| | <<<Back 1 day (to 2012/08/15) | 2012/08/16 |
| sebras | tor8: also we might want to update those cmaps, UniCNS-UTF16-H, UniGB-UTF16-H, UniJIS-UTF16-H and UniKS-UTF16-H have all been updated when I compare adobe's latest versions with mupdf's. | 00:34.27 |
| robin_watts_mac | Morning all | 07:46.49 |
| kens | morning robin_watts_mac | 07:47.35 |
| sebras | morning paulgardiner | 10:09.50 |
| paulgardiner | hi. Saw your message in the logs. Will check out that file. | 10:10.16 |
| | I thought I thought through the one-page case. Obviously not carefully enough. :-) I had a max(val, 1) in there at first and then wrongly worked out I didn't need it, | 10:13.11 |
| | Ah. I carefully thought through the smax == 1 case. Doh! | 10:19.09 |
| sebras | yes, exactly. there will be no one-page pdfs. ;) | 10:21.08 |
| paulgardiner | I guess we shouldn't really display a seek bar for one-page document, but for now I'll just make it behave consistently with other cases. | 10:28.09 |
| | sebras: there, pushed a fix and I've uploaded another apk. Thanks for pointing that out. | 10:32.07 |
| sebras | paulgardiner: sure, no problem. I just stumbled over it. :) | 10:32.24 |
| | good thing too, otherwise we'd likely have a mupdf 1.1.1 ;) | 10:32.50 |
| paulgardiner | yeah, I really should have tested the one-page case, especially as I'd realised there was at least a potential problem. | 10:34.33 |
| sebras | paulgardiner: agh! I'm trying to provoke an array index out of range bug in the Type3 font rendering. | 10:35.31 |
| | the sumatra guys have pointed this out, but can not provide the bogus file. | 10:35.52 |
| robin_watts_mac | paulgardiner: A patch on my casper repo on the forms branch. | 10:36.26 |
| paulgardiner | sebras: that'll be painful to find. | 10:36.36 |
| sebras | basically they claim that they have a file where a string contains a codepoint that gets mapped to a gid which is outside the valid range for type3 fonts. | 10:36.42 |
| paulgardiner | robin_watts_mac: oh hi Robin. I'll have a look | 10:36.49 |
| sebras | so now I'm trying to recreate such a file... | 10:36.54 |
| robin_watts_mac | Mountain Lion has a different compiler or default warning options or something. | 10:37.02 |
| | sebras, paulgardiner: I've introduced an UNUSED(x) macro to indicate that a variable is unused. | 10:38.20 |
| | Just doing e=e; now gives a nasty warning message. | 10:38.33 |
| sebras | robin_watts_mac: ok. tired of compiler warnings? | 10:38.38 |
| | why don't we either just remove those variables or use them? | 10:39.09 |
| paulgardiner | If I introduced e=e somewhere that would have been to put a breakpoint on and wasn't intended for commit. | 10:39.17 |
| robin_watts_mac | paulgardiner: The use of e=e; is in fz_var(); | 10:39.34 |
| paulgardiner | oh i see | 10:39.43 |
| robin_watts_mac | but other places we calculate ctx, and then don't use it currently. | 10:39.48 |
| | but I'm sure we will when we flesh the code out. | 10:40.00 |
| paulgardiner | so how come only on forms? | 10:40.01 |
| robin_watts_mac | 1) The majority of the changes are actually with const char *'s. | 10:40.24 |
| | s/1)// | 10:40.35 |
| | C++ dislikes casting string literals to non const char *'s | 10:41.03 |
| | so I had to push some consts through the code. | 10:41.22 |
| paulgardiner | oh right. | 10:42.17 |
| | So does it compile cleanly now? | 10:42.31 |
| robin_watts_mac | There are more UNUSED things appearing on master on my repo now. | 10:42.36 |
| sebras | I have seen a few unused variables in the xps-code as well. I know tor8 parsed out some attributes but never got around to actually use them (might not make sense to implement certains parts eiter). | 10:42.40 |
| robin_watts_mac | The only warnings left on forms are ones that are are still on master, and a few in thirdparty. | 10:43.01 |
| paulgardiner | Magic. | 10:44.41 |
| | The unused vars would probably be better removed, but I can do that. | 10:44.59 |
| robin_watts_mac | oh, ok. | 10:46.22 |
| paulgardiner | robin_watts_mac: changes look fine, trusting you on the use of %lld in a fprintf because I am not familiar with it. | 10:46.41 |
| robin_watts_mac | That's what the compiler told me. | 10:46.54 |
| kens | Oh, I see Harry Harrison has died :-( | 10:48.13 |
| robin_watts_mac | kens: Indeed. | 10:49.06 |
| | I remember "The Men from PIG and ROBOT" fondly. | 10:49.23 |
| kens | I spent some time looking for a copy, I haev one now :-) | 10:49.44 |
| sebras | paulgardiner, robin_watts_mac: ehm, mujstest_count is an int, then you add 46702919800 which is bigger than an int on a 32-bit system and the print it as a long long... | 10:51.05 |
| | maybe it ought to be declared as a uint64_t and the literal suffixed with ULL? | 10:51.39 |
| robin_watts_mac | sebras: No. It should stay as an int, and we should explicitly cast it. | 10:52.35 |
| | 46702919800 + (int64_t)mujstest_count | 10:53.08 |
| | although... we may not have int64_t defined. | 10:53.25 |
| sebras | robin_watts_mac: there's a patch on sebras/master that will add 64-bit types to fitz.h | 10:53.42 |
| robin_watts_mac | and the upcast is implicit, I think, so it can probably stay as it is. | 10:53.46 |
| | sebras: This is only for generating .mjs files, and I'm the only person who ever does that :) | 10:54.12 |
| sebras | robin_watts_mac: still! ;) | 10:54.28 |
| | I'm not sure about the upcast. why would one otherwise declare a literal ULL? | 10:54.59 |
| robin_watts_mac | For values that are dangerously close to the 32bit limits? | 10:56.22 |
| paulgardiner | robin_watts_mac: commit 28541e46 attempts to cause mjs generation to get stuck in an infinite loop, but hoepfully just manages to fail to do so. I hope I've reasoned better about that than about the divide by 0 in the android app. | 10:57.21 |
| robin_watts_mac | I don't have that SHA ? | 10:58.38 |
| | brb | 10:58.54 |
| paulgardiner | on paul/forms | 11:00.14 |
| robin_watts_mac | Looks good to me. | 11:05.28 |
| paulgardiner | ta | 11:06.35 |
| robin_watts_mac | actually, I was looking at the wrong thing. | 11:08.44 |
| | let me look at the right thing. | 11:08.50 |
| tor8 | sebras: ULL is for values that are outside the 32-bit limits | 11:09.19 |
| robin_watts_mac | Ah, a tor8. I didn't see you lurking there. | 11:10.26 |
| tor8 | robin_watts_mac: how's your vacation? | 11:10.55 |
| sebras | tor8: yes, and 46702919800 > 2**32 == True (hello python) | 11:10.58 |
| robin_watts_mac | tor8: Hot. | 11:11.11 |
| | Have just moved to Gozo. Much nicer hotel. Waiting for room to be ready. | 11:11.33 |
| sebras | tor8: however, given what I read in C99 robin_watts_mac is perfectly correct in that the implicit type for constant literals is chosen by what type can fit the numerical value, hence it will automagically be long long which is normally 64-bits. | 11:12.37 |
| robin_watts_mac | paulgardiner: So... we ask for some text, and we try to set it. If it fails to set, we go back to the text input routine with a 'retry' flag. | 11:14.11 |
| | and currently the X and test versions just return NULL then. | 11:14.33 |
| paulgardiner | yep | 11:14.34 |
| robin_watts_mac | but the windows one handles it properly? | 11:14.43 |
| paulgardiner | yep and that stops the loop | 11:14.46 |
| | yep | 11:14.49 |
| robin_watts_mac | seems plausible to me. | 11:15.21 |
| paulgardiner | good good | 11:15.30 |
| robin_watts_mac | Pushed (along with a version of my patch without the UNUSED) | 11:21.13 |
| paulgardiner | Oh my commits too. Thanks. | 11:22.03 |
| robin_watts_mac | Crumbs. Is the cluster testing far too many revisions? Or did I push more than I thought ? | 12:00.36 |
| kens | tor8 pushed a whole load | 12:03.12 |
| | I see 20 commit emails | 12:03.42 |
| robin_watts_mac | kens: Right. If it was tor8 that pushed them, that's fine. | 12:04.10 |
| kens | Looks to be approximately the expected number of tests | 12:04.52 |
| robin_watts_mac | jbig2_decode_generic_template3 doesn't appear to ever be called. | 12:12.14 |
| | but jbig2_decode_generic_template3_unopt is called twice. That looks like a typo to me. | 12:12.48 |
| kens | shouldn't you be in a pool or somethign robin_watts_mac ? | 12:16.22 |
| robin_watts_mac | Still waiting for room to be ready. | 12:16.52 |
| kens | :-( | 12:17.00 |
| | Oh great some free user has 'attached a Visual Stusdio project' which is 58Mb, zipped..... | 12:58.16 |
| robin_watts_mac | kens: Don't knock it if it's the simplest way of giving a coherent bug report. | 13:01.06 |
| kens | It inludes the objects and pdb files, but *not* the executable.... | 13:04.10 |
| | Oh, and it also includes the Ghostscript DLL.... | 13:05.22 |
| | Both 32 and 64 bit versions | 13:05.39 |
| | What the heck is a '.sdf' file ? | 13:06.20 |
| | 65Mb of the content is in that file and Windows doesn't know what to do with it.... | 13:06.44 |
| | Hmm, google seems to think its a SQL database file. | 13:07.20 |
| robin_watts_mac | .sdf is the vs2010 equivalent of .ncd | 13:09.51 |
| | apparently. OK. off to pool. bbl. | 13:10.05 |
| kens | Great so *many* megabytes of useless binary cryft | 13:10.13 |
| | bye robin_watts_mac | 13:10.32 |
| | Ah, apparently I don't have VS 2010 express installed, so I can't use that solution he's sent | 13:16.39 |
| sebras | tor8: I believe I fixed the LZW issue. it's over at sebras/master | 13:28.04 |
| tor8 | grrr. latest chrome has these annoying "safe browsing" warnings that throw up big scary alerts if you download a zip with an exe in it. that won't be good for our download pages unless we figure out what arcane crap needs to be done to placate the google overlords... | 13:52.53 |
| | same thing with IE9 I hear too | 13:53.09 |
| sebras | tor8: btw I ran sebras/master through sane/insane without any diffs. | 13:53.09 |
| mvrhel_laptop | bbiaw | 14:18.05 |
| apineda | I'm using a pdf into tiff32nc with -dUseFastColor=true and CM is still being applied! GS 9.06 | 14:19.54 |
| mvrhel_laptop | apineda: open a bug report on this | 14:22.54 |
| | and I will look into it | 14:23.28 |
| apineda | Alright | 14:31.33 |
| tor8 | chrisl: mupdf 1.1 is uploaded | 14:48.52 |
| sebras | tor8: cool! :) | 14:57.49 |
| | tor8: I'm harvesting sumatrapf's bugzilla for issues we can fix. seems like there's quite a few fixes that never made it to us. | 14:58.19 |
| robin_watts_mac | sebras: Yes, Zeniko (sorry, zeniko) seems to have an aversion to actually opening individual bug reports and attaching patches. | 15:02.43 |
| robin_watts_mac | carries laptop to pool. bbs. | 15:02.59 |
| sebras | robin_watts_mac: yeah, I know. that's why I used to patrol their bugzilla. I haven't done that for the last 900 bugs though. :-/ | 15:03.43 |
| chrisl | tor8: thanks | 15:14.01 |
| robin_watts_mac | tor8: Any objection to me creating a 'pdf_obj *me' entry in a pdf_page ? | 15:19.46 |
| chrisl | tor8: ghostcript.com downloads page updated, and mupdf "customer release archive" done, too. | 15:28.25 |
| apineda | http://bugs.ghostscript.com/show_bug.cgi?id=693282 I created this report for mvrhel however, if you see my last note it's simply a syntax problem on my side. However, it is possible you guys may want to make changes in the future so I don't want to resolve it(can I even?). The bottom line is that I don't want to waste mrvhel's time reading through the large first comment, only to find out in... | 15:59.42 |
| | ...the last comment what the problem was. Thanks! | 15:59.44 |
| kens | Hmm, that comment #2 on bug#693282, that really shouldn't happen..... | 16:05.27 |
| | I wonder why it needs a space in front | 16:05.39 |
| chrisl | kens: I wondering that myself - -dUseFastColor should be sufficient, anyway | 16:12.25 |
| kens | chrisl that's probably why it works with a space... | 16:12.43 |
| | its equivalen to -dUseFastColor | 16:12.52 |
| chrisl | It certainly shouldn't be a blocker...... | 16:12.59 |
| kens | No, that seems reasonable | 16:13.11 |
| | unreasonable* | 16:13.17 |
| chrisl | I get the same output with "-dUseFastColor=true" and "-dUseFastColor", and "-dUseFastColor= true" throws an error..... | 16:18.38 |
| kens | OK off now, goodnight all | 16:18.39 |
| mvrhel_laptop2 | chrisl: you there? | 18:13.13 |
| chrisl | mvrhel_laptop2: sort of...... | 18:17.44 |
| mvrhel_laptop2 | hey. just curious about the -dUseFastColor bug | 18:18.04 |
| | so things worked fine for you? I don't understand how it could work with -dUseFastColor= true | 18:18.30 |
| | The same code for all the boolean options is used for this as everything else | 18:18.56 |
| chrisl | It all worked as I expected, yes. With the space, it gave me an error | 18:18.56 |
| mvrhel_laptop2 | ok | 18:19.00 |
| chrisl | I'm wondering if he's calling through the API rather than from the command line | 18:19.33 |
| mvrhel_laptop2 | thanks. enjoy your evening chrisl | 18:19.33 |
| | oh that could be | 18:19.39 |
| chrisl | I guess it might be worth clarifying that - oh, and I tested on Linux, not on Windows, but as you say, it's the same code for all those options | 18:20.37 |
| sebras | tor8: zeniko is not happy about my png patch. signedness issues. :-/ | 18:30.24 |
| Louis_ | hello? | 20:06.48 |
| | I just downloaded the fresh gxps-9.06-win32.exe hoping it had solved some things preventing me from using the earlier versions. | 20:08.43 |
| | Thought I'd ask here before filing a bug... | 20:09.29 |
| | If any humans are listening? | 20:10.04 |
| | too bad there's no user forum or so, eather chat or bugzilla, no inbetween.... | 20:11.00 |
| | bye. | 20:11.14 |
| | Forward 1 day (to 2012/08/17)>>> | |
IRC Logs