| <<<Back 1 day (to 2012/12/19) | 2012/12/20 |
Robin_Watts | sebras: Any reason why 1.5.1 and not 2.0 ? | 00:13.13 |
JakeSays | so i'm augmenting the txtwrite driver to output rects and paths. is there some sample code i can look at to get an idea of how to get at the path data? | 00:19.58 |
Robin_Watts | JakeSays: gs or mupdf? | 00:21.18 |
| gs I guess. | 00:21.25 |
JakeSays | Robin_Watts: gs, but i found the trace driver | 00:21.33 |
Robin_Watts | fab. | 00:21.39 |
JakeSays | :) | 00:21.44 |
| heh. trace_path | 00:22.02 |
| perfect | 00:22.04 |
zygew | sebras,i used to read a little code or mupdf,it rotated CContext or CCanvas. so i guess it is possible in theory to rotate in text viewer.. | 00:30.58 |
| not very professional on windraw | 00:31.59 |
| blind guess | 00:32.37 |
| things like totalcommand uLister | 00:36.03 |
sebras | Robin_Watts: because 1.5.1 was a simple upgrade. :) | 00:42.54 |
| Robin_Watts: also I didn't know whether you opted out of 2.0 for interface reasons. | 00:43.12 |
| zygew: ok, I'm not on windows so I know nothing about what you write. but as long as you do then it's fine! :) | 00:44.47 |
| Robin_Watts: found the culprit that caused the regression. | 00:44.57 |
| in thirdparty/openjpeg/libopenjpeg/j2k.c::j2k_read_sod() there's a new section commented "chop pading bytes" this section causes my regression. | 00:45.30 |
Robin_Watts | We haven't used 2.0 in the past cos it was unstable and rapidly changing. | 00:46.24 |
| but they've actually released it now, which means, I suspect, we should at least look at it. | 00:46.40 |
sebras | this was introduced in r1702 in openjpeg back on tuesday 02592012 with the comment: "Applying old patch from r1642, with further modification from winfried. Also enable failing test. Everything seems to be working well now." | 00:46.46 |
| I have extracted the jp2-file from the pdf and will send them a bug on 1.5.1. | 00:47.21 |
| Robin_Watts: btw... is there any reason that the font rendering is a bit bad right now? | 00:47.49 |
Robin_Watts | sebras: It is ? | 00:48.08 |
sebras | it is using mupdf HEAD here... | 00:48.23 |
Robin_Watts | In what way bad? | 00:48.31 |
| I just updated type 3 rendering, but only 2 files changed appearance. | 00:49.03 |
| There hasn't been a change that has affected font rendering for ages, according to the regression tests | 00:49.35 |
sebras | maybe I have messed up. | 00:50.12 |
Robin_Watts | I'm off to bed now. I may be being woken by my builder early tomorrow... night! | 00:52.17 |
sebras | Robin_Watts: sleep weel! | 00:52.30 |
| well. | 00:52.32 |
| Robin_Watts: d'oh. | 00:56.27 |
| I think the bug in openjpeg is rather obivous. tries to remove any zeroes from the end of the file (some creators pad for whatever reason). those zeroes are interpreted as invalid marker by openjpeg which causes problems. | 01:11.56 |
| however, this truncation of zeroes is not only done at the end of the file but also whenever parsing a tile, which means that the any tile whose data has at least two zero bytes at the end will have its data truncated. | 01:13.10 |
| Robin_Watts: reported -- http://code.google.com/p/openjpeg/issues/detail?id=205 | 01:56.25 |
Robin_Watts | kens. tor8: Either of you have experience of ftview ? | 10:10.02 |
kens | only slightly | 10:10.10 |
| wwOTOH I can debug FreeType | 10:11.24 |
Robin_Watts | Well, I have a fix. | 10:11.38 |
| The freetype guys would like it reproduced with ftview though. | 10:12.08 |
kens | Really ? Werner never asked for that before | 10:12.21 |
Robin_Watts | Are you on the freetype mailing list? | 10:12.42 |
kens | Yes | 10:12.45 |
| ft-devel | 10:12.50 |
Robin_Watts | No, this is vanilla 'freetype'. | 10:13.00 |
| I very rarely see postings on ft-devel. | 10:13.16 |
kens | I get lots | 10:13.25 |
| I normally send to ft-devel or direct to Werner | 10:13.43 |
Robin_Watts | kens: What's the easiest way to build ftview for windows ? | 10:19.08 |
kens | There's a project isn't there ? | 10:19.18 |
Robin_Watts | I've just tried the VS2005 one, and it builds the freetype lib only. | 10:19.35 |
kens | 1 sec | 10:19.53 |
Robin_Watts | I'll try building under linux. | 10:21.07 |
kens | Hmm, the last time I built FT was on my old PC and I didn't carry it across | 10:21.49 |
| I'll boot up old PC this will take a minute | 10:22.11 |
Robin_Watts | kens: Don't panic for a bit. | 10:23.08 |
| If I can make it run under linux that'll do. | 10:23.20 |
kens | I'll copy it over to new PC anyway :-) | 10:23.43 |
Robin_Watts | but I can't make it work under unix - at least not trivially. The configure step dies missing some files :( | 10:25.19 |
kens | OK give me a minute | 10:25.29 |
| Umm, and it looks like I trashed that folder when chrisl took over FT. Which I guess is why I didn't copy it over. | 10:27.18 |
| OK so let me go fetch a current copy | 10:28.15 |
| I can't even find ftview in the source archive | 10:34.15 |
| I bet the demos aren't in that \archive | 10:35.01 |
Robin_Watts | http://www.artifex.com still seems broken for me. | 10:40.04 |
kens | Its fine for me, I'm using it to write a support reply | 10:40.17 |
chrisl | Robin_Watts: I don't think ftview can work under Windows, because it pokes the X structures "directly".... | 10:41.47 |
Robin_Watts | chrisl: Ah. kens, have you run it on linux before? | 10:43.30 |
kens | Robin_Watts : yes, though some bits I have run on Windows | 10:43.50 |
| Not graphical bits | 10:44.00 |
chrisl | So, IIRC, the "example" apps don't have a configure of their own, they pull in the results of the Freetype lib's configure | 10:45.03 |
kens | And the makefile doesn'tg work with nmake. I do remember that I hand-built Visual Studio projects for the few apps I wanted to run under Windows. | 10:46.04 |
chrisl | Robin_Watts: on Linux, ftview can probably be installed via the package manager, that *might* be enough to demonstrate it crashing | 10:47.41 |
Robin_Watts | chrisl: Thanks. | 10:48.00 |
chrisl | On Ubuntu, it's in a package called "freetype2-demos" | 10:48.27 |
| Robin_Watts: I take it that the font is actually broken, though? | 10:49.33 |
Robin_Watts | Not necessarily. | 10:49.43 |
| It's possible that it just contains a large number of points in a glyph. | 10:50.01 |
| The underlying problem is that they try to cram an int into a signed short. | 10:50.18 |
| and so num_points ends up negative. | 10:50.29 |
chrisl | I've never seen a glyph with enough points to overflow a signed short! | 10:50.36 |
kens | No, that sounds broken | 10:51.15 |
Robin_Watts | It may be a broken font, but that's why it's dying. | 10:51.35 |
chrisl | FWIW, I have a "local" build of ftview for ft 2.4.4 here, if you want to send the font to me to try out | 10:52.14 |
Robin_Watts | thanks. I'll try to extract the font. | 10:55.50 |
chrisl | Oh, I can extract the font from a PDF if required - I do it often enough! | 10:56.22 |
kens | chrisl aren't you supposed to be on holiday ? :-) | 11:20.37 |
chrisl | kens: I am, but squash training cancelled, horrid weather, can't/won't go out on my bike, and expecting my parents in an hour or so (taking cat to the vet)...... | 11:22.19 |
| The FAPI indeterminism(s) were playing on my mind, so I thought I would have a play with that | 11:23.07 |
kens | Mad... Take a break | 11:25.24 |
chrisl | Yeh, I know, but I was bored...... | 11:26.20 |
Robin_Watts | tor8: ping | 13:01.47 |
tor8 | Robin_Watts: hey. | 13:25.02 |
Robin_Watts | Got a bug here to do with overflows when translating a bbox. | 13:27.38 |
| If we translate a bbox by (say) INT_MIN, INT_MIN, infinite bboxes become finite ones etc. | 13:28.24 |
| I propose to add an fz_translate_bbox function that checks for overflows. | 13:28.41 |
| This is particularly painful on MSVC as the result of: x = floorf(f) is INT_MIN if f won't fit in a int. | 13:30.03 |
tor8 | Robin_Watts: hang on. how often do we call translate_bbox? | 13:31.27 |
| (and should those places be using rects instead) | 13:31.38 |
| reminds me, I ought to do that rect/bbox thing | 13:31.52 |
Robin_Watts | This is for draw_glyph | 13:33.10 |
tor8 | right, so not solvable that way | 13:34.31 |
Robin_Watts | Also... in fz_intersect_bbox and friends... | 13:38.08 |
| we check first for the box being infinite, then for it being empty. | 13:38.22 |
| I think we should check for it being empty before we check for it being infinite. | 13:38.33 |
| Consider the box with: x0 = 0, x1 = 0, y0=0, y1=-1 | 13:39.05 |
| That's empty (because of the x range). | 13:39.25 |
| but we detect it as being infinite because of the y range. | 13:39.35 |
kens | chrisl ping ? | 13:57.42 |
| opps oh no not there | 13:57.56 |
| tor8 Robin_Watts ? | 13:58.01 |
tor8 | Robin_Watts: infinite and empty rects *ought* to be limited to [0 0 0 0] and [0 0 -1 -1] | 14:16.39 |
| kens: yes? | 14:16.48 |
kens | :-) | 14:16.52 |
| Werner is looking for funding: | 14:17.13 |
| http://pledgie.com/campaigns/18808 | 14:17.14 |
| Given that GS and MuPDF use FreeType I was wondering if we could approach Miles | 14:17.31 |
tor8 | kens: I think that's a worthwhile idea | 14:18.27 |
kens | Thanks to8 I'll send Miles anemaikl | 14:18.55 |
tor8 | fatter fingers than usual today? ;) | 14:19.09 |
kens | think my keyboard doesn't like the weather | 14:19.26 |
tor8 | *I* don't like the weather! | 14:19.38 |
kens | Well its terribly wet here | 14:19.46 |
| been raining all day | 14:19.53 |
| I suspect Robin's house may have a moat by now | 14:20.09 |
tor8 | it thawed here, now there's slush everywhere :( | 14:20.30 |
| well, beats rain I guess, rain in this weather is dreadful | 14:20.59 |
Robin_Watts | slush is worse than rain. | 14:36.24 |
Robin_Watts | pops to post office. | 14:36.49 |
| oh, eck... This monitor is doing wierd color strobing. | 14:39.56 |
| tor8: Reviews up. | 15:05.25 |
| 1 file shows differences, you can see it in my bmpcmp. I think they are progressions. | 15:05.54 |
| tor8: When do you stop for Chrimble? | 15:06.35 |
tor8 | Robin_Watts: saturday through wed | 15:07.25 |
| back on thursday | 15:07.30 |
| Robin_Watts: um, you added win32/Profile/*.obj in one of those commits :) | 15:08.20 |
Robin_Watts | tor8: fab. | 15:08.23 |
| tor8: eek. Will fix. | 15:08.31 |
tor8 | Bug 693503: Fix SEGV in glyph painting due to bbox overflow LGTM | 15:09.33 |
Robin_Watts | tor8: fix pushed. | 15:12.03 |
Viorel | Hi guys! I recently checked out the muPdf project as I changed my workstation, and it seems that /xps/xps_xml.c is missing - can you please verify? | 16:21.02 |
Robin_Watts | Viorel: That file got moved... | 16:22.23 |
| to fitz/base_xml.c I believe. | 16:22.50 |
| What git revision are you on? And how are you trying to build it ? | 16:23.07 |
Viorel | um... and what can I do about this error when trying to ndk-build? "make: *** No rule to make target `jni/../../../mupdf/xps/xps_xml.c', needed by `obj/local/armeabi/objs-debug/mupdfcore/__/__/__/mupdf/xps/xps_xml.o'. Stop." | 16:23.33 |
| I got the latest revision, cloned it about 3 hours ago, and tried a pull a few minutes ago, after I saw the error | 16:24.24 |
| Commit:af6fd921f12c0e718d2a984789997098ef098e0c | 16:26.30 |
Robin_Watts | Viorel: Let me check. | 16:27.34 |
| When I grep in android, I can't find any mention of xps_xml.c | 16:28.58 |
| There used to be a reference in android/jni/Core.mk, but that's gone now. | 16:29.14 |
Viorel | Great, thanks! I then must update the local core.mk to reflect the current situation. | 16:31.23 |
Robin_Watts | Viorel: Oh, so you're building your own app based on our core ? | 16:31.51 |
Viorel | I had the old core.mk, from the other workstation | 16:32.49 |
| yeah, I'm making some slight changes to it, I've told you about the 2-page issues I encountered, they're solved now | 16:34.04 |
Robin_Watts | cool. | 16:34.14 |
ray_work | everybody at my house is saying how cold it is. It's down to 4, but the grass has frost on it. | 16:49.56 |
kens | 4 fahrenheit | 16:50.13 |
| ? | 16:50.17 |
| THat is cold | 16:50.22 |
ray_work | nope, 4 C | 16:50.24 |
kens | still relatively cool | 16:50.32 |
ray_work | 38 F | 16:50.34 |
kens | I'm fairly sure its warmer than that here, but very, very wet | 16:51.22 |
ray_work | it's cold enough that my office is actually not overheated. The thermostat is in another office and usually when it's cold outside, the heater switches on -- which I don't need with peeves and my other system always on | 16:51.51 |
Robin_Watts | ray_work: Have you done the DNS change forhttp://www.artifex.com yet? | 16:52.33 |
ray_work | Robin_Watts: oops. Nope. | 16:53.34 |
| Robin_Watts: what's the new IP, again, please ? | 16:53.48 |
Robin_Watts | 188.121.46.128 = new.artifex.com | 16:54.12 |
ray_work | Robin_Watts: OK. Thanks. I'll get to that when I finish an email reply... | 16:55.00 |
kens | OK off I go. Goodnight everyone | 17:02.22 |
JakeSays | i'm studying the svg output device, and i notice its mostly implemented by functions in gdevvec.c. is gdevvec.c a 'generic' vector driver? i'm trying to understand the relationship | 17:04.45 |
Robin_Watts | JakeSays: Yes, AIUI. | 17:05.12 |
JakeSays | hmm. that might be a better option for what i'm trying to accomplish then | 17:05.46 |
| er, except the svg driver appears to be very broken | 17:11.05 |
Robin_Watts | JakeSays: svgwrite? | 17:11.34 |
| henrys is actively working on that now. | 17:11.43 |
| I don't know whether he's committed his code yet. | 17:11.59 |
JakeSays | Robin_Watts: is svgwrite the same as -sDEVICE=svg ? | 17:12.03 |
Robin_Watts | yeah, I think so. | 17:12.11 |
JakeSays | ah ok. cool. | 17:12.23 |
| henrys: ping | 17:13.02 |
| hmm. i'm curious as to why fprintf is not used in gdevtxtw.c | 17:21.24 |
Robin_Watts | ray_work: New artifex site appears live for me. | 18:45.36 |
| Should artifex.com redirect tohttp://www.artifex.com (hint: it should, really...) | 18:46.16 |
| ? | 18:46.20 |
| tor8: ping | 19:10.02 |
| tor8: Various patches for review. The last one (Fix problems with tiling) may benefit from some new bbox/rect functions. what do you think? | 19:11.49 |
ray_work | Robin_Watts: For that I think I need to change the DNS servers away from cnchost.com (use networksolutions DNS). Since XO isn't doing mail _or_ web hosting anymore, we need to anyway. | 20:50.58 |
| Robin_Watts: the only issue is that NS warns that it will cause "service interruption" for up to 48 hours, so I was thinking of doing that during XMAS | 20:51.42 |
| Robin_Watts: then we can drop XO altogether. | 20:52.11 |
| Robin_Watts: I will contact NS to make find out what the best way to do this is. | 20:53.05 |
| henrys: also see above discussion | 20:53.16 |
sebras | Robin_Watts: I read on your career page that I should continue to study mandarin. ;) | 20:54.53 |
| Robin_Watts: also when reading the licensing pages I'm thinking that they probably seem a bit alien to an android app developer. maybe something to state that commercial licensing is available both for large corporations as well as one man android apps...? | 20:57.02 |
JakeSays | is there a way to convert a path to nothing but line segments? | 21:32.58 |
| sebras: iirc artifex is not too interested in one man shops | 21:33.48 |
saper | JakeSays: you mean no arcs etc.? | 21:34.04 |
JakeSays | saper: yes | 21:34.12 |
| saper: i have paths that render as straight lines, but they use curves | 21:34.48 |
| and i need to detect the straight lines | 21:34.57 |
saper | some math should do | 21:35.09 |
JakeSays | lol yes it would do nicely | 21:35.30 |
| but that would require me having the math | 21:35.58 |
saper | ghostscript is not mathematica :) | 21:36.06 |
JakeSays | wha..?? | 21:36.15 |
| i was lied to! | 21:36.20 |
| o | 21:36.21 |
| er, | 21:36.24 |
saper | wrong IRC channel then | 21:36.31 |
| JakeSays: btw. there is a simple patch to svg driver in bugzilla | 21:36.50 |
JakeSays | saper: but yeah, i was just hoping there was a utility function somewhere in the code base | 21:36.53 |
| saper: oh? i'll check it out | 21:37.14 |
saper | http://bugs.ghostscript.com/show_bug.cgi?id=692527 | 21:37.24 |
JakeSays | although i dont think i need svg at this point | 21:37.24 |
saper | helped me to generate some simple SVGs | 21:37.33 |
| I am not an expert but from what I've seen most of the vector image approaches follow ps model of moveto, lineto, arcto etc. | 21:38.09 |
JakeSays | saper: ah yeah, the svg issues i'm seeing go way beyond the fill= issue | 21:39.05 |
saper | SVG files generated by the master are completely unusable | 21:39.32 |
JakeSays | saper: i've managed to extend the txtwriter driver to capture the shapes i need though | 21:39.34 |
saper | Please try to apply the patch and see if it helps. | 21:39.41 |
JakeSays | saper: that patch only fixes the broken fill attribute, as near as i can tell. the svg i'm getting has that issue and many many more. | 21:41.44 |
saper | maybe those "many more" can be fixed, too | 21:43.40 |
JakeSays | they can, yes | 21:43.50 |
| they're just bugs | 21:43.51 |
saper | there is not much support for anything more advanced in svg output | 21:44.11 |
JakeSays | but for what i'm doing svg is overkill anyway | 21:44.12 |
sebras | JakeSays: no, I know, but unless they are mentioned people might assume that commercial licensing for the small players is not possible and contradict the licensing instead... | 22:04.26 |
JakeSays | sebras: well, i dont know a whole bunch about the licensing, but having quarterly minimums to meet makes it pretty much out of touch for small players | 22:25.10 |
sutula | http://mupdf.com/ pointed me here...apologies in advance if my question is off-topic. | 23:04.45 |
| Does anyone here know whether there's a page similar to: | 23:04.56 |
| http://svn.ghostscript.com/ghostscript/trunk/gs/doc/Commprod.htm | 23:05.03 |
| but that covers mupdf? | 23:05.11 |
| I'm looking to point a potential commercial user of mupdf to some document that might explain when they would need a commercial license. | 23:05.54 |
sutula | finds the page he wants in Google cache, but the live page seems to be missing: | 23:14.40 |
| Live: http://www.artifex.com/indexlicense.htm | 23:14.52 |
| Cached: http://webcache.googleusercontent.com/search?q=cache:5gZd6HcmwKEJ:www.artifex.com/indexlicense.htm+&cd=1&hl=en&ct=clnk&gl=us | 23:15.04 |
| Again, I'd appreciate any pointers from anyone with Artifex connections | 23:15.47 |
sebras | sutula: there's this page too: http://www.artifex.com/page/licensing-information.html | 23:20.18 |
sutula | sebras: That looks great, thanks! | 23:21.07 |
sebras | sutula: I'm just a hang-around, I don't work for artifex. just for the record. :) | 23:21.33 |
| sutula: though I know that when having questions about commercial licensing then Scott Sackett (mentioned on the page) is an appropriate contact. | 23:22.23 |
sutula | probably should have introduced himself better...I try to help teams in a big company understand the licenses and obligations so that we don't get into trouble or compromise relationships with open source developers. | 23:23.19 |
| sebras: Thanks much for the pointers. I'll leave this channel open in case others have guidance, but I thought Artifex's position was pretty clear. | 23:24.16 |
ray_work | JakeSays: BTW, PostScript (and thus ghostscript) has path flattening control that converts the Bezier curves into line segments | 23:58.41 |
| JakeSays: I thought you were the guy that only needed Redmon -- do I have you confused with someone else ? | 23:59.19 |
Robin_Watts | sutula: Hi. MuPDF is released under the GNU GPL. As such, if you can abide by the terms of the GNU GPL, then you should be fine. | 23:59.46 |
| Forward 1 day (to 2012/12/21)>>> | |