| | <<<Back 1 day (to 2013/06/03) | 2013/06/04 |
| ray_laptop | well, that's a surprise. My change for using fill adjust with all clipping paths shows NO bmpcmp differences. | 00:09.55 |
| | committing.... | 00:10.01 |
| | done. e0ba422 Fix clipping bugs 693509 and 690036. | 00:12.58 |
| | I'll wait for the "final" commit regression before closing bug 693509 -- bug 690036 is already closed as a duplicate. | 00:13.55 |
| | Interesting that the regression testing showed so many differences, but bmpcmp didn't. I hope my commit is OK, but at the most I expect 1 pixel diferences. | 01:14.49 |
| henrys | marcosw:looks like we've gone a month without a Bug Report or am I missing them somehow? | 01:49.18 |
| | hmm I guess the report should be html so links can be clicked | 04:20.45 |
| kens | chrisl ping | 07:31.30 |
| chrisl | kens: pong | 07:34.12 |
| kens | chrisl can you tell me where Marcos is keeping these 'fuzzing' files ? I can't remember.... | 07:34.31 |
| chrisl | they are in the svn-private repo..... | 07:34.52 |
| kens | ah, so I can get them from Peeves ? | 07:35.02 |
| | I don't think I have subversionset up here | 07:35.24 |
| chrisl | svn-private/ghostpcl/trunk/tests_private/fuzzing | 07:35.30 |
| kens | Hmm svn-private doesn't seem ot be on Peeves | 07:36.28 |
| | OK No Subversion and I can't find the source directory, any clues ? | 07:37.09 |
| tor8 | kens: svn co on casper and scp the files | 07:38.03 |
| kens | tor8 no subversion | 07:38.14 |
| chrisl | Erm, get subversion? | 07:38.21 |
| tor8 | or use git-svn with a shallow clone, but if you don't have svn I don't think git-svn will work | 07:38.22 |
| kens | Oh onCasper, that sounds bad | 07:38.23 |
| tor8 | kens: svn export will checkout without a repository, if you just want the files | 07:38.43 |
| kens | Why on earth did Marcos stick them in subversion..... | 07:38.46 |
| chrisl | kens: that's the policy we *all* decided on many moons ago | 07:39.10 |
| kens | I gvuess I wasn't paying attention. I thought we stopped using Subversion so I'm not set up for it. | 07:39.42 |
| chrisl | kens: since you can't checkout a sub-repo in git, and we rarely want the *entire* test repo, it was thought best to keep it in subversion. | 07:40.43 |
| kens | would prefer that it be available in a directory somewhere.... | 07:41.03 |
| chrisl | kens: don't you have svn in msys? | 07:41.14 |
| kens | chrisl not as far as I'm aware | 07:41.27 |
| tor8 | msys-git comes with "git svn" at least on my install it did | 07:41.46 |
| chrisl | If you want to e-mail a list of file(s) you want, I can stick them up on casper | 07:41.51 |
| kens | Well svn in msys says 'command not found' | 07:41.51 |
| | chrisl thanks, give me a minute while I compile them, its only about 5 files | 07:42.14 |
| | I'm sure I used ot be able to get files out of subversion somewhere without actually running SVN.... | 07:43.26 |
| chrisl | You can probably get the "normal" regression tests from the cluster nodes, but the fuzzing tests aren't part of the normal cluster test, so aren't pulled onto every node | 07:44.39 |
| kens | No I mean I actually pulled them out of the Subversion directory without running SVN, I just can't remember how | 07:46.40 |
| chrisl | kens: on casper : /home/chrisl/fuzzing-20130604.zip | 07:52.14 |
| sebras | chrisl: who created those fuzzing files? are they pdfs? have they been run though mupdf? | 07:54.19 |
| chrisl | sebras: they come from a group at google, I believe, and I guess they ran them through mupdf - we got a bunch of files from them for both mupdf and gs. | 07:55.35 |
| sebras | chrisl: aha. do you know how they created them? are they private or public? | 07:58.19 |
| kens | chrisl, thanks | 08:04.13 |
| | chrisl I have the file but I can't delete the original (not the owner obviously). Feel free to get rid of it, and thanks again | 08:06.46 |
| chrisl | kens: cool | 08:10.33 |
| | sebras: I believe the tests were created with a fuzzing tool of google's own, but I'm not sure. I *believe* there is nothing private in the files, they're in svn-private simply because we (and google) don't want the security implications spread over the net! But don't quote me on any of that - Robin may know more | 08:12.00 |
| kens | Looking at Alex's patch I think it needs to be modified, there are three possible returns; error, not present and found, but the code only caters for found or not found. Also there are several other places with teh same fault | 08:12.04 |
| chrisl | kens: was it perhaps the intention that the "error" case be treated as "not found"? | 08:13.04 |
| kens | chrisl, I don't think so, it tries to add broken data to a dicitonary I think,but I need to run the files to be sure, which is what I'mdoing now | 08:13.36 |
| chrisl | kens: I just wondered if treating error was missing was intended behaviour, but implemented wrongly | 08:14.19 |
| kens | No, I think its an oversight, whoever did it didn't realise there was a possible error return I believe | 08:14.43 |
| | At the very least its dangerous and may lead to more subtle problems. | 08:15.12 |
| | THe fault is that the page number is -2.... | 08:16.14 |
| sebras | chrisl: ok. I'm just curious, because if there are pdfs that cause problems that I can look at then I'm happy to do so. | 08:27.39 |
| | chrisl: also it could be interesting to run the fuzzing tool on another set of pdfs, so if we know what tool was used and how it was configured we could do some more testing. | 08:28.36 |
| | chrisl: I have the zzuf tool, which does fuzz-testing, but it has no context-knowledge so it just does random fuzzing which take unnecessary execution time. | 08:29.31 |
| chrisl | sebras: I don't think we can get the fuzzing tool, and I think the mupdf ones are all either fixed or in-hand - but check with Robin | 08:29.59 |
| kens | At least one of the tools was 'address sanitizer' I believe, but I don't think that did the actual fuzzing | 08:30.32 |
| sebras | chrisl: ok, will do. thanks. :) | 08:31.15 |
| Robin_Watts | kens: On peeves, look in /home/marcos/cluster/tests_private/ | 09:40.37 |
| | there is a fuzzing dir in there. | 09:40.41 |
| kens | Robin_Watts : thanks, if I need more I'll look there | 09:40.56 |
| kens | lunches | 11:20.53 |
| | tor8 Robin_Watts another one on Stack Overflow. I'm not really sure what he wants, though it seems like a daft question. Also prossibly another one needing a pointer to the GPL: | 13:15.17 |
| | http://stackoverflow.com/questions/16917787/how-to-view-multiple-pdfs-in-a-single-pdf-in-mupdf-pageview-android | 13:15.17 |
| Robin_Watts | https://my.applause.com/?app_id=com.artifex.mupdfdemo | 13:19.21 |
| kens | OK I'mbaffled, what does it do ? | 13:20.31 |
| | ah, it has comments, they work | 13:20.48 |
| Robin_Watts | kens: It collates reviews/comments etc to let you track your app. | 13:22.34 |
| kens | what puzzled me is the graph on entry, which seems very bare | 13:22.52 |
| Robin_Watts | kens: I suspect it will make more sense when it's been running for longer. | 13:24.13 |
| kens | Yeah that could be,the other bits seem more sensible :-) | 13:24.31 |
| | anyone know how to read valgrind logs ? | 13:33.54 |
| Robin_Watts | kens: ish. | 13:34.50 |
| kens | OK I've got a log which I *think* I know what its telling me, but I'm not certain, jst a second. | 13:35.15 |
| | ==22390== Invalid read of size 8 | 13:35.30 |
| | ==22390== at 0x7F074D: pdf_compute_BaseFont (gdevpdtf.c:855) | 13:35.30 |
| | ==22390== by 0x801388: write_font_resources (gdevpdtw.c:655) | 13:35.30 |
| | ==22390== by 0x7A58D6: pdf_close (gdevpdf.c:2298) | 13:35.30 |
| | ==22390== by 0x937C72: gs_closedevice (gsdevice.c:647) | 13:35.30 |
| | ==22390== by 0x535D76: gs_main_finit (imain.c:867) | 13:35.30 |
| | ==22390== by 0x5361B4: gs_to_exit_with_code (imain.c:931) | 13:35.30 |
| | ==22390== by 0x468527: main (gs.c:139) | 13:35.31 |
| | ==22390== Address 0x9f5a550 is 3,280 bytes inside a block of size 20,048 free'd | 13:35.31 |
| | ==22390== at 0x4C24D68: free (vg_replace_malloc.c:325) | 13:35.32 |
| | ==22390== by 0x951E3B: gs_heap_free_object (gsmalloc.c:347) | 13:35.32 |
| | ==22390== by 0x9271FB: alloc_free_chunk (gsalloc.c:1998) | 13:35.33 |
| tor8 | the first stack trace is where it bombed, the second is a stack trace from when the memory was allocated/freed | 13:36.12 |
| kens | I think that's telling me that pdf_compute_BaseFont was trying to read from free'd memory and that the memory was free'd in gs_closedevice | 13:36.16 |
| | tor8 that's what I was thinking, thanks | 13:36.26 |
| Robin_Watts | kens: you would be right. | 13:36.29 |
| kens | So now I have to track down why the memory was freed but the resource was not discarded (I think) | 13:37.05 |
| | Kind of an odd sequence.... | 13:37.22 |
| | and of course its a large file with many fonts. | 13:37.52 |
| tor8 | kens: I don't follow... how did you read that trace to deduce that the memory was freed in gs_closedevice? | 13:42.29 |
| sebras | Robin_Watts: kens: but in the stacktrace there is no evidence that gs_closedevice() freed the memory...? | 13:42.42 |
| tor8 | the free trace is only 3 lines | 13:42.44 |
| kens | from gs_closedevice | 13:42.44 |
| tor8 | (unless irc truncated your paste) | 13:42.59 |
| kens | gs_closedevice calls pdf_close etc | 13:43.06 |
| | Oh yes, freenode did truncate it | 13:43.16 |
| | 'Throttled due to flooding' | 13:43.25 |
| | Having said that, I think valgrind is wrong anyway | 13:43.43 |
| | THe memory that's freed is not related to the font, but that probably means it was relocated by the garbage collector or something | 13:44.06 |
| sebras | kens: valgrind might also truncate stacktraces sometimes. --num-callers=50 is really handy then. | 13:44.15 |
| | kens: if you can run gs without its own allocator it probably would make sense. we have to do the same for gstreamer (which relies on glibs slice-allocator e.g.) | 13:45.01 |
| kens | You can't run GS without its own allocater | 13:45.14 |
| | Sorry allocaters, plural | 13:45.26 |
| | Well, the Val;grind warning is incorrect, at least for my current source, under Windows. I strongly suspect that the memory has been relocated which is either confusing Valgrind or making it impossible for me to see the problem because I don't have the same version of source running on the same OS etc. | 13:53.21 |
| | ROFL never seen this one before : | 13:55.43 |
| | ==26718== Warning: silly arg (-393168) to malloc() | 13:55.43 |
| sebras | kens: (for the logs) I believe that it is possible to instrument allocator code to inform valgrin what is going on. it might be worthwhile to do that for gs perhaps. | 14:13.44 |
| tor8 | henrys: Robin_Watts: paulgardiner: I may be a few minutes late for the mupdf meeting today. | 14:23.21 |
| henrys | okay | 14:23.50 |
| Robin_Watts | sebras: In what way? You mean marking stuff as undefined/defined within our own allocators? | 14:25.17 |
| | MuPDF does that. | 14:25.28 |
| | sorry, Memento does that. | 14:25.37 |
| | but I'm not sure the chunk allocator does. | 14:25.49 |
| | It means we need to mark block headers as accessible/unaccessible as we walk through them. | 14:26.18 |
| | henrys: A couple of topics for the meeting agenda... | 14:27.26 |
| | 1) Get sebras an NDA. | 14:27.32 |
| | 2) Investigate the NVidia path rendering on GPU stuff. | 14:27.51 |
| henrys | yes I spoke with him privately earlier | 14:27.55 |
| | and added that. | 14:28.03 |
| Robin_Watts | oh, ok. | 14:28.10 |
| | Also, the agenda works better in meetings when items are numbered, I think? Otherwise we lose track of where we are? | 14:28.49 |
| sebras | I need to go, but I'll be sure to check the logs when I get home. | 14:31.43 |
| henrys | Robin_Watts:yes the agenda is a work in progress right now. I'm actually using an online outliner which allows me to cross stuff off which fouls numbering, but yes I'll think of something. | 14:33.01 |
| | I could use an md5 sum ;-) | 14:34.15 |
| Robin_Watts | henrys: put it in a shared git repo. | 14:38.52 |
| henrys | It wouldn't be a problem if I broke it into subtopics, which I'd prefer, but that promotes snoozing ;-) | 14:39.02 |
| Robin_Watts | I'm confused by bug 694290 | 14:40.45 |
| | I can't reproduce it. Marcos says in comment 8 that it's been fixed. | 14:41.09 |
| | so why does he then ask me to try to figure out where it has the problem in comment 10 ? | 14:41.27 |
| kens | No he means the /Nums problem is fixed | 14:41.28 |
| | the reporter has 2 problems in one bug | 14:41.38 |
| Robin_Watts | kens: right. | 14:41.45 |
| kens | I oculd reproduce the crash | 14:41.52 |
| Robin_Watts | I'm trying on peeves and failing to make it crash. | 14:42.07 |
| kens | crashed on WIndows for me | 14:46.00 |
| henrys | I'm fine skipping meetings today we'll have plenty of meeting in Miami. | 14:56.17 |
| kens | Sounds good to me.... The only things I have to discuss are on the agenda | 14:56.39 |
| henrys | the only thing I'd say is let's rush to get stuff checked in, steady pace is best. | 14:56.55 |
| Robin_Watts | henrys: I was hoping to talk about the mupdf fuzzing work that shelly and zeniko did. | 14:57.11 |
| | kens: won't reproduce in either windows or on peeves for me. | 14:58.05 |
| kens | Well it was a slightly earlier version of the code, just a minute and I'll try with my current build | 14:58.26 |
| Robin_Watts | what SHA were you testing on ? | 14:58.28 |
| | Thanks. | 14:58.30 |
| henrys | Robin_Watts:good enough if you, alexcher_ and I talk about it? | 14:58.37 |
| Robin_Watts | henrys: sure. | 14:58.43 |
| henrys | alexcher_:are you here yet? | 14:59.00 |
| | we may have to wait for the ghostscript meeting time, Robin_Watts | 14:59.56 |
| Robin_Watts | essentially, we asked shelly/simon to look at the remaining fuzzing SEGVs from mupdf (which we believed to all be in openjpeg2) | 15:00.02 |
| | We have 1 SEGV left that we know of, in the jbig2 stuff. | 15:00.22 |
| tor8 | henrys: we'll snooze regardless ;) | 15:00.40 |
| Robin_Watts | So I think they've fulfilled their remit, and we should sort out paying them some sort of bounty for their efforts. | 15:00.53 |
| | The majority of the fixes came from zeniko, but shelly was working on openjpeg2 integration. | 15:02.20 |
| kens | Robin_Watts : current code works fine, I have no idea why | 15:02.28 |
| Robin_Watts | so both deserve something. | 15:02.33 |
| | kens: I'll try to backtrack a bit, | 15:02.52 |
| kens | a week or so should be enough | 15:03.02 |
| Robin_Watts | I'd be tempted to kick the remaining jbig2dec bug to shelly and to pay them both a bounty. | 15:03.32 |
| henrys | Robin_Watts:I have been watching and it "feels" like 1000 and 1500.00 (shelly and zeniko) is that about right? | 15:03.32 |
| Robin_Watts | I don't think either could complain at that being ungenerous. | 15:03.59 |
| henrys | okay I'll arrange that with miles. | 15:05.21 |
| tor8 | henrys: mupdf meeting time? | 15:05.59 |
| Robin_Watts | tor8: henrys just said to skip the meetings as we have miami looming. | 15:06.24 |
| | tor8: did you have anything for the meeting? | 15:06.34 |
| tor8 | just sebras huge list of gpl violations on android marketplace | 15:06.52 |
| henrys | if there is reason to have the meeting I'm here. | 15:06.56 |
| | tor8:I'll add that to the agenda. | 15:07.18 |
| | one good thing about mobile is we won't see many dodging the GPL with forked process. | 15:08.06 |
| tor8 | like I mentioned yesterday to kens, it seems a lot of the dumb "how do I use my computer" questions regarding mupdf on android on stackoverflow are probably related | 15:08.54 |
| | henrys: a few of those seem to be cropping up as well | 15:09.02 |
| | one of the more popular apps that used to be gpl is going closed source and using a forked process to dodge the gpl... | 15:09.40 |
| | ebookdroid I think it wa | 15:09.47 |
| | s | 15:09.48 |
| mvrhel_laptop | sorry I am a bit late | 15:10.07 |
| | oh no meetings? | 15:10.52 |
| henrys | tor8:really I thought that would too heavyweight on mobile | 15:11.03 |
| tor8 | so we should probably probe all the stackoverflow "halp me plz fix! urgent! how do I compile mupdf on android??? I am clueless!" people about whether they're aware of licensing issues | 15:11.14 |
| mvrhel_laptop | I had a couple questions. | 15:11.16 |
| henrys | mvrhel_laptop: right - but go ahead. | 15:11.25 |
| Robin_Watts | our expectations of the capabilities of mobile are changing radically. | 15:11.36 |
| henrys | mvrhel_laptop: we want to get your stuff cleared for the windows store. Are you ready? | 15:12.36 |
| mvrhel_laptop | so once I get the things that Robin_Watts mentioned to me (about the not checking for !=NULL) and figure out my open xps issue (obviously something I am doing here), and test that all is well when running on the device itself are we fine with submitting to the store? | 15:12.45 |
| | henrys: other than checking and fixing those things, yes I think so | 15:13.05 |
| Robin_Watts | mvrhel_laptop: The not checking for != NULL thing was just an observation. | 15:13.31 |
| | don't feel you need to change your code purely for my sensibilities :) | 15:13.42 |
| mvrhel_laptop | Robin_Watts: no it is cleaner. I already fixed it :) | 15:13.53 |
| henrys | mvrhel_laptop: is there a packaging scheme like droid? | 15:13.58 |
| | you should post the package and let someone else test the final thing | 15:14.19 |
| mvrhel_laptop | yes. I have to go through the whole process. I will copy the info that we have from the droid app | 15:14.21 |
| | oh. good idea. I | 15:14.40 |
| | I will start to go through the process and see if I can package and have someone test henrys | 15:15.04 |
| | next question is, do we want to make a windows mobile version too | 15:16.15 |
| henrys | post it to tech and ask for testing. We'll talk about folks getting windows 8 at the meeting in Miami, but for now I think a few people can give it a go.. | 15:16.25 |
| mvrhel_laptop | it should not be too hard, but it has to be written in c# as opposed to c++ :( | 15:16.47 |
| henrys | ugh | 15:17.09 |
| mvrhel_laptop | at least the ui part. I should still be able to call into the winrt library | 15:17.11 |
| | as I have it set up | 15:17.18 |
| tor8 | Robin_Watts: mudraw on xps files leaks images according to valgrind :( | 15:17.30 |
| mvrhel_laptop | I really don't think it will be too dificult | 15:17.33 |
| henrys | why do we have the c++ version at all then? | 15:17.44 |
| mvrhel_laptop | well, you want to write the winrt stuff in c++ | 15:18.07 |
| Robin_Watts | tor8: bugger. will look. | 15:18.23 |
| mvrhel_laptop | and you can call from c++, javascript, c# | 15:18.32 |
| henrys | mvrhel_laptop:I guess that was my question we either need 3 languages for the new windows ecosystem or 2 (C# and C), 2 is better than 3. | 15:19.26 |
| mvrhel_laptop | writing the UI portion in xaml/C# will be pretty easy | 15:19.28 |
| | henrys: I see. I suppose, I could have written the windows 8 viewer in c#, but I still would need to write new UI code for the windows mobile device. This way at least we demonstrate that yes indeed from our mupdf winrt library we can create application in the various languages | 15:22.53 |
| | so I don't see it being too much of an issue. | 15:23.17 |
| Robin_Watts | The core lib is in C. | 15:23.34 |
| henrys | mvrhel_laptop: fair enough. | 15:23.49 |
| Robin_Watts | AIUI in order to expose the API to other things we need a small wrapper veneer. That needs to be in C++, right? | 15:24.02 |
| mvrhel_laptop | yes. and the winrt interface to it is in c++. And one can write UI front ends in c++, c# and javascript to call into the winrt interface | 15:24.15 |
| Robin_Watts | And then apps can be written in any language we like. | 15:24.22 |
| mvrhel_laptop | Robin_Watts: yes. | 15:24.24 |
| Robin_Watts | right, so we are tied to C/C++ for the core winrt libs, but app writers are free to choose. | 15:24.52 |
| henrys | Robin_Watts:my point we could use C# in lieu of C++ and have one evil language to deal with ;-) not 2 | 15:24.57 |
| mvrhel_laptop | c++ is not as evil as c# | 15:25.10 |
| Robin_Watts | Can we write the veneer in C# ? | 15:25.28 |
| mvrhel_laptop | but I think it is good to demonstrate the use of both. | 15:25.39 |
| | Robin_Watts: no I think the winrt stuff has to be in c++ | 15:25.52 |
| Robin_Watts | C# can call anything that's been wrapped up in the right way - but we need to use C++ to wrap our C. | 15:25.54 |
| mvrhel_laptop | but I would need to double check that | 15:25.59 |
| Robin_Watts | right, that's what I thought, hence I think henrys' assertion is wrong. | 15:26.07 |
| mvrhel_laptop | Robin_Watts: ah I see your logic | 15:26.22 |
| tor8 | these wrappers, do they have to be classes and stuff or why can't you just call the C directly? | 15:26.23 |
| mvrhel_laptop | even if I had written the viewer in C# I would have had c++ anyway | 15:26.40 |
| | tor8: they have to be very special winrt types | 15:26.55 |
| | not even standard c++ types | 15:27.03 |
| Robin_Watts | tor8: AIUI, in order for the magic interoperability to work, it needs to be an object based interface. | 15:27.17 |
| tor8 | yuck. but oh well. | 15:27.31 |
| mvrhel_laptop | that part is done and working though, so getting a windows mobile solution working should be pretty easy | 15:28.18 |
| | but we can chat about it at the meeting | 15:28.33 |
| tor8 | the cross-language interfaces to mupdf (jni and winrt, and maybe more in the future?) should probably be designed and documented. something for the agenda? | 15:28.46 |
| mvrhel_laptop | yes. that would be a good idea | 15:29.16 |
| henrys | okay the cancelled meeting has 1 minute left ;-) | 15:29.56 |
| mvrhel_laptop | :) | 15:30.02 |
| | henrys: so can you add windows mobile to the agenda? | 15:30.16 |
| tor8 | and foreign language mupdf interfaces. | 15:30.35 |
| henrys | yes and yes | 15:30.47 |
| Robin_Watts | foreign language mupdf interfaces? | 15:31.03 |
| | LeMuPDF ? | 15:31.09 |
| mvrhel_laptop | oui oui | 15:31.16 |
| tor8 | Robin_Watts: precis! | 15:31.32 |
| henrys | mvrhel_laptop: I'm a bit confused, didn't you have something running on the surface? | 15:31.34 |
| mvrhel_laptop | yes. that is working fine. but is different than windows phone.... | 15:31.53 |
| | sad but true | 15:32.06 |
| Robin_Watts | gotta love ms. | 15:32.09 |
| henrys | i.e. I was just lumping windowrt viewer all in the same category and I guess it should be split up. | 15:32.27 |
| Robin_Watts | everyone else is having problems with fragmentation of the device space. So we should design our new system to have the same problems! | 15:32.37 |
| tor8 | jni, winrt, objc, and some basic set of apis that look similar for them all | 15:32.45 |
| mvrhel_laptop | so we really have 3 windows things | 15:32.47 |
| | desktop, windows 8, windows 8 mobile | 15:32.58 |
| | the surface falls in the windows 8 category | 15:33.07 |
| henrys | okay I'll fix the agenda. | 15:33.46 |
| Robin_Watts | kens, ray_laptop: It seems that it was rays clipping change that's made this stop crashing. | 15:34.49 |
| kens | THat's kind of odd | 15:35.48 |
| henrys | whilst many of you are now here (and for the logs) I was going to suggest Artifex spend a bit more money and have meetings on weekdays. I won't suggest it if there are loud objections. I think family obligations are most likely to fall on a Saturday or Sunday. I'm definitely in the doghouse going out of town this weekend, summer art market for Sabrina. Thoughts? | 15:36.16 |
| kens | henrys its really a lot more for us to fly not over a weekend. THree time sthe price as I recall | 15:37.20 |
| Robin_Watts | henrys: The europeans would still HAVE to stay over a saturday night. | 15:37.39 |
| kens | We don't ahev to stay a saturday robin | 15:37.53 |
| | Its just way more expensive if we don't | 15:38.03 |
| Robin_Watts | kens: Unless we want to pay 4x the price. | 15:38.08 |
| kens | yes, that's what I was saying :-) | 15:38.18 |
| | THough I thought it was 3 times, but .... | 15:38.28 |
| henrys | oh I didn't know it was that much. | 15:38.35 |
| Robin_Watts | 600 vs 2400 last time I looked. | 15:38.48 |
| kens | yeah its like £1500 vs £500 | 15:38.48 |
| | well RObin may be right | 15:38.54 |
| | Its really a lot | 15:39.02 |
| henrys | okay well we don't want to do that. | 15:39.26 |
| Robin_Watts | So either miles would need to pay a lot more (and there are 5 of us now), or we'd need to stay extra nights (which costs more in hotel bills) | 15:39.43 |
| | Idea: Do US folks pay the stupid saturday night premium when they fly to london? :) | 15:42.06 |
| mvrhel_laptop | I don't believe so | 15:42.20 |
| Robin_Watts | All future meetings to be held here then! Result! | 15:42.35 |
| mvrhel_laptop | hehe | 15:42.41 |
| | I don't mind the weekend meetings for the most part. | 15:43.06 |
| | as long as we have plenty of notice | 15:43.16 |
| Robin_Watts | I'm still trying to parse how "avoiding family obligations" is a bad thing... | 15:43.31 |
| mvrhel_laptop | hehe | 15:43.37 |
| henrys | I see denver to london weekday for about 1200 US | 15:43.54 |
| | and don't see a change for the weekend, I'm just using google flight though | 15:44.51 |
| Robin_Watts | ok, to and from sf in december, looking at virgin atlantic. | 15:45.04 |
| kens | henrys miles sadi that too | 15:45.14 |
| | Its not the case here | 15:45.19 |
| Robin_Watts | 483.75 if we stay saturday night, 1514.75 if we don't. | 15:45.21 |
| kens | right 3 time slike I said | 15:45.32 |
| henrys | insane | 15:45.42 |
| Robin_Watts | sorry, that was for JFK not SFO. Don't know how I did that. But you get the idea. | 15:46.41 |
| henrys | is it just a london thing or all of europe is like that? | 15:47.34 |
| Robin_Watts | 2056.75 vs 785.75 | 15:47.38 |
| | henrys: I thought it was a worldwide thing, so I clearly have no idea how to answer that question. | 15:47.59 |
| kens | Miles said to me it was different for US folks, but when he aqctually tried it,it turned out not to be | 15:48.23 |
| | But that was some time ago | 15:48.29 |
| henrys | how much is a one way? | 15:48.42 |
| kens | I'm not sure immigration likes you arriving on a one-way ticket ;-) | 15:49.10 |
| Robin_Watts | henrys: it would work out >= the 2K price, I'm sure | 15:49.33 |
| henrys | from what I'm grokking a one way here and a round trip back where you didn't use the 2nd leg of the round trip would be cheapr. | 15:49.54 |
| Robin_Watts | We could fly on saturday, have the meeting sunday/monday and then fly home. | 15:49.54 |
| kens | Robin_Watts : we've done that before (or at least, I hacve) | 15:50.12 |
| Robin_Watts | kens: yeah, me too. | 15:50.35 |
| paulgardiner | How about if we all live in the US and visit the UK for 4 almost 3 month periods a year with 2 day gaps? | 15:50.45 |
| kens | Canslo do Thurs/Friday and fly back Sunday | 15:51.00 |
| Robin_Watts | That would mean that USicans could maybe fly in sunday morning? | 15:51.01 |
| | i.e. only lose one day of their weekend. | 15:51.15 |
| | henrys: 1 way flight = 1085.15 | 15:51.28 |
| marcosw | the easiest solution is to use nested tickets, but that would require miles to plan two staff meetings in advance. | 15:51.57 |
| Robin_Watts | marcosw: Cunning. | 15:52.19 |
| marcosw | I used to fly a lot on business for cheap companies :-) | 15:52.40 |
| henrys | marcosw:well if you can figure out a was to keep us working on the workday and off on the weekdays at a reasonable cost I'll talk miles into it. | 15:53.28 |
| Robin_Watts | Hmm. I could fly from Johanessburg to Chicago with a 4 week layover in london... | 15:53.28 |
| marcosw | Robin_Watts: we also used to buy round trip tickets and throw away the return, that way a trip is never more than 2x the cheapest fare. | 15:54.10 |
| henrys | of course he probably deducts it so the hit is not as large as it seems. | 15:54.21 |
| Robin_Watts | marcosw: Right. That only works if you're not flying *from* europe/ | 15:55.02 |
| | cos if you're flying from europe, you have the stupid taxes etc on the outbound leg. | 15:55.26 |
| henrys | okay 5 minutes until the ghostscript cancelled meeting. I'll remind everyone at the half hour that it is over ;-) | 15:55.29 |
| | marcosw:from the logs - I haven't seen a bug report in about a month - is it me? | 15:56.09 |
| marcosw | henrys: they are being produced but apparently the emails aren't making it. I've started looking into it. | 15:56.37 |
| | Robin_Watts: I don't see how the taxes a problem, if a round trip from LHR to JFK costs 500 pounds with a saturday stay and 1500 pounds without and a roundtrip from JFK to LHR costs $1000 with a saturday night stay isn't it cheaper to buy the two round trips? Or does the 500 pound fare not include taxes? | 15:58.21 |
| Robin_Watts | marcosw: Oh, I see. | 16:00.00 |
| marcosw | and in theory you should be able to get back at least the various departure/security/customs fees on the two return tickets you throw away, but I suspect that's a fools errand. | 16:02.10 |
| | or maybe it is possible (at least in the UK): http://www.skyscanner.net/news/apd-refunds-how-claim-back-airline-tax | 16:04.28 |
| | "Congress or the FAA should end the practice of not refunding aviation ticket taxes on unused commercial airline tickets," says the National Business Travelers Association. | 16:06.35 |
| henrys | I have to think teleconferencing has gotten good enough that we could have US'rs meet at one location and Euros at another and have an experience not much different than our meeting. | 16:13.25 |
| kens | is unconvinced | 16:13.43 |
| Robin_Watts | would have to check the rules on bacon. | 16:14.10 |
| | and milkshakes. | 16:14.26 |
| henrys | there must be milkshakes in London like the US no? | 16:15.31 |
| Robin_Watts | henrys: harder to find. | 16:15.42 |
| | but calories in a different timezone don't count. | 16:16.09 |
| henrys | the gamma radiation of flight converts the mass to energy. | 16:17.19 |
| Robin_Watts | sounds convincing to me. | 16:18.52 |
| henrys | They've set up nature cams at one of my favorite running spots, not sure if I really want to look at them anymore: http://www.thedenverchannel.com/news/local-news/mountain-lion-mugs-for-remote-wildlife-camera-near-mount-sanitas | 16:19.51 |
| | I can tell that's a bad kitty | 16:20.21 |
| Robin_Watts | http://www.telegraph.co.uk/earth/wildlife/8535055/Lion-steals-photographers-camera.html | 16:21.40 |
| kens | 1OK I'm off goodnight all | 16:28.33 |
| marcosw | henrys: I found the missing bug emails, google identified them as spam so they were being held for moderation. I'm trying to figure out how to avoid this in the future but am thwarted by the fact that if google tells people how to avoid having their emails identified as spam the spammers will win. | 16:50.07 |
| | that was easy, they have a whitelist by sender. | 16:51.39 |
| henrys | I'm moving the agenda to a live document I wonder if that couldn't be done with the bug report - update it every hour and write it to a web page and folks can look at it when they want. | 16:52.10 |
| marcosw | isn't that what bugs.ghostscript.com is? | 16:52.33 |
| henrys | yes if we enable all the stats so we had reporting | 16:53.13 |
| | if there a place on the bug page where say miles can go and see a summary of the bug universe? | 16:53.49 |
| marcosw | a bug dashboard? That shouldn't be too hard to whip up. | 16:54.23 |
| henrys | I think bugzilla gives you all kinds of reporting we just don't have it turned on right? | 16:55.10 |
| marcosw | lists bugs counts, current customer bugs, and the latest bugs. Bugs that are stale (i.e. no engineer comment in X days) are in red, etc. | 16:55.20 |
| | a combination of the monday and tuesday bug emails. | 16:55.35 |
| henrys | yes that would be better for me. | 16:56.10 |
| marcosw | I'll have to check into bugzilla, I didn't think it was quite that versatile. | 16:56.19 |
| henrys | probably not. | 16:56.29 |
| | is it slow to generate the report - can it be done in real time? | 16:56.59 |
| marcosw | this is interesting: | 16:57.30 |
| | http://bugs.ghostscript.com/report.cgi?x_axis_field=bug_status&y_axis_field=assigned_to&z_axis_field=&query_format=report-table&short_desc_type=allwordssubstr&short_desc=&resolution=---&longdesc_type=allwordssubstr&longdesc=&bug_file_loc_type=allwordssubstr&bug_file_loc=&keywords_type=allwords&keywords=&bug_id=&bug_id_type=anyexact&emailassigned_to1=1&emailtype1=substring&email1=&emailassigned_to2=1&emailreporter2=1&emailqa_contact2=1&emailc | 16:57.31 |
| | &emailtype2=substring&email2=&emaillongdesc3=1&emailtype3=substring&email3=&chfieldvalue=&chfieldfrom=&chfieldto=Now&j_top=AND&f1=noop&o1=noop&v1=&format=table&action=wrap | 16:57.31 |
| SpNg | what is the best way to read the colorspace of an eps file? | 16:57.34 |
| marcosw | oops: http://tinyurl.com/k93s7gd | 16:57.51 |
| | lots of people have bugs who probably shouldn't (i.e. Igor, jack, dan, ...) | 16:58.51 |
| henrys | I'll add that to the agenda - I can probably go through those. | 16:59.21 |
| | but we do want some kind of report that contains all the info we are getting now from both emails. | 16:59.55 |
| marcosw | add an agenda item to discuss a bug dashboard and/or updated weekly emails. | 17:00.10 |
| henrys | okay | 17:00.30 |
| marcosw | thc | 17:06.10 |
| | thx | 17:06.11 |
| henrys | gawd 822 total issues. | 17:15.20 |
| marcosw | henrys: ghostpcl won't build with "-DENABLE_VALGRIND -DPACIFY_VALGRIND". do you want to take a quick look or shall i open a bug? | 17:25.33 |
| henrys | I'll look now. thanks | 17:26.02 |
| tor8 | henrys: one way airline tickets are even more expensive than business class... | 17:26.18 |
| marcosw | plplatf.c:42: undefined reference to `VALGRIND_HG_DISABLE_CHECKING' | 17:26.42 |
| mvrhel_laptop | henrys: wow about the mountain lion | 17:35.20 |
| henrys | mvrhel_laptop:just have to make sure the group has a few slower than me. | 17:37.02 |
| mvrhel_laptop | exactly. | 17:37.11 |
| ray_laptop | Robin_Watts: I saw that you thought my clipping change fixed a crash ??? | 17:39.12 |
| Robin_Watts | ray_laptop: There is a bug that no longer happens after your fix. | 17:39.48 |
| henrys | marcosw:hmm can't reproduce? | 17:39.49 |
| Robin_Watts | and it does involve clipping. | 17:39.58 |
| henrys | make -k pcl-debug CFLAGS="-DENABLE_VALGRIND -DPACIFY_VALGRIND" | 17:40.13 |
| | marcosw:is that your line. | 17:40.26 |
| marcosw | henrys: probably depends on what libraries are installed. let me try on peeves | 17:40.27 |
| henrys | ? | 17:40.28 |
| Robin_Watts | but it could easily be a heisenbug. | 17:40.29 |
| marcosw | make pcl-debug XCFLAGS="-DENABLE_VALGRIND -DPACIFY_VALGRIND" | 17:40.39 |
| Robin_Watts | marcosw: on what machine ? | 17:41.11 |
| | peeves has an old valgrind, IIRC. | 17:41.21 |
| | ISTR I built my own one. | 17:41.27 |
| marcosw | Robin_Watts: on my x6 cluster node. | 17:41.41 |
| henrys | marcosw:trying with XCFLAGS but I doubt that will matter. | 17:41.48 |
| Robin_Watts | ~robin/sauce/valgrind-3.8.1 | 17:41.50 |
| marcosw | but it fails on other cluster nodes as well (i.e. inches) | 17:42.07 |
| | so in what library is VALGRIND_HG_DISABLE_CHECKING normally found? | 17:42.32 |
| Robin_Watts | It should be a macro defined in valgrind.h | 17:42.55 |
| | or valgrind/helgrind.h | 17:43.01 |
| mvrhel_laptop | Robin_Watts: you available for a sec? | 17:43.26 |
| henrys | indeed with XCFLAGS I do get a compile fail | 17:43.29 |
| Robin_Watts | valgrind/helgrind.h | 17:43.41 |
| henrys | but CFLAGS works | 17:43.43 |
| Robin_Watts | mvrhel_laptop: I am here. | 17:43.45 |
| marcosw | I don't think CFLAGS does anything. | 17:43.46 |
| mvrhel_laptop | I can wait until you are finished I mean | 17:43.57 |
| sebras | henrys: see PM. | 17:44.03 |
| mvrhel_laptop | Robin_Watts: don't want to interupt. ping me when you are available | 17:44.21 |
| Robin_Watts | mvrhel_laptop: Go for it | 17:44.29 |
| | marcosw: Anything that uses VALGRIND_HG_... in gs should include our own valgrind.h header. | 17:45.10 |
| | That defines VALGRIND_HG_DISABLE_CHECKING if the nodes don't have it. | 17:45.29 |
| mvrhel_laptop | ok. so if mupdf ends up doing a throw due to some issue, how do I get things set up so my viewer does not crash. Right now I am ending up in base_error.c throw exit(EXIT_FAILURE) | 17:45.37 |
| Robin_Watts | So around every call into the lib, you should have: fz_try(ctx) {... do the call here ... } fz_catch(ctx) { ... clean up ... } | 17:46.17 |
| mvrhel_laptop | I figured out the open xps issue. I was passing along the .oxps extension | 17:46.21 |
| | oh problem here is that this is in the streaming stuff | 17:46.37 |
| Robin_Watts | mvrhel_laptop: Well, we should probably cope with .oxps. | 17:46.38 |
| mvrhel_laptop | file streaming set up | 17:46.46 |
| Robin_Watts | mvrhel_laptop: What streaming stuff? | 17:46.55 |
| marcosw | Robin_Watts: there is a #include <valgrind/helgrind.h> in pl/plplatf.c | 17:46.57 |
| mvrhel_laptop | when we are attempting to open the file, | 17:47.05 |
| Robin_Watts | marcosw: Right. That should be #include "valgrind.h" and the problem should go away. | 17:47.19 |
| mvrhel_laptop | if the extension is unknown, we try to open it as a pdf | 17:47.21 |
| | oh I see what I need | 17:47.29 |
| Robin_Watts | mvrhel_laptop: Right. We should cope with .oxps meaning the same as .xps | 17:47.41 |
| mvrhel_laptop | I need to put the catch around my call to fz_open_document_with_stream | 17:47.58 |
| Robin_Watts | The try, yes. | 17:48.06 |
| mvrhel_laptop | anyway | 17:48.07 |
| | yes | 17:48.09 |
| | plus it would be nice to add in the oxps option | 17:48.26 |
| | into fz_open_document_with_stream | 17:48.33 |
| | I will go ahead and do that | 17:48.40 |
| Robin_Watts | and you may want to use the fz_try { ... } fz_always { ... } fz_catch { ... } form. | 17:48.40 |
| | to avoid leaks. | 17:48.43 |
| mvrhel_laptop | Robin_Watts: will do | 17:48.47 |
| Robin_Watts | fab. | 17:48.53 |
| mvrhel_laptop | thanks | 17:49.00 |
| | for getting me back on track | 17:49.05 |
| Robin_Watts | no worries. | 17:49.06 |
| | marcosw: Are you trying the #onclude "valgrind.h" fix now? If not, I will. | 17:49.27 |
| marcosw | Robin_Watts: yes, compiling as we speak (type) | 17:49.39 |
| | Robin_Watts: nope, didn't work. | 17:50.28 |
| | never mind, there is a typo in the file. | 17:50.48 |
| Robin_Watts | in which file ? | 17:51.18 |
| marcosw | Robin_Watts: all sorted thx | 17:51.39 |
| Robin_Watts | ok :) | 17:51.47 |
| mvrhel_laptop | ugh. a P1 softmask bounding box bug | 17:55.43 |
| | must be my lucky day | 17:55.55 |
| | oh and it was from a commit from igor... | 17:56.15 |
| tor8 | Robin_Watts: oh... open_document and open_document_with_stream have different duplicated mime-type/file extension detection code :( | 17:58.03 |
| Robin_Watts | tor8: You say that like it's a bad thing :) | 17:58.23 |
| | We should possibly move to a more data driven way of working; a table of suffix -> type, and type -> functions. | 17:59.01 |
| tor8 | Robin_Watts: yeah. | 17:59.29 |
| marcosw | mvrhel_laptop: one of Igor's infamous "part 4 of 6" commits. | 17:59.43 |
| Robin_Watts | tor8: XPS leak fix on robin master | 18:18.55 |
| mvrhel_laptop | Robin_Watts: if you can look over the 3 commits in my repos, I would appreciate it | 18:58.12 |
| | now let me see if this thing still works on the surface with the ARM build.... | 18:59.11 |
| | have not done that in a while | 18:59.23 |
| | or tor8 above, if Robin_Watts is out | 19:00.11 |
| Robin_Watts | looking now. | 19:00.58 |
| mvrhel_laptop | ok thanks | 19:01.35 |
| Robin_Watts | mvrhel_laptop: In the first one, you've added "<ResourceCompile Include="mupdf_cpp.rc"> | 19:03.15 |
| | but not included mupdf_cpp.rc itself? | 19:03.26 |
| mvrhel_laptop | yes. that was a mistake | 19:03.27 |
| | and it was fixed later | 19:03.35 |
| Robin_Watts | likewise resource.h | 19:03.37 |
| mvrhel_laptop | yes. same | 19:03.41 |
| Robin_Watts | ok, so those 2 commits exactly cancel one another out? | 19:04.12 |
| mvrhel_laptop | yes | 19:04.16 |
| | no | 19:04.18 |
| | they cancel out that part | 19:04.25 |
| | but there is other stuff in there | 19:04.29 |
| | sorry | 19:04.31 |
| Robin_Watts | no, there isn't. | 19:04.37 |
| mvrhel_laptop | hmm hold on | 19:04.45 |
| | ok so 444bc introduced them, plus other stuff | 19:05.17 |
| Robin_Watts | http://git.ghostscript.com/?p=user/mvrhel/mupdf.git;a=commitdiff;h=444bcbc1b423b870d492c8be2bc81640e3b72600 | 19:05.32 |
| mvrhel_laptop | and d20d removed the resources | 19:05.37 |
| | Robin_Watts: I guess I don't understand your question | 19:06.05 |
| Robin_Watts | Look at that link. | 19:06.14 |
| mvrhel_laptop | the two commits do not cancel one another. | 19:06.19 |
| Robin_Watts | It shows the changes in commit 444bc | 19:06.27 |
| mvrhel_laptop | the second commit removes some stuff that was introduced by accident in the first commit | 19:06.40 |
| ray_laptop_ | henrys: There is a commit of yours from May 1999 that has a check that looks funky ;-) | 19:06.51 |
| Robin_Watts | and those are restricted to the changes in mupdf_cpp.vcxproj.* | 19:06.55 |
| mvrhel_laptop | yes, as far as those files are concerned the commits cancel | 19:07.15 |
| Robin_Watts | mvrhel_laptop: Are you saying there is more in that commit than the link above shows? | 19:07.18 |
| mvrhel_laptop | ok let me start at the beginning | 19:07.30 |
| henrys | ray_laptop_: that was before I learned how to code ;-) | 19:07.44 |
| mvrhel_laptop | oh wierd | 19:07.49 |
| ray_laptop_ | henrys: actually I think it's from Peter, but I can't be sure. All you did was a massive commit to our repository (I think we were using CVS back then) | 19:08.07 |
| mvrhel_laptop | wtf | 19:08.10 |
| | 5ab3dae was the commit I was thinking of where the resources were added | 19:08.48 |
| ray_laptop_ | The History 5 does mention: - Images that lay entirely outside the page could cause a memory access error. (gxclimag.c) | 19:08.48 |
| mvrhel_laptop | where did 444bcbc come from | 19:09.02 |
| Robin_Watts | Ah, I understand. | 19:09.16 |
| mvrhel_laptop | d20d94a6d did cancel out 444bcbc | 19:09.24 |
| Robin_Watts | OK. so git rebase -i HEAD~10 | 19:09.26 |
| | then remove those 2 lines and the cancelling out commits will vanish, leaving just 1 commit. | 19:09.55 |
| mvrhel_laptop | oh and 9820d3cf removed them | 19:09.58 |
| | both d20d94a6 and 444bcbc1 should not be there | 19:10.16 |
| Robin_Watts | I'll do that for you if you want, but I've got to go walk the dog now. | 19:10.19 |
| ray_laptop_ | But the code that I don't think makes sense w.r.t. the comment is in write_image_end_all: | 19:10.24 |
| | /* | 19:10.26 |
| | * We need to check specially for images lying entirely outside the | 19:10.27 |
| | * page, since FOR_RECTS doesn't do this. | 19:10.29 |
| | */ | 19:10.30 |
| | if (rheight <= 0) | 19:10.32 |
| mvrhel_laptop | I have no idea how I managed that one Robin_Watts | 19:10.33 |
| ray_laptop_ | return 0; | 19:10.34 |
| Robin_Watts | mvrhel_laptop: I suspect it's an artifact of rebasing. | 19:10.44 |
| | it's easy to fix. | 19:10.50 |
| mvrhel_laptop | so the only commit that is of interest to review is the top one in my repos | 19:10.59 |
| | sorry for the mixup | 19:11.02 |
| Robin_Watts | I'll do that, and look over the last commit when I get back. | 19:11.02 |
| | no worries :) | 19:11.08 |
| | mvrhel_laptop: Do you need tea this time? | 19:11.23 |
| mvrhel_laptop | oh yes. that would be great | 19:11.30 |
| ray_laptop_ | This doesn't make sense to me since an image that is entirely outside the page has ymax < 0 or ymin > dev->height | 19:11.34 |
| Robin_Watts | I will see what I can manage. | 19:11.40 |
| mvrhel_laptop | Robin_Watts: just regular English Breakfast if you can | 19:11.49 |
| Robin_Watts | mvrhel_laptop: Sure. The twinings stuff OK? | 19:12.01 |
| mvrhel_laptop | that is great. or the store brand was great too | 19:12.12 |
| Robin_Watts | ok. | 19:12.19 |
| mvrhel_laptop | thanks a bunch | 19:12.27 |
| ray_laptop_ | henrys: or Robin_Watts or anybody. Can I have a spare set of eyes on the check for image off the page in write_image_end_all (line 2217 in gxclimag.c) ??? | 19:15.22 |
| | specifically the check on line 2230 | 19:16.03 |
| henrys | ray_laptop_:the comment and the code seem unrelated. | 19:21.41 |
| ray_laptop_ | henrys: thanks. That's what I thought | 19:26.39 |
| henrys | gosh the FOR_RECT macro was removed but many comments cite it. | 19:26.58 |
| mvrhel_laptop | Robin_Watts: another question for you when you return.... | 19:29.23 |
| | problems with openjpeg stuff when compiling for the ARM with Visual studio | 19:32.00 |
| | not sure how we handle third party lib issues like that | 19:32.27 |
| Robin_Watts | mvrhel_laptop: For openjpeg, we fix it ourselves. | 19:34.41 |
| | and we offer the fix upstream. | 19:34.49 |
| | if (code == S_ISOK) | 19:38.01 |
| | { | 19:38.03 |
| | InitialRender(); | 19:38.05 |
| | return code; | 19:38.07 |
| | } | 19:38.09 |
| | else | 19:38.10 |
| | return code; | 19:38.12 |
| | Why not just: | 19:38.27 |
| | if (code == S_ISOK) InitialRender(); | 19:38.28 |
| | return code; | 19:38.30 |
| | Also: fz_open_document_with_stream(ctx, ext, stream) takes a reference to the stream that's passed in. | 19:41.30 |
| | so I'm not sure you need to keep mu_stream about at all. | 19:43.40 |
| | i.e. rather than doing this->mu_stream = blah; then calling fz_open_document_with_stream, why not get rid of this->mu_stream entirely? | 19:44.25 |
| | and do: fz_stream *mu_stream = fz_new_stream; .... fz_try { mu_doc = fz_open_document_with_stream(... mu_stream); } fz_always { fz_drop_stream(stream) } fz_catch { return E_FAILURE; } | 19:46.05 |
| | mvrhel_laptop: ^ | 19:51.14 |
| | ray_laptop_: I won't get time to look tonight, but I will do so during the day tomorrow if you send me an email to remind me :) | 19:51.42 |
| mvrhel_laptop | Robin_Watts: ok let me look this over | 19:52.48 |
| Robin_Watts | mvrhel_laptop: http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=a5402710b7f1dc658439be0d27936a09733805f9 | 19:58.36 |
| mvrhel_laptop | Robin_Watts: oh ok. You already did the changes to the stream stuff | 19:59.56 |
| Robin_Watts | mvrhel_laptop: I thought it was easier to show you what I meant than to try to describe it in english :) | 20:00.16 |
| | I suspect it's probably worth you trying a memento build to check for leaks. | 20:00.41 |
| mvrhel_laptop | I understood and was making the changes here, but I am fine with it | 20:00.44 |
| Robin_Watts | I found loads in the android stuff that I wasn't expecting. | 20:00.59 |
| | It's ever so easy to miss a 'drop' and leak loads, as I've proved on many occasions. | 20:01.32 |
| mvrhel_laptop | Robin_Watts: ok. I will look over how memento is built in win32 project and see if I can replicate in winRT | 20:02.13 |
| | have to fix a few things with respect to the ARM build | 20:02.37 |
| | also | 20:02.40 |
| Robin_Watts | in theory you should just need to predefine MEMENTO and rebuild. | 20:03.13 |
| mvrhel_laptop | need to step out for a bit now though. I may bug you a bit tomorrow if I run into memento issues | 20:03.15 |
| Robin_Watts | having said that I've never tried it with C++ etc. :( | 20:03.24 |
| | sure. ttyl. | 20:03.28 |
| mvrhel_laptop | have you committed to golden? | 20:03.34 |
| | I guess I should do a hard reset to that to avoid any merge issues | 20:03.48 |
| | when I have a commit here and there is a slightly different version in golden it would seems issues occur | 20:04.17 |
| | bbiaw | 20:06.39 |
| | Robin_Watts: are you going to go ahead and commit your altered version of my commit? | 23:41.22 |
| | or do you want me to fix up mine and amend the last commit | 23:41.38 |
| | and then you can check it | 23:41.43 |
| Robin_Watts | mvrhel_laptop: I was planning to let you do it, cos you can test it :) | 23:41.47 |
| mvrhel_laptop | good idea | 23:41.52 |
| | thanks | 23:41.55 |
| Robin_Watts | no problem. | 23:42.02 |
| | Forward 1 day (to 2013/06/05)>>> | |
IRC Logs