IRC Logs

Log of #ghostscript at

 <<<Back 1 day (to 2013/07/29)2013/07/30 
robin_watts_web Hi all.10:10.42 
chrisl robin_watts_web: Jeez, really can't keep away, eh? So where are you now?10:11.15 
robin_watts_web 50 yards up the side of a sand dune this morning in a really strong wind, it ripped my glasses off my face and out into space :(10:11.17 
  Fortunately I have a spare pair somewhere.10:11.35 
  Was just emailing my opticians to sort out a replacement pair for when I get home.10:11.55 
chrisl All else going well, though?10:12.32 
robin_watts_web chrisl: I think we're at sossiflei (have no idea how it's spelt)10:12.44 
  Yeah, having a great time so far other than that (and other than the usual hire car fiasco etc)10:13.06 
  I shall miss todays meetings 10:13.44 
chrisl Wikipedia seems to suggest: Sossusvlei10:13.53 
robin_watts_web Have a good one!10:13.54 
chrisl You too, stay safe!10:14.03 
robin_watts_web that sounds plausible.10:14.03 
tor7 paulgardiner: argh. it appears that you can't have a stencil buffer without a depth buffer :(10:18.36 
  so wasting 24 bits of unused depth buffer to get 8 bits of stencil buffer in a texture :(10:18.55 
paulgardiner tor7: Hmmm, not ideal10:20.35 
tor7 you can probably do it if you render to a Renderbuffer object instead of rendering to a texture10:20.44 
  hm. or maybe... it's possible to mix rendering to textures and renderbuffers10:21.11 
  meh, same story there... GL_FRAMEBUFFER_UNSUPPORTED if I've attached only stencil and color attachments. add a depth attachment, and it works :(10:27.27 
paulgardiner tor7: Two commites on paul/master. The changes to pdf_measure_text, in addition to solving the bounding box problem, have given us better positioning of text in form fields in general.11:01.17 
tor7 paulgardiner: first one is good.11:09.52 
  paulgardiner: in the second one, calling pdf_measure_text just to get the ascender seems a bit overkill11:10.08 
paulgardiner Oh yeah. Weird. I'll sort that.11:11.34 
  tor7: New version pushed11:20.15 
tor7 paulgardiner: looks good.11:28.26 
paulgardiner tor7: ta11:30.45 
tkamppeter chrisl, hi11:31.40 
chrisl tkamppeter: hi11:51.48 
tkamppeter chrisl, I want to change one thing in the upcoming GS 9.08: I want to move the gstoraster CUPS filter from GS to cups-filters as now it will need libcupsfilters. Perhaps I also move gstopxl and all associated files, leaving only gdevcups.c and cups.mak in the cups/ directory of GS.11:54.39 
chrisl tkamppeter: okay, that would be good from our perspective, too. Do you want to do the GS changes, or shall I?11:55.36 
tkamppeter chrisl, this way all PS/PDF->XXX filtering will be controlled by the now required cups-filters package.11:55.36 
  chrisl, I will remove the files and the lines in cups.mak but I am grateful if you could do the changes outside the cups/directory, especially documentation.11:57.00 
chrisl tkamppeter: I wasn't aware there was any documentation pertaining to that stuff, except the changelog, which I generate from the git log11:58.04 
tkamppeter chrisl, I also do not know whether there is doc to change, changelog and history naturally does not need to be changed.11:59.15 
chrisl tkamppeter: yeh, the only places I see them mentioned is in the change log. If you could note in your commit message where you're moving them to, that would be useful for anyone wanting to do something with them.12:00.55 
tkamppeter chrisl, OK, I am proceeding now.12:08.57 
sebras tor7: a few patches over at sebras/master to move x11 code around a bit and make it possible to show warnings in the statusbar.12:22.10 
  tor7: I made several separate patches to show my intent, if you want to you can squash them.12:22.37 
henrys mupdf meeting time?15:00.12 
mvrhel_laptop yes. sounds like Robin_Watts is having a good time15:00.39 
henrys oh I haven't been on fb lately is he posting there?15:01.22 
mvrhel_laptop oh I was just reading his comment on the logs15:01.51 
tor7 henrys: I've got path rendering using the nvidia extension working. I had to rewrite quite a bit, since the path rendering extension uses all the old deprecated opengl crap.15:01.59 
henrys I bought the good reader app it is really quite nice if you like marking up your pdf's15:02.02 
mvrhel_laptop climbing sand dunes15:02.02 
tor7 other than that, no real news.15:02.25 
  the release candidate is up, so if you could give that a spin would be good15:02.45 
henrys tor7:what does that mean in terms of rendering pdfs? we tiger? does that include text?15:03.11 
tor7 henrys: we tiger, in a different way.15:03.19 
henrys is it on a public branch somewhere, like to see some of that stuff - just curious.15:03.53 
tor7 going to try path rendering a third way, then on to text.15:05.07 
  henrys: tor/opengl2 and tor/opengl315:05.12 
  opengl2 has the fixed function stuff tigering with opengl 2.0 + extensions15:05.33 
henrys I didn't really have anything for the meeting this time about - just reviewing the agenda stuff … the meeting is coming up.15:05.35 
paulgardiner henrys: knowing what you found most useful from the good reader app that we don't yet support might be handy15:05.40 
tor7 opengl3 has the pure core profile (but not tigering, that code was too crap to push)15:05.46 
henrys paulgardiner: so far sketching with my finger 15:06.25 
  but I may like to doodle more than most 15:06.44 
paulgardiner We can do that but we've yet to get around to allowing other than one color and one line thickness.15:07.42 
henrys one of the things I'd like to do is pitch mupdf to them. Their pdf seems to be lacking some stuff and they are a top selling app.15:07.52 
paulgardiner I have text appearance stream creation via the pdf device working now (just for base14 fonts), so at last I can actually get back to the main thrust on signatures15:09.25 
mvrhel_laptop henrys: I should easily have the windows phone app in the store by meeting time. I hit a wall using the ready made controls in getting the rects for text search and hypelinks to work with the animations of the page movements so I am going to work on my own custom control this week15:09.45 
henrys sounds like it should be ready for chicago also.15:10.18 
  are you bringing a windows device?15:10.27 
mvrhel_laptop I will bring the phone and the surface both15:10.36 
henrys paulgardiner: seems like signatures will be tight schedule wise probably not a big deal we should have plenty to demo.15:12.18 
paulgardiner For Chicago?15:13.07 
henrys yes assuming 42 more interruptions between now and then.15:13.31 
paulgardiner I don't know. I'm still hopeful.15:15.50 
henrys anything else for the meeting? have a look at the agenda and see if there is any low lying fruit to be picked.15:16.07 
  getting that bountiable thing to zeniko looks fairly easy.15:16.44 
  also I am looking at patches to openjpeg and it doesn't look like they are being done by the real developers. They look like just in time fixups starting in column one which always makes me nervous about the future of a system.15:18.10 
  hi tkamppeter, it's mupdf meeting time anything to add?15:22.37 
  well let's call it done,gs meeting in 1/2 hour15:26.26 
  jeez... a county of dumbass rednecks in my state is trying to secede and create a 51st state. 15:31.43 
chrisl henrys: for a second I thought you meant secede from the Union.....15:33.47 
henrys these are the same dumbasses shooting at drones...15:34.23 
tkamppeter henrys, no, nothing new about MuPDF. For Ubuntu Touch it loks like that Popler will be the choice, especially as it is also part of desktop and we especially want convergence, mobile devide with monitor getting a desktop.15:34.28 
henrys tkamppeter: thanks I understand.15:34.57 
tkamppeter henrys, but continue with any support of using MuPDF in the (mobile) printing stack. Perhaps things will change or other mobile platforms will show up.15:38.43 
henrys well we are driving on with windows mobile, android and begrudgingly ios15:39.47 
ray_laptop morning, all15:52.20 
  henrys: did you see the email from cust 801 ?15:52.33 
  about PCL tray selection ?15:52.44 
henrys yes but I let marcosw1 preprocess everything.15:53.56 
ray_laptop henrys: OK. Where do I look for PCL tray selection, or do we need to ask them for a file to see how they are requesting tray 2 ?15:55.54 
henrys it would just be a device parameter, does marcosw want me to handle it?15:59.35 
  so I guess a release discussion is in order for the meeting?16:00.15 
marcosw1 morning henrys, I haven't read my email yet...16:00.37 
henrys chrisl:I didn't see any response to your release announcement, I'm ready to go.16:00.56 
chrisl henrys: I haven't seen any replies, either, but then, that's not unusual16:01.27 
henrys they'll just tell you to wait after you prepare everything ;-)16:01.56 
chrisl Yeh, and I'm tired of playing that game :-(16:02.12 
henrys marcosw1:I should have an update on the blue plane issue later today but I haven't started anything with the the new issue.16:02.56 
chrisl TBH, all I wanted to check today is whether ray_laptop wants to get "saved page" in, or whether we should go ahead without it16:03.01 
henrys in case you want to update them.16:03.04 
marcosw1 henrys: if you could handle the tray selection email from customer 801 that would be helpful, I have lab meeting at uni today.16:03.28 
  henrys: okay will email re. the blue plane issue.16:03.39 
henrys marcosw1: will do.16:03.41 
  ray_laptop:you were saying you were surprised how anything worked in the logs… but I didn't understand the context of that. Was PCL doing something weird?16:04.55 
  mvrhel_laptop: anything for the release?16:05.18 
ray_laptop henrys: I was talking about the gxclmem problem on Windows. When we switched to gs_sprintf (trio sprintf) it wasn't compatible with sscanf on Windoze16:06.28 
mvrhel_laptop nothing from me. 16:06.53 
chrisl ray_laptop: sorry about that, I didn't realise they behaved differently16:06.58 
ray_laptop PCL builds with memory based clist by default and pattern-clist is always memory based clist16:07.00 
  chrisl: neither did I16:07.08 
  chrisl: I just added the one function I needed gs_sscanf Feel free to populate it with more16:07.49 
chrisl ray_laptop: that's fine by me, we'll just add any extras as we need to, I think16:08.24 
  ray_laptop: the other thing I was going to mention about that is that trio is unlikely to ever get an upstream update, so we can pretty much hack on it any way we want to suit our needs16:08.33 
ray_laptop chrisl: no on the saved-pages. I'll save it for post release16:08.49 
chrisl ray_laptop: okay, thanks16:09.03 
henrys I don't think we have blockers - I have a very minor fix to go in today.16:09.24 
  it doesn't matter if it is released or not.16:09.37 
chrisl henrys: on that basis, I reckon I'll take the release branch tomorrow - I'd like to check with Ken first16:09.40 
ray_laptop chrisl: the way it works is internally consistent, so it's OK. I have no problem with having the leading 0x on pointers come from the %p16:09.42 
  so the openjpeg segv is resolved ? If not, I would call that a blocker16:10.30 
chrisl That's not a regression, is it?16:10.50 
ray_laptop I was still getting the segfaults, but I hadn't updated from master since yesterday16:10.57 
henrys there is a fix in openjpeg2 but the integration is stalled.16:11.38 
  it is not a regression16:11.53 
chrisl henrys: Shelly has made progress on OpenJPEG2, but is on holiday this week16:12.05 
henrys if you are talking about the tile size div by 016:12.08 
chrisl henrys: is the OpenJPEG2 fix something we could easily backport?16:12.52 
henrys I don't think any of zeniko's summatra pdf test files should hold up the release.16:12.54 
ray_laptop so we are punting on openjpeg2 for the release as well ?16:13.37 
henrys probably but I wouldn't fool with it.16:13.44 
  yes unfortunately mupdf will have openjpeg2 and gs will not.16:14.18 
ray_laptop chrisl: were you thinking to do it after the release candidate ? That sounds dicey, but I guess if it passes regression I'd just leave it up to you16:14.51 
henrys moot for custormers16:14.53 
chrisl ray_laptop: well, almost all the distributions still use Jasper, customers use luratech, so.......16:15.35 
ray_laptop henrys: true -- they use luratech. I haven't looked at the weekly luratech to see if it ran since the sumatra files went it16:15.39 
  but they don't use jasper with gs do they? didn't we rip out the hooks ?16:16.13 
chrisl ray_laptop: we have now, yes, good point16:17.20 
henrys has anyone looked at the publisher of zeniko's pdfs openjpeg2 is rare enough but pathologically broken ones, really?16:17.21 
ray_laptop chrisl: so if a distro wants to use gs with jasper, they have to pull in the hooks from old gs. Not too likely16:18.02 
marcosw1 ray_laptop: the luratech build runs on Monday16:18.13 
chrisl ray_laptop: yes, I'd just forgotten you'd ripped out the jasper cr*p since 9.0716:18.34 
  ray_laptop, henrys: if feelings are strong on the subject, I can take a look tomorrow - we might be able to safely throw an error instead of crash?16:19.41 
ray_laptop marcosw1: Monday's luratech run ALSO has 29 segfaults and 148 new errors16:19.44 
  looks like it's common to the decoder, so detecting it is probably possible to do in our code16:20.23 
  s/common to/common among all/16:20.57 
henrys chrisl: I'd prefer leave it as is without a reported regression.16:20.58 
marcosw1 the Monday errors were almost all in the pxlcolor/pxlmono path or tiffscaled. I'm surprised by that.16:21.10 
chrisl henrys: okay. I must admit, I wasn't too happy about adding a bunch of new test files days before a release!16:21.51 
henrys they are grouped together and easily ignored.16:22.14 
  other than nagging everyone including me to look at the agenda and do a few items, I got nothin' else.16:22.51 
chrisl henrys: on the OpenJPEG2 front, I did some work with Shelly before he went on holiday, and I think we know what the remaining problems are..... fixing them is the next thing16:25.09 
ray_laptop henrys: is there a PCL command which is supposed to perform tray selection? or is it PJL ? I grepped for 'tray' in the pcl and pl .c files and didn't see anything16:25.41 
mvrhel_laptop I did fix one segv that was happening with psdcmyk16:25.55 
chrisl mvrhel_laptop: I saw you'd got that done - nice work!16:26.15 
ray_laptop henrys: I do see ESC & l <num> H that sets %MediaSource16:27.50 
henrys right there is an XL command also16:28.06 
  I have to correspond with them to understand what they want.16:28.18 
marcosw1 figured out why the Monday errors were all in pxlcolor/pxlmono path or tiffscaled, I had added all the errors to expected_errors.txt but neglected those.16:28.19 
henrys ray_laptop:If you prefer to talk with them I can send you a paragraph of stuff and you can rephrase it if you like. Let me know.16:29.09 
ray_laptop henrys: PXL also has a way to set MediaDestination16:29.28 
henrys right media source and destination.16:29.52 
  from the assembler code: 'MediaSource' : [ 'eDefaultSource=0', 'eAutoSelect=1', 'eManualFeed=2',16:30.17 
  'eMultiPurposeTray=3', 'eUpperCassette=4', 'eLowerCassette=5',16:30.18 
  'eEnvelopeTray=6', 'eThirdCassette=7', 'External Trays=8-255' ],16:30.19 
  'MediaDestination' : [ 'eDefaultDestination=0', 'eFaceDownBin=1', 'eFaceUpBin=2',16:30.20 
  'eJobOffsetBin=3', 'External Bins=5-255' ],16:30.21 
  tools/ better than the docs ;-)16:31.14 
ray_laptop henrys: thanks. So that seems to mean that to take paper from a specific tray, you use MediaSource16:31.40 
ray_laptop goes to check what that does in the library or devices16:32.00 
henrys pxlcolor or pxlmono should do something interesting.16:32.33 
  ray_laptop:so do you want to take that over, I'm fine with that.16:33.02 
ray_laptop henrys: fine. I will do it16:33.32 
henrys I imagine there is PJL also, we don't have it but it is trivially added.16:34.07 
  let me look that up for you.16:34.14 
ray_laptop since I already did some messing around for them with tray selection for PS16:34.21 
  henrys: I didn't see anything in PJL16:34.44 
  grep Media pl/* came up blank16:35.46 
henrys We don't implement all of PJL but there are documented commands that are easily added, if they want to use them.16:36.04 
ray_laptop henrys: if they ask for that, I'll let you handle that :-)16:37.14 
  hmm... my saved-pages code came up with "Differences in 23119 of 59847 non-pdfwrite/ps2write test(s):" but bmpcmp (which only checks 1000) came back with no errors. Now what16:40.44 
  The few I've spot checked all look identical as well16:41.06 
  although I've only checked on windows. I guess I'll move to peeves and see if I can see differences with cmp of the files16:41.51 
  since the test doesn't have --saved-pages mode enabled, they aren't supposed to differ16:42.37 
  marcosw1: when this does (eventually) go in, I've added a --saved-pages-test option that will run the job in an automatic saved-pages mode where all the clist files for the job are collected before any output starts16:43.59 
  marcosw1: so we will want a weekly run of that mode16:44.16 
chrisl ray_laptop: indeterministic behaviour? I had one a while back that would only wrong on older versions of gcc16:44.27 
ray_laptop marcosw1: It is implemented both in PS/PDF and plmain (PCL6 and XPS)16:45.01 
  chrisl: going to check now, but if it's indeterministic I'd expect at least a few bmpcmp errors to show up out of 100016:46.01 
chrisl ray_laptop: yes, I'd expect that, too. Although, another issue I had recently produced different output on older vs newer gcc versions - but each was consistent with itself.16:48.31 
marcosw1 ray_laptop: if you need help figuring out why there were cluster differences that bmpcmp didn't find let me know; I have to run out now a but will have time early this afternoon.17:08.02 
ray_laptop marcosw1: no, I don't know why bmpcmp didn't report errors, but I am able to reproduce it (for the first file I checked)17:30.16 
tophyr hello hello - i have some questions about mupdf and licensing, have i come to the right place?18:40.53 
henrys tophyr:go ahead and ask19:07.19 
tophyr i'm wondering if a potential use i'm thinking of would count as "dynamically linked" or "saas". Android allows one app to communicate with another, as complete units. i have an app that i'd like to add pdf rendering capabilities to, but i can't open source it and don't have the coin for commercial licenses.19:11.44 
  i'm considering building a no-frills reader based upon mupdf, that can accept "intents" - basically android app-to-app RPC calls - and will generate a .png from a .pdf upon demand19:12.33 
  the closed-source app would, if the converter/reader is installed, talk to it and use it to generate images that it can then itself render on screen19:13.03 
  they wouldn't be distributed together, however, and would have no dependencies on each other, other than "something on the device must be able to handle this kind of intent, in order to render a pdf. here's one option: <link to google play store>"19:14.23 
henrys tophyr:we cannot condone something like that … keep things GPL or get a license. You can probably find loopholes in the licenses as well as I can and do something with shared libraries but we don't endorse that. Contact scott.sackett at artifex dot com maybe he can come up with a workable license for you. 19:16.01 
tophyr well, and that's really the limits of my understanding, what constitutes a "shared library" or not.. but that's obviously a very gray area anyway. i'll definitely get in touch with scott, thanks for that!19:17.44 
ablemike Hello All. 21:17.16 
  I am looking for a package that will assist me in some operations on a PDF file.21:17.18 
  Essentially, I need to crop a PDF to a fixed width and height.21:17.25 
  And then scale that PDF and it's embedded artwork to another fixed height and width.21:17.33 
  Am I in the right neck of the woods? 21:17.43 
ray_laptop found what I messed up with the saved-pages changes (not committed yet). It worked fine *only* in saved-pages mode, I broke clist rendering without it. :-~21:22.47 
  running regression now... bbiaw21:23.30 
henrys ablemike: usually folks come here after they've tripped over stuff they already tried. a google of cropping and ghostscript will get you this:, probably your other question is easily found also, then if you have problems please come back and ask questions.21:46.49 
ablemike i'm on day 4 of googling :)21:47.10 
  i've hit that page you suggested a 1000 times21:47.17 
  the issue I have with this gs command, is it sets the cropbox correct and renders in viewers correctly21:47.43 
  however trimbox, mediabox, etc are eluding me21:48.06 
  when I open my CropBoxed pdf in Illustrator (like in the command in one of the answers), I get all kinds of extra space21:48.51 
  I want to take the contents of the cropbox and just force it to that size21:49.05 
  and I can't find anything out there that will do it21:49.12 
  PDFCrop / Briss / Poppler ... pdftoeps to jpg to blah 21:49.35 
ablemike is feeling a bit hopeless 21:50.14 
henrys is there an associated stack overflow question?21:54.19 
  ablemike: ^^^21:54.38 
ablemike I haven't posted one yet no. 21:54.57 
  i guess i view SO as "heres my bug, help me debug" 21:55.46 
henrys I'm not quite grokking exactly what you want to do and I was hoping for a more detailed question and perhaps a preliminary response from the experts over there.21:57.09 
  (i.e. the exact command line you are using stuff like that) are you producing pdf. Saves us time if we have a complete description of the issue like on stack overflow first.21:58.30 
ablemike yup yup21:59.23 
  I'll post one when I get home. 21:59.27 
dogisfat Quick question, is there a way to print the libpng version used by ghostscript?22:01.31 
henrys dogisfat: I don't think so, you'd have to look at the source22:03.07 
dogisfat Thanks22:28.36 
sebras tor7: (for the logs) you mentioned that you planned on looking at sebras/master, but I haven't heard from you since and there's nothing in the logs. holler when you're done reviewing.22:51.40 
ray_laptop dogisfat: The version of libpng will vary if you are using .so packages. For obvious reasons, we only test and respond to issues with the source we distribute since what random sites have installed is "who knows what"23:21.02 
  dogisfat: and that goes for all other third party libs as well: expat freetype jbig2dec jpeg jpegxr lcms* (libpng already mentioned) openjpeg* tiff trio zlib23:24.25 
  jbig2dec was developed by us, and we support it (assuming someone else hasn't mucked with it when generatiing a .so)23:25.26 
  but the others are true third party apps, NOT particularly well tested, especially with Ghostscript/GhostPDL so "all bets are off" with distro .so libs23:26.28 
  ALLRIGHT! No bmpcmp differences23:27.44 
  marcosw1: is there a way for me to test my branch with an additional command line option before the file? I want to test with --saved-pages-test immediately before the command line (the important part is that it follows the -sDEVICE=)23:29.21 
  I guess I could just commit as is :-/23:31.19 
  have to pick my daughter up from her show choir camp. bbiaw23:32.16 
 Forward 1 day (to 2013/07/31)>>>