needs testing, or something like that. It'll just handle it, though | 09:49.27 |
kens | It doesn't handle it, that's for sure, bisect stops and tells me the fault exists between to commits (and givces the commits) | 09:49.53 |
chrisl | That's probably not related to the branching warning | 09:50.34 |
kens | Its something to do with merges, but I don't know what, and to be honest, I don't think I care enough to chase it any further | 09:51.03 |
| I 'suspect' that its the introduction of the getdeviceparams stuff. | 09:51.19 |
| Which is now (of course) fixed. | 09:51.32 |
| Hmm, no it can't have been that, I did that back in March 2011 | 09:53.22 |
| OK the error message is 'bisecting a megre base must be tested' | 09:56.09 |
jogux | kens: I donât think thatâs an error as such, more an âokay, I gave you something slightly different from normal, but I just need you to go ahead and test it and do a git gisect good/bad on itâ. | 09:59.43 |
kens | jogux, I htought that, butthen it got *really* upset afterwards | 10:01.12 |
| Of course, that could have been the failure to remove a directory.... | 10:01.24 |
jogux | :( | 10:01.28 |
kens | I may have a solution | 10:01.52 |
| It gave methe SHA for the tow commits, so I'm trying to bisect between those 2 | 10:02.14 |
chrisl | kens: all that means is that one of the commits you want to test is on a branch, so it will test the parent of commit of the branch first to see if the change is on the branch, or on master - then it will continue the bisect based on the result of that | 10:02.18 |
kens | chrisl, OK but then, like I said, it refused to bisect further | 10:02.39 |
| Gave me the two SHA though | 10:02.54 |
chrisl | I'm not sure that's related, you *always* get that merge base warning when bisecting with a release tag because the release tag is on a branch | 10:03.42 |
kens | Hmm, I didn't think I was using the release tag by then, but.... | 10:04.16 |
| Looks like the two symptoms are unrelated, so I may have to do 2 bisects as well | 10:04.48 |
chrisl | You said you were looking at the 9.09 code, which I assume you got from the tag | 10:05.09 |
kens | Yes, but I'd moved away from that | 10:05.26 |
| I started with the 9.07 and 9.09 tags | 10:05.35 |
| But swiftly moved on | 10:05.40 |
| Its behaving itself now | 10:05.53 |
| I just need to decide whether this is good data or bad data :-( | 10:06.10 |
chrisl | If the SHAs were on different branches, you'd get the warning, too | 10:06.38 |
kens | It seems to be happy now so I'mprogressing. I'll have to rerun it to find the initialisation slowdown, I'mchansing the processing slowdown first | 10:07.20 |
mattchz | Hi all | 10:40.46 |
| Iâm Matt, and Iâm working with emobix on mupdf | 10:41.02 |
kens | Hi Matt | 10:41.10 |
mattchz | The Android readme says to use the r14 SDK on Windows, and on the Mac an older version. Is there a recommended version to use? | 11:07.30 |
| (on the mac) | 11:07.39 |
kens | sorry Matt, I don't know, you probably need Robin_Watts to answer that one | 11:14.28 |
mattchz | ah, okay, np, thanks! | 11:14.40 |
kens | wonders where Robin_Watts is today | 11:14.56 |
mattchz | I spoke to him earlier; heâs having to sort some issues out with a workman at home | 11:15.20 |
kens | OK thanks | 11:15.36 |
mattchz | He said heâll be back in a bit | 11:16.32 |
kens | chrisl the slowdown in 9.09 for pdfwrite was caused by commit ecc7a19 'Bug 693488: Improved glyph name alisasing' | 11:16.45 |
chrisl | kens: that's what marcosw concluded, too | 11:18.45 |
kens | chrisl as far as I can tell that's *not* the commit which caused the slowdown in initialisation | 11:19.17 |
| That one only affected certain files, and it was a prcoessing hit, not an initialisation hit | 11:19.36 |
chrisl | Oh, that's odd.... it *should* have had some impact on initialisation | 11:20.02 |
kens | Oh it may well have had 'some' effect, but the primary effect I see is that a particular PDF file runs at half the speed to the commit preceding it | 11:20.34 |
| I've still got to look into the intialisation change | 11:20.57 |
sebras | thinks Robin_Watts and tor8 have both run awol | 11:21.58 |
mattchz | just off for some lunch, bbiab | 11:22.50 |
jogux | mattchz: tbh I suspect both Robin and the other likely Androider Paul will have been building on Windows :-) | 11:22.55 |
Robin_Watts | mattchz: Back! | 11:31.13 |
| mattchz: I think the notes say "this is how I have been building" rather than "this is how you should build" | 11:32.02 |
| so later versions of the SDK are probably fine. | 11:32.10 |
Robin_Watts | feels sure that mattchz already has a version installed- just use that unless you have problems. | 11:32.52 |
miyalys | I'm using mupdf and wanted to display the page number. Ultimately I found out it's shown if you press Shift P (or a large P), but this doesn't seem to appear in its man page, even though many other keybindings are listed which I assumed are all of them? At least it appears so in archlinux? Shouldn't it be added? | 11:34.20 |
Robin_Watts | miyalys: Possibly. Feel free to update the man page and submit us a patch at bugs.ghostscript.com | 11:34.59 |
miyalys | Right. | 11:35.38 |
jkroon | hi, i'm not sure this is the place to ask about this, since i'm not convinced this is a ghostscript issue, but alas, it's the closest clue I've got. | 11:38.05 |
kens | chrisl yes it seems that commit affects both the initialisation time *and* the processing time. | 11:38.31 |
jkroon | when printing from an application called "quasar", even when printing to a file I end up with this is the generated .ps file: No embeddable font for Times-Bold found | 11:38.36 |
kens | jkroon how is Ghostscript involved in this ? What operating system are you using ? | 11:39.12 |
jkroon | now, when comparing the header of the .ps file from a system where the print looks correct there is no such "No embeddable font" message, and the actual fonts differ. | 11:39.14 |
| gentoo linux. | 11:39.19 |
chrisl | kens: well, it does extra stuff for every type 1 and TTF font (from disk) we load | 11:39.20 |
kens | chrisl yes I was assuming that was the reason. | 11:39.33 |
| jkroon Linux printing is kind of out of my territory, but its 'unlikely' Ghostscript is produing the PostScript file. Not impossible, but you need to look into the print papth | 11:40.18 |
sebras | jkroon: is quasar this accounting software from linuxcanada.com? | 11:40.42 |
jkroon | kens, kens - as stated, i'm not sure it's a gs issue. but I've no idea where to start and since I suspect the people here know more than I do about this stuff in general ... | 11:40.42 |
kens | However it sounds like you just don't have a Times-Bold font installed on your system | 11:40.43 |
jkroon | sebras, yes. | 11:40.47 |
| how can I trace the printing path? They still stuck on Qt3 ... | 11:41.13 |
kens | jkroon can't you take this up with the supplied of the accounting software ? | 11:41.36 |
jkroon | which afaik uses CUPS, which does utilize ghostscript in some form or another. | 11:41.36 |
| kens, i've tried, but will try again. | 11:41.54 |
kens | jkroon its true that CUPS 'may' use Ghostscript, but if you print to file I very much doubt that ti does | 11:42.00 |
jkroon | kens, even when printing to a .ps file? | 11:42.17 |
kens | To me it sounds lik eyou just don't have a Times-Bold font installed | 11:42.20 |
sebras | jkroon: can you print from another program (that also uses ghostscript) on the same computer? | 11:42.24 |
kens | jkroon Ghostscript *consumes* PostScript, it doesn't manufacture it | 11:42.33 |
jkroon | sebras, that's an affirmative. | 11:42.34 |
| kens, ah ok. that explains it better for me thanks. | 11:42.50 |
chrisl | jkroon: look in the cups log files: /var/log/cups | 11:42.57 |
jkroon | as stated - i'm not 100 % sure what to dig it. | 11:43.02 |
| chrisl, will start there then thanks :) | 11:43.11 |
| and sorry for going completely off-topic. | 11:43.18 |
kens | Not a problem | 11:43.26 |
| We do try to help where we can, but other people's software can be as much of a mystery to us as to you :-) | 11:43.45 |
chrisl | jkroon: but, to be honest, it sounds like the software in question is emitting it's own Postscript, and can't find a Times-Bold font to embed in the output | 11:43.47 |
jkroon | %%Creator: Qt 3.3.8b <-- just spotted this :) so yea. wrong location here. | 11:43.52 |
| on the working machine it seems to embed NimbusRomNo9L-Medi NimbusRomNo9L-Regu and NimbusSanL-Regu ... which I do have. | 11:44.21 |
| so something else is going wrong. | 11:44.28 |
chrisl | jkroon: are those fonts on the same paths on both machines? And known to fontconfig? | 11:45.40 |
jkroon | kens, i completely understand. i'm not asking to get people to solve my problems on my behalf, but very often someone else will just spot (or know) something that I don't that just gets one back on the right path. | 11:45.44 |
kens | D'oh, network dies | 11:45.47 |
jkroon | chrisl, as far as I know - how do I confirm? | 11:46.03 |
chrisl | Well, I assume you know how to check the paths..... | 11:46.23 |
kens | jkroon can you post a PostScript file for us to take a quick look at ? I cna tell quickly if we made it. Please don't post it here, you can send us a URL, and please make very sure its not private! | 11:46.27 |
jkroon | kens, let me generate one without actual financial data in it by pre-dating it quick. | 11:46.54 |
kens | THa't'll be fine, thanks | 11:47.04 |
chrisl | jkroon: the fc-list command will list all the fonts known to fontconfig (note: I don't actually know if Qt uses fontconfig....) | 11:48.10 |
jkroon | chrisl, yes, both machines has the same list for "Nimbus Roman No9" (grep No9), a more general grep Roman however does not match. | 11:50.18 |
| kens, http://www.uls.co.za/quasar-tauri.ps (working) http://www.uls.co.za/quasar-greyscale.ps (broken) | 11:53.05 |
kens | OK That PostScritp was definitely *not* produced by Ghostscript, it looks pretty much like it was produced by the application itself. | 11:54.10 |
| Its pretty clear that its Qt producing the output behind the scenes | 11:54.37 |
jkroon | or by Qt3 ... | 11:54.54 |
kens | I'd have to go with what CHris and I said earlier, the application can't find the Times family of fonts, so it can't embed them. | 11:56.03 |
jkroon | ok, i now know this is not the right place to ask, but is it possible to point me somewhere where I might be able to get more information? | 11:56.05 |
| that makes sense. | 11:56.16 |
kens | TO be honest, I think the supplied of the software is your only likely recourse | 11:56.22 |
chrisl | The Qt print developers? | 11:56.26 |
kens | supplier* | 11:56.29 |
jkroon | kens, yea, that does make sense. or I need to figure out which fonts I require. | 11:56.46 |
kens | suspects Trolltech won't be able to help, presumably Qt needs to be told where the fonts are, and presumably that's done by the application (I could be wrong....) | 11:57.09 |
chrisl | I would expect there to be a system wide configuration for Qt, as well as app specific | 11:57.53 |
kens | I guess that makes sense | 11:58.02 |
chrisl | But I'm not immediately finding any documentation on Qt font substitution configuration | 11:58.38 |
kens | According to SO 'Qt uses your window system as a font provider' | 11:58.47 |
| Tihs is embedded Linux,but it might be relevant: | 11:59.21 |
| http://doc.qt.digia.com/4.6/qt-embedded-deployment.html | 11:59.21 |
chrisl | I doubt it uses X for fonts | 11:59.31 |
kens | If you find where Qt is isntalled, then look for a sub-directory /lib/fonts there may be a fontdir file. | 12:00.05 |
| Apparently there's also a QT_QWS_FOTNDIR environment variable which should poin to the fonts (presumably the QT /lib/fonts folder) | 12:01.39 |
jkroon | thanks guys | 12:04.34 |
| i'll keep digging. | 12:04.37 |
kens | I have to go out and buy a body protector, will be back in a bit. | 12:09.38 |
jkroon | kens, problem was exactly that the Times and Helvetica fonts *are* on the broken system, they're found and used, but since they're bitmap fonts they are not embedded. | 12:24.10 |
| uninstall those and they too use the more sane fallbacks. | 12:24.28 |
| on-screen display of the application itself though now sucks. | 12:24.40 |
chrisl | jkroon: you may want to look into whether you can enable hinting and/or anti-aliased rendering for fonts | 12:30.56 |
purn | hii.... anyone there? who can help me on annotation problem? | 12:38.55 |
Robin_Watts | purn: Ask the question, don't ask to ask. | 12:43.59 |
mattchz | robin_watts: ta. Yes, I have the latest version installed. Iâll try that, thanks. Iâll use the old version of the NDK though, as I know that can be a bit more senstitve | 12:50.44 |
Robin_Watts | I think someone was building with r9 the other day. | 12:51.04 |
| I'd try it and see. | 12:51.08 |
mattchz | okey doke. | 12:51.15 |
Robin_Watts | In general, I'd like to ensure that it builds with the latest versions. | 12:51.26 |
mattchz | indeed; sounds a good plan | 12:51.47 |
purn | hi robin_watts: how can I create a new annotaion using a existing annotation property? | 13:22.06 |
Robin_Watts | purn: On what platform? For what app? | 13:25.11 |
purn | on windows platform. | 13:25.29 |
| I want to create a new Stamp annotion | 13:26.22 |
| using a existing one | 13:26.45 |
Robin_Watts | I'm not sure we have support for Stamp annotations, currently. | 13:26.58 |
purn | So, can I copy a existing stamp annotation and past at one another place........... | 13:28.12 |
| is it possible? | 13:28.22 |
Robin_Watts | anything is possibly using the underlying pdf manipulation functions. | 13:28.35 |
| but in terms of pdf_annot functions, it's more limited. | 13:28.53 |
| You may be able to extend the pdf annotation stuff to include stamp support, in which case, we would love to see a patch... | 13:29.19 |
purn | How can I do this? | 13:30.50 |
| Hoe to extend the pdf annotation stuff to include stamp support? | 13:31.21 |
Robin_Watts | purn: That's a non trivial question, and one paulgardiner is probably better equipped to answer than me. | 13:32.35 |
| BUT... both paulgardiner and myself are buried under a mound of other work at the moment... | 13:32.52 |
purn | So, can you suggest me how to copy anyone annotation and past it mannualy in another place using any fuction?????/ | 13:34.49 |
Robin_Watts | purn: How much do you understand of the underlying PDF format? | 13:35.16 |
purn | function means internal any function? | 13:35.20 |
| Now I can create a new type on annotion in my end..... | 13:36.08 |
| like line, text | 13:36.18 |
| I am familiar with AP part...... but how to copy a Stream part | 13:37.06 |
| In case of Stream part I am facing some problem | 13:37.35 |
Robin_Watts | I'm sorry, I just don't have time for this at the moment. | 13:38.43 |
purn | I am understand so many thing about PDF file structure.................. But I don't understand how to copy any one annotation | 13:40.46 |
paulgardiner_lap | purn: the function pdf_create_annotation can create any type of annotation, but will not set up any details. Luckily for stamp annotations, the only additional step is to create the AP. | 13:40.52 |
| The usual way to create thge AP is via pdf_set_annot_appearance. That takes a display list, which in turn can created by making mupdf graphics calls into a mupdf display-list device | 13:42.45 |
purn | I am already use this pdf_set_annot_appearance for line annotation | 13:44.02 |
paulgardiner_lap | Typically, if I've wanted to copy the AP from an existing file, I've loaded the pdf graphics commands from the stream into vi and edited it into a set of mupdf graphics calls | 13:44.32 |
| purn: okay, then same technique but with more complicated graphics commands to create the display list | 13:45.31 |
| That's as much as I can help for now. As Robin said we're snowed under with work on other things at the moment. | 13:46.10 |
purn | I am understand the use of pdf_set_annot_appearance function. But in case of stamp we have to create BBA object | 13:46.22 |
| IT may be only BBA or have to create another object BBA1 | 13:47.47 |
| Each object have Stream information | 13:48.33 |
| hi.........paulgardiner_lap: How to create Stream information of each object of Stamp annotation | 14:02.52 |
| How to genarate the stream information | 14:03.25 |
| Is there any documentation of Mupdf?????????/ | 14:30.51 |
Robin_Watts | purn: The docs directory. | 14:32.08 |
| but nothing pertaining to annotations. | 14:32.13 |
sebras | Robin_Watts: I guess the future jni-api ought to be really well documented. | 14:35.07 |
Robin_Watts | sebras: Indeed. | 14:35.15 |
| sebras: Meet mattchz. He's an old friend who is doing some contract work on MuPDF for us. | 14:36.05 |
henrys | Robin_Watts: purn looks commercial to me ? | 14:36.20 |
mattchz | Hi Sebras | 14:36.32 |
| Nice to meet you. | 14:36.38 |
sebras | mattchz: hi! likewise! | 14:36.44 |
Robin_Watts | specifically Android and iOS UI work to start with. | 14:36.46 |
| Ah, henrys, didn't see you here. | 14:36.54 |
| mattchz: henrys is the PHB. | 14:37.14 |
mattchz | Heh, hi henrys | 14:37.28 |
henrys | hi mattchz | 14:37.35 |
sebras | Robin_Watts: mattchz: aha, nice. are you going to simply fix bugs in that area or are you going to do the entire interface overhaul as well..? | 14:37.41 |
| or maybe this has not been decided yet. :) | 14:37.50 |
Robin_Watts | henrys: I didn't pick up the commercial vibe from him... | 14:37.56 |
| sebras: I don't think the android app needs an interface overhaul, does it? | 14:38.54 |
| The ios one might - I don't do ios. | 14:39.00 |
sebras | Robin_Watts: there might be an inverse correlation between the number of question marks and the level of commercialness..? | 14:39.10 |
| Robin_Watts: API-interface, not UI-interface. | 14:39.29 |
| Robin_Watts: at least that's something tor8 has mentioned a few times to me (and that no one has had time to work on it of course). | 14:39.59 |
Robin_Watts | sebras: Ah, right. I don't think that's been decided yet. | 14:40.05 |
sebras | Robin_Watts: ok. | 14:40.34 |
Robin_Watts | does the Tor summoning dance | 14:41.20 |
kens | Not seen tor all day | 14:43.47 |
| I suspect a more powerful rite maybe necessary | 14:44.06 |
henrys | Robin_Watts, kens vacation 16 - 27 | 14:44.39 |
kens | Then a *very* powerful summoning will be required :-) | 14:45.03 |
Robin_Watts | henrys: oh, he's away? That's... inconvenient. | 14:45.35 |
kens | I think he said he'd check email daily if its really important | 14:45.56 |
henrys | he copied in tech miles and joann | 14:45.58 |
Robin_Watts | I was hoping that Tor would be pointing Matt at the most important MuPDF ios/Android UI problems. | 14:46.09 |
henrys | I already went through that with jogux I thought | 14:46.42 |
| after talking to tor | 14:46.52 |
Robin_Watts | oh, ok, great. | 14:47.00 |
mattchz | Iâve got the bugzilla list of bugs | 14:47.04 |
Robin_Watts | Ah. right, I wasn't aware that had been taken care of. | 14:47.28 |
jogux | henrys : yep, matt has the list/instructions we talked about | 14:47.35 |
sebras | tor5: welcome! | 14:48.10 |
tor5 | hi Robin_Watts (and sebras) | 14:48.20 |
Robin_Watts | tor5: Sorry, didn't realise you were on holiday. | 14:48.31 |
mattchz | â[10:04pm] henrys: jogux: start with any crash. looks like several of | 14:48.34 |
| these are not really bugs at all, like 693719 | 14:48.34 |
| [10:06pm] henrys: jogux: anything that isnât a UI problem should | 14:48.35 |
| probably have the Hardware field changed to âallâ just to keep your list | 14:48.36 |
| simple" | 14:48.37 |
Robin_Watts | Was just going to introduce to mattchz. | 14:48.48 |
| and to ask what the most important bugs for him to fix were, but I think that may have been sorted already. | 14:49.03 |
mattchz | hi tor5! | 14:49.05 |
tor5 | Robin_Watts: np. I'm just sitting at home sniveling from a cold... just perfect timing. | 14:49.07 |
jogux | tor5: :( | 14:49.13 |
Robin_Watts | urgh :( | 14:49.18 |
mattchz | :( | 14:49.19 |
henrys | tor5: change your vacation days ! | 14:49.32 |
tor5 | the worst seems to have passed, I expect to be back to my normal self by tomorrow. shame about losing the weekend though. | 14:50.19 |
| mattchz: hi! | 14:50.30 |
mattchz | hi jogux | 14:51.10 |
jogux | robin: Iâve not used the Android mupdf (yet) but as a seasoned iOS user there are a few tweaks etc we could make I think there. Not sure theyâre fully formed ideas I could write down yet though. | 14:52.26 |
tor5 | mattchz: Robin_Watts: I haven't looked at the android bugs in enough detail to be able to say anything about priorities there. | 14:52.27 |
henrys | paulgardiner, Robin_Watts : perhaps you two should review the bug list mattchz has and see if something important is missing. | 14:53.08 |
| tor5 has been through it. | 14:53.32 |
Robin_Watts | Well, if mattchz has stuff to be getting on with from tor5, I think he should do that first. | 14:53.54 |
mattchz | I was looking at this list: | 14:54.32 |
| http://bugs.ghostscript.com/buglist.cgi?columnlist=product%2Ccomponent%2Cassigned_to%2Cbug_status%2Cresolution%2Cshort_desc%2Cchangeddate%2Cpriority&list_id=15649&op_sys=iOS%204&op_sys=iOS%205&op_sys=iOS%206&op_sys=iOS%207&op_sys=Android&order=priority%2Cbug_severity&product=MuPDF&query_based_on=&query_format=advanced&resolution= | 14:54.33 |
| oh, weird. did that link come through? | 14:54.58 |
jogux | is not sure it came through right as itâs showing resolved bugs for me | 14:55.13 |
mattchz | ok, my irc client is doing very odd things. | 14:55.17 |
kens | works for mer | 14:56.03 |
paulgardiner_lap | hi mattchz | 14:56.08 |
tor5 | I have saved searches for mobile, 31 bugs + 7 enhancements | 14:56.09 |
Robin_Watts | http://bugs.ghostscript.com/buglist.cgi?bug_status=UNCONFIRMED&bug_status=CONFIRMED&bug_status=IN_PROGRESS&columnlist=product%2Ccomponent%2Cassigned_to%2Cbug_status%2Cresolution%2Cshort_desc%2Cchangeddate%2Cpriority&list_id=15799&op_sys=iOS%204&op_sys=iOS%205&op_sys=iOS%206&op_sys=iOS%207&op_sys=Android&product=MuPDF&query_format=advanced maybe? | 14:56.39 |
tor5 | just filtered on os=iOS *, Android | 14:56.39 |
henrys | mattchz: yes your list seems to include fixed bugs. | 14:56.49 |
paulgardiner_lap | Nothing that's missing occurs to me so far | 14:56.51 |
tor5 | Robin_Watts: yeah, that one matches what I have | 14:57.03 |
paulgardiner_lap | mattchz: Did jogux mention moving the current Android app from the old Play account to the new account? | 14:57.54 |
jogux | no :-) does henrys want us to pick that up too? | 14:58.16 |
Robin_Watts | paulgardiner_lap: That sounds like an SOT thing, not a MuPDF thing ? | 14:58.32 |
paulgardiner_lap | jogux: I'm probably confused. I thought you suggested so. | 14:59.17 |
mattchz | Hi paul | 14:59.22 |
jogux | we should move the iOS mupdf from Torâs account into the new artifex one (probably). | 15:00.09 |
mattchz | Robinâs link looks good - filters out the fixed bugs | 15:00.20 |
paulgardiner_lap | I expect there were several conversations going and I misunderstood what you said "Matt could do that too" to | 15:00.21 |
jogux | doesnât remember the conversation :-S | 15:01.33 |
tor5 | jogux: yes, someone should do that. | 15:01.59 |
| (move the mupdf app from my account into the new artifex one) | 15:02.11 |
henrys | jogux, tor5 agreed | 15:02.21 |
jogux | okay, weâll stick that on Mattâs list then? shouldnât take too long. | 15:03.36 |
paulgardiner_lap | jogux: It was SOT related. You passed me a link to a page describing how to move apps from one account to another, but the part regarding Matt was a misunderstanding I'm sure (or I entirely imagined it :-) ) | 15:03.39 |
jogux | paul: I remember that part :-) Iâd have thought youâd have done it by now ;-) | 15:04.02 |
Robin_Watts | mattchz has committed his first fix already :) | 15:07.23 |
paulgardiner_lap | No... since I'd hidden it within an SEP field | 15:07.23 |
mattchz | it was an easy one :) | 15:07.42 |
| so should I just pick any crash from that list? | 15:08.43 |
Robin_Watts | mattchz: yeah. | 15:09.26 |
mattchz | k, cool | 15:09.38 |
rayjj | henrys: who is going to do the gsview website ? I've put a dummy up on casper /var/www/gsview.com that I cloned from the ghostxps.com as a placeholder (and to make sure that it's working) | 15:17.28 |
| can someone in .eu try http://www.gsview.com ? | 15:17.51 |
Robin_Watts | yup. works. | 15:18.01 |
rayjj | Robin_Watts: thanks | 15:18.05 |
kens | agreed | 15:18.06 |
rayjj | kens: thank you, too | 15:18.18 |
henrys | rayjj: sort of falls into chrisl_awayâs jurisdiction I suppose | 15:18.30 |
rayjj | henrys: I didn't know if Miles was going to have an 'expert' do one (with input from mvrhel) | 15:19.05 |
henrys | rayjj: I think zoltan is after you again | 15:19.27 |
rayjj | in any case "my job here is done" | 15:19.31 |
henrys | rayjj: indeed we could have ron look at it. | 15:19.42 |
tor5 | Robin_Watts: mattchz: is there anything more you need from me now, before I log off? | 15:19.46 |
henrys | rayjj: wow big commit | 15:20.08 |
Robin_Watts | tor5: No. You go get better/enjoy your break. | 15:20.11 |
mattchz | tor5: no, I think weâre good thanks. I hope you feel better! | 15:20.32 |
tor5 | Robin_Watts: mattchz: thanks. bye! | 15:20.47 |
rayjj | henrys: yes, I saw. Same strange question. How do to private PS hooks for transparency work. That is a black hole of questions that I am loathe to even start since each response will surely lead to more questions | 15:20.53 |
mattchz | tor5: bye | 15:21.01 |
rayjj | henrys: I'll get to him after cust 532's issue with performance. Getting close to what I hope will be a major speedup | 15:21.30 |
kens | Seems to me they are undocumented GS-specific extensions that he *should not* use | 15:21.36 |
Robin_Watts | kens: I thought they *were* documented in the language.htm stuff as extensions? | 15:22.14 |
henrys | rayjj: I think that is above and beyond if youâd like Iâll respond with a request for money to field that which should scare him away. | 15:22.20 |
kens | Robin_Watts : then we should remove the docs :-) | 15:22.28 |
Robin_Watts | Does matt need special bugzilla permissions to be able to assign himself bugs etc? | 15:24.21 |
rayjj | kens: oh, I didn't even realize there were docs there. | 15:24.41 |
kens | Robin_Watts : pointed it out, not me.... | 15:24.55 |
| But I still think we should remove the documentation, or say 'don;t use these, thaty are not part of the API and subject to change without notice, we will not support use of these opedrators by 3rd parties' | 15:25.57 |
kens | has no inclination to support Zoltan trying to do bizarre stuff with these operators | 15:27.35 |
| Hmm, one of Ray's commits triggered a new seg fault | 15:30.37 |
rayjj | kens: It did ? I had just run a regression and the only one that showed up was the old one Bug695090.ps | 15:34.29 |
kens | ***************************************************************************** | 15:34.47 |
| Seg faults with current rev: | 15:34.47 |
| tests_private/comparefiles/Bug688845.eps.pkmraw.300.1 plancks Seg_Fault | 15:34.47 |
rayjj | kens: I wonder why that didn't show up for my run. I'll have a look... | 15:35.20 |
kens | ordinarily I'd say indeterminism, but the seg faults have been pretty stable for ages | 15:35.45 |
rayjj | kens: doesn't fail for me on win 32, trying linux 64-bit now | 15:41.21 |
kens | jjst goes by the emails.... | 15:41.57 |
rayjj | kens: thanks for spotting that. | 15:41.59 |
| henrys: kens: I sent Zoltan a short reply, telling him that he's on his own, and also pointing out that if he had looked at the code he wouldn't have needed to ask that particular question. | 15:52.38 |
kens | :-) | 15:52.47 |
rayjj | henrys: I hope it's not too impolite | 15:52.51 |
kens | I think we need to be clear this is 'not for meddling with' | 15:53.19 |
rayjj | but given the time frame during which these guys used the product without paying ... | 15:53.33 |
| hmm... doesn't fail on peeved either (at least a debug build). Trying a release build next. | 15:55.39 |
| chrisl: I assume you saw in the logs that you are responsible for the gsview.com website ;-) | 15:57.20 |
| (chrisl probably wishes he had stayed "away") | 15:57.48 |
chrisl | rayjj: I'm conscientiously ignoring it because I don't want to do it....... | 15:57.51 |
rayjj | chrisl: sounds like a plan. :-) I think mvrhel wants to do it :-) | 15:58.20 |
Robin_Watts | :would volunteer, but feels he won't be allowed :) | 15:58.55 |
chrisl | rayjj: you may well right about mvrhel :-) Besides, there's not really anything to do until nearer September | 15:59.05 |
| Robin_Watts: whatever you did in a past life, can you find out what it was, so I can avoid it?? | 15:59.35 |
rayjj | chrisl: that's a point, talking about a product that isn't yet released may be pointless. I'll add something to the existing site that mentions that we plan a release in September, and check back closer to then | 16:00.17 |
| that much html I can do :-) | 16:00.41 |
chrisl | rayjj: I guess it also depends on if mvrhel wants to do a beta release before Sept | 16:01.03 |
rayjj | darn. a release build doesn't segfault either -- trying the actual cluster binary that in on there | 16:02.43 |
kens | Hmm, maybe its just a weird indeterminism | 16:03.25 |
| Off for dinner now, goodnight all | 16:16.42 |
mattchz | how do I build the android code with memento enabled? I tried ndk-build MEMENTO=1 NDK_DEBUG=1, but that failed | 16:23.30 |
| make: *** No rule to make target `jni/../../../fitz/memento.c', needed by `obj/local/armeabi-v7a/objs/mupdfcore/__/__/__/fitz/memento.o'. Stop. | 16:23.46 |
Robin_Watts | mattchz: Urm... | 16:24.28 |
| That should have worked. | 16:25.40 |
mattchz | Hmm, I didnât build the generated stuff with MEMENTO. | 16:25.47 |
| not sure if thatâs relevant. | 16:25.57 |
Robin_Watts | doesn't matter | 16:25.59 |
| Fails for me too. | 16:27.26 |
| In Core.mk | 16:27.33 |
| there is an ifdef MEMENTO | 16:27.41 |
| Ah, that's the problem. | 16:27.58 |
| LOCAL_SRC_FILES += $(MY_ROOT)/fitz/memento.c | 16:28.05 |
rayjj | chrisl: I updated the gsview.com site slightly to mention that it is not yet released and there will be a beta sometime prior to the release planned for Sept | 16:28.07 |
Robin_Watts | should be | 16:28.08 |
| LOCAL_SRC_FILES += $(MY_ROOT)/source/fitz/memento.c | 16:28.14 |
rayjj | chrisl: feel free to change it' | 16:28.21 |
chrisl | rayjj: no, sounds fine, thank you | 16:28.30 |
mattchz | robin> ah, that fixes it, thanks! | 16:31.08 |
Robin_Watts | mattchz: Bug 694967.... I'd guess that glo is probably NULL. | 16:33.27 |
| glo is the globals object. | 16:33.43 |
mattchz | Right, yeah. I wondered if it was happening when things were shutting down | 16:33.57 |
| as it happens when pressing the back button. | 16:34.20 |
henrys | rayjj: sorry lost internet briefly, Iâm fine with the zoltan response | 16:34.26 |
mattchz | robin: that seems likely, as onDestroy() in MuPDFCore is setting globals to 0 | 16:40.22 |
Robin_Watts | mattchz: Possibly the correct response is to spot glo being NULL and bale out then. | 16:40.55 |
mattchz | cool | 16:41.37 |
| although, it looks like that is probably getting called from an async task, so Iâm guessing that async task has not been shutdown properly when the document is closed? | 16:43.21 |
Robin_Watts | possibly. | 16:44.01 |
| That would be a java question, and hence once that paul may be in a better position to answer. | 16:44.24 |
| but it's possible that we can't easily kill outstanding async tasks. | 16:44.42 |
| In which case having the async task just exit would be a good solution. | 16:44.57 |
mattchz | yeah. it looks like we try and cancel it possibly. I canât remember offhand if itâs guaranteed to work though. Iâll check the AsyncTask API | 16:45.29 |
| and is if by magic, the shopkeeper appeared! | 16:45.54 |
Robin_Watts | paulgardiner_lap: See the recent logs :) | 16:46.08 |
paulgardiner | mattchz: highly likely. The documentation says something like "this is too hard to explain, please look at this app". I never did. | 16:46.53 |
mattchz | heh | 16:47.00 |
| Iâve had my own angst with AsyncTask in the past. Itâs not a nice API. | 16:47.30 |
paulgardiner | I was always expecting to see crashes due to not shutting them down, but never did (until now of course). I guess we'll need to wait for the background thread. I doubt we can nuke it. | 16:48.18 |
mattchz | Iâll add some logging to confirm this is actually the problem. | 16:48.53 |
| I worry that just checking for NULL and bailing out would just cause the crash to occur later (if the globals object is deleted while the task is still running) | 16:49.54 |
Robin_Watts | mattchz: The globals object is fundamentally linked to the MuPDFCore object, IIRC. | 16:51.30 |
mattchz | ah, so wonât be deleted. | 16:51.49 |
Robin_Watts | If there is no globals object, then the C can't do anything. | 16:51.58 |
| If there is no MuPDFCore object then the java can't do anything. | 16:52.08 |
| so, basically in C, we should bale if there is no glo, and in java we bale if there is no Core. | 16:52.32 |
| Possibly you can see a nicer way to achieve what I've done there. | 16:52.49 |
| I was learning JNI from StackOverflow pretty much at that point :) | 16:53.25 |
mattchz | heh. You almost certainly know more than I do about JNI tbh :) | 16:53.43 |
Robin_Watts | (Hmm, that implies I'm past the SO stage now, and sadly... probably not :( ) | 16:53.59 |
mattchz | :) | 16:54.19 |
pedro_pc | we should just do OpenGL UI and use an android NativeActivity for the apps ;) | 16:54.50 |
mattchz | or UE2 | 17:03.35 |
| good night all | 17:38.29 |
| Forward 1 day (to 2014/06/17)>>> | |