| <<<Back 1 day (to 2012/03/26) | 2012/03/27 |
robin_watts_mac | leaves nice hotel. 8 hour drive to dodgy one. Wish me luck :) | 02:11.27 |
ray_laptop | my laptop is seriously sick. Sometimes I haven't even done anything on it since it booted and it crashes and reboots :-( | 05:26.50 |
| it is REALLY frustrating trying to get stuff done. I guess I'm going to have to get a new laptop, load all my stuff over and send it to Dell for diagnostic/repair :-( | 05:29.46 |
| it is really bad trying to do performance testing when after a reboot I get drastically different timings for about the first 1/2 hour -- if it doesn't reboot before that | 05:30.58 |
| the only good news is that laptop prices have come down a lot since I bought this one -- I can probably get an i7 instead of an i5. The only thing I give up is screen resolution (currently 15.6" 1920x1080) | 05:33.36 |
jmari | hi? | 09:06.25 |
henrys | we've managed to avoid VDP I wonder if we should give it a go, I was just reading http://www.globalgraphics.com/news/2012/03/21/unprecedented-speed-of-new-digital-harlequin-rip/ | 14:34.44 |
kens | I wonder what a 'regular PDF file containing variable data' is ? | 14:36.31 |
| So they can do multi-threaded rendering, they've made their transparency faster (which was slow before) and they've added variable data printing. Typical Harlequin hype. | 14:38.05 |
| Mind you, I suspect getting multiple threads to run on the Harlequin display list might have been quite a tiask :-) | 14:38.39 |
henrys | kens:I don't know about the vdp on a regular file. It's a request I hear about once per show and then seeing that announcement we have to at least consider it. | 14:41.47 |
kens | We've seen it a lot in the past, someone has a 'real need' for it. Until they discover more about it.... | 14:42.27 |
| No doubt there are some people who really need it, but the last I heard (some time ago)_ the real high speed merchants were rendering the pages offline and seinding pre-rendered bitmaps to the high-speed printers. | 14:43.07 |
| Because it was the only wy they could keep up. We did pitch variable printing, but they weren't interested. Of course things may have changed | 14:43.43 |
| XPLOR would be the place to find out I would think | 14:43.54 |
paulgardiner | Is HHR a competitor? | 14:44.09 |
kens | GGS ? | 14:44.17 |
| paulgardiner : ? | 14:44.45 |
paulgardiner | I was wondering what Harlequin Host Renderer was. | 14:45.17 |
kens | Oh, its a version of the Harlequin rip | 14:45.28 |
| Which is a competitor | 14:46.39 |
henrys | maybe marcosw can poke around drupa and find out if it is real or still an unused feature checkbox. | 14:47.22 |
paulgardiner | Does MuPDF have compatibility layers for competitor's APIs - i.e., thin layers that would allow a client to drive MuPDF via a competitors API. Maybe that's a hard thing to do, but I was just thinking that one barrier to a potential client swapping is the amount of work in changing to a different API. | 14:49.11 |
henrys | paulgardiner:no neither does ghostscript. We really don't have access to competitor API's for both GG and Zoran we'd have to sign NDA's to get the code if they'd be willing to give it to us at all. | 14:51.35 |
kens | Harlequin doesn't really have an API. | 14:51.57 |
paulgardiner | Oh dear! :-) | 14:52.12 |
kens | Jaws does and it was deliberately written (at the high level) to be similar to the CPSI API | 14:52.15 |
henrys | kens:oh that's interesting I didn't know that. | 14:53.36 |
paulgardiner | Henrys: Yeah, I sort of imagined there would be legal issues. | 14:54.14 |
kens | Often hard to prove. | 14:54.49 |
| If you change the names and muddle with the parameters a bit. | 14:55.04 |
henrys | I doubt that is a relatively big barrier to switching though. | 14:55.26 |
paulgardiner | A barrier to giving it a quick try though. | 14:56.04 |
kens | Mainly the problem is with custom development which is usually not portable at all, since the internals of interpreters vary so widely | 14:56.07 |
| Even 'simple' things like colour management | 14:56.29 |
| Or (real workld example) image processing | 14:56.42 |
chrisl | Hmm, HQN has been using threads for several years, so if not for proper MT rendering, I wonder what they were doing......? | 14:59.39 |
kens | Runnig hte GUI probably :-) | 14:59.52 |
chrisl | No, it was definitely a performance thing | 15:00.37 |
kens | Dunno then | 15:00.52 |
| Like I said Harlequin hype | 15:00.58 |
chrisl | Yeh, like "Harlequin's custom, synthetic font substitution...." | 15:02.04 |
| Or whatever their MM font thingy was called | 15:02.31 |
henrys | chrisl:they've been MT'ing bands or front/back end in parallel for years? | 15:02.42 |
| or both? | 15:02.46 |
| they seem to have gotten the the big HP presses maybe we should try hyping? | 15:03.57 |
chrisl | henrys: I don't know, that's what I was saying. I *know* they've been using threads internally for a number of years, I had assumed it was for parallel band rendering, but in the light of what's mentioned above, maybe not | 15:04.12 |
| henrys: you have *no idea* how much time and effort they ended up devoting to that HP deal! | 15:04.42 |
henrys | When I read it I thought they had MT bands and now they've added front/back. | 15:04.51 |
kens | Looks like henrys is right. | 15:06.33 |
| THey are talking about processing mutiple pages at once. | 15:06.44 |
| Which you can only do with PDF of course. | 15:06.59 |
| Or DSC-compliant files. | 15:07.06 |
chrisl | henrys: That's a possibility - my take with Jaws was that *thread safety* was the biggest thing we needed - the fastest solution is going to be multiple rips in parallel (again for PDF) | 15:07.16 |
kens | Actually, I take that back. | 15:07.24 |
| Looks like htey are handing off the rendering to the renedeering threads, and then starting a new display list with the next page. | 15:07.46 |
| Which presumably isn't hard if you already have multi-threaded sisplay list rendering. | 15:08.14 |
chrisl | Whether that's worthwhile probably depends on how complex your display list format is, and much effort it takes to create it | 15:08.39 |
kens | Also depends on the complexity of the input. If interpretation is significant then it could be a useful saving. | 15:09.15 |
chrisl | Anyway, I always take HQN performance claims with a large pinch of salt - I was rarely able to reproduce the claims with real world jobs. | 15:10.57 |
chrisl | wonders what a "digital" RIP is.......... surely, by definition....... | 15:16.47 |
henrys | hmm a qnx employee porting mupdf to qnx. Intereting. | 15:21.05 |
sebras | henrys: not surprising since qnx was bought by rim, and to compete with android... | 15:22.25 |
henrys | sebras:right. | 15:27.43 |
Salil | Hey I am using GhostScript API and during the call to initialize with file gs_init.ps I get an error e_rangecheck (-15). What could be the reasons for this? I am trying to port GhostScript to iOS. | 15:31.03 |
chrisl | Salil: rangecheck is a Postscript error, it could be coming from innumerable different places. It's pretty much impossible to guess with being able to debug it | 15:33.58 |
Salil | chrisl: Thanks. I am using 8.63 as I was only able to compile framework on Mac for that. though I can compile for arm I can't include framework in my XCode project due to some restrictions from Apple I guess. So I am working with a static library. As I am working with static I have added all ps files from lib/ folder in the project and passing the path of that folder in -I. API is able to read gs_init.ps for sure. | 15:37.21 |
| chrisl: Now which other files should be included in the project? I can't find font files :( | 15:37.48 |
henrys | Salil:will you be able to contribute your work? | 15:37.57 |
Salil | I would love to. I am working for a corporate so I am not sure how that would work out. I hope you understand. | 15:39.29 |
henrys | MuPDF is already ported to iOS, we believe that technology is more suited to iOS, of course if you need postscript MuPDF won't help. | 15:41.46 |
Salil | henrys: I need to convert to pcl from pdf. I think MuPDF won't solve that. | 15:42.34 |
| henrys: Is there anyother library than GS I can use? | 15:43.09 |
henrys | the default pcl output by ghostscript is really just a wrapper around raster unless you use the higher level gdevpx devices for pcl6 - what device are you using? | 15:44.31 |
chrisl | Salil: The font files are include in 8.63 in Resource/Font | 15:45.49 |
| henrys: we probably need to think about our position on the Mac framework build - currently it is pretty much moribund....... | 15:46.40 |
henrys | robin_watts_mac did an xcode thing right? | 15:47.16 |
chrisl | I thought that just built a normal executable | 15:47.53 |
henrys | paulgardiner:without Robin_Watts_Mac and unless tor8 has something I think we can wait until next Tuesday. Of course if you need anything please ask. | 15:47.56 |
Salil | henrys: I have used commandline to produce output using hpdj340 (-sDEVICE) which worked for me. I wanted pcl3 but somehow it didn't work though on one of linux distribution it worked for me. | 15:48.06 |
| chrisl: Thanks but it results in the same error. | 15:48.45 |
henrys | chrisl:I'm able to be use the project in the ide - debug and stuff what else does it need to do? | 15:48.49 |
| we just need a dyld target? | 15:49.22 |
chrisl | henrys: is that all a "framework" is, then? | 15:49.39 |
henrys | I am not sure I haven't fooled with it. | 15:50.29 |
Salil | I am not sure but I could run the app on my device when I added the Framework or shared library (.dylib). | 15:50.30 |
| I am not sure but I could *not* run the app on my device when I added the Framework or shared library (.dylib). | 15:50.50 |
chrisl | Salil: my comment was really aimed at the existing MacOS build than a (currently) non-standard port | 15:51.59 |
henrys | chrisl:I would think the framework is completely bitrotted I've heard nothing about it in years. | 15:52.15 |
chrisl | henrys: which is what I meant - should we officially deprecate it, or it actually working again? | 15:52.58 |
henrys | I believe the future is starting from autoconf -> xcode and then doing what's needed. | 15:53.33 |
chrisl | Salil: you said it definitely reads gs_init.ps - do you know if it actually interprets that file, and if so, do you know how far it gets? | 15:53.46 |
| henrys: yes, the framework target has never been in the autoconf build - my feeling is we should remove the bitrotted makefile, then. | 15:54.51 |
henrys | https://developer.apple.com/library/mac/#documentation/Porting/Conceptual/PortingUnix/compiling/compiling.html | 15:55.36 |
Salil | chrisl: I am trying to debug that. I haven't included the files from GS to XCode so debugging by including print statement. Trying to find other ways to debug efficiently with gdb | 15:55.36 |
chrisl | Salil: when it's Postscript, your only real option is to add print operations into the PS file | 15:56.32 |
henrys | chrisl:I agree we should deprecate it. | 15:56.43 |
chrisl | henrys: I'll put it on my list for 9.06. | 15:57.09 |
ray_laptop | chrisl: Salil: sometimes the information you get from -dOSTACKPRINT -dESTACKPRINT can identify the problem. But at the least you need the "offending command" from the Error output | 15:58.17 |
Salil | chrisl: I am trying to convert from pdf to pcl and then send that to a printer. I couldn't understand adding print operations into the PS file. I am sorry I don't have enough background on this subject :( | 15:58.18 |
henrys | need coffee before the meeting. | 15:58.18 |
kens | best be quick | 15:58.33 |
ray_laptop | we can finish the meeting before henrys gets back with his coffee ;-) | 15:59.02 |
Salil | ray_laptop: Thanks I will look into that. I will report if I find something. | 15:59.17 |
chrisl | Salil: gs_init.ps is a Postscript file - it would be useful to establish if it's even reaching the point of the trying to interpret your PDF. | 15:59.25 |
henrys | hah I made it. | 15:59.34 |
| but now I can't remember what I was going to talk about | 15:59.51 |
ray_laptop | Salil: addding -dDEBUG will print out the progress while processing the gs_init.ps (and other init files) | 16:00.14 |
henrys | ah right alexcher how is the portfolio issue, marcosw sent you email? | 16:00.32 |
Salil | ray_laptop: thanks a ton! let me see | 16:00.39 |
ray_laptop | let's see if my laptop will stay alive through the entire meeting :-( | 16:01.18 |
alexcher | henrys: It's almost ready. I'll check it after the meeting. | 16:01.25 |
henrys | and kens I noticed a very noticeable problem with server mode if you run the %d thing with say the plrm and -Z? it will print a memory problem. I think that shouldn't be hard to track down should I create a bug? | 16:02.09 |
paulgardiner | henrys: Yes sure: next Tues will be fine. | 16:02.14 |
henrys | tor8:anything for paulgardiner? | 16:02.27 |
kens | henrys, isn't that already covered by the server crash bug ? | 16:02.30 |
| I'm flailing at Till's latest one at the moment | 16:02.43 |
| So I haven't had a chance to look further at these problems | 16:02.54 |
henrys | kens:I suspect that causes the crash using -Z? will shave about 42 hours off of finding it though. | 16:03.19 |
kens | Add it to the existing bug ? | 16:03.36 |
henrys | will do. | 16:03.41 |
| I haven't been following Till's latest anything the group can help with? | 16:04.08 |
kens | Its a pdfwrite problem. The PostScript relies on the /.notdef glyph being used when it asks for a space, because there is no /space glyph in the font. | 16:04.44 |
henrys | mvrhel:when do you think you'll get around to the planar stuff? Can we hand off any of your customer bugs? | 16:04.49 |
ray_laptop | kens: what does Acrobat do ? | 16:05.18 |
kens | IMO its very bad PostScript, and can never be ued to (eg) create a PDF/A file, so it should be fixed by the original application, but I'm trying to mimic what Distiller does. | 16:05.19 |
| ray_laptop : uses the /.notdef width | 16:05.29 |
| But if I do that, 2 of our other test files show regressions | 16:05.39 |
| They (of course) are PDF files, so I can't etst them against Acrobat | 16:05.53 |
| (Distiller) | 16:05.58 |
mvrhel | henrys: I am hoping to kick it into overdrive here starting Friday. I have 10 full days with no wife or kids. I will look over my customer bugs today | 16:05.59 |
ray_laptop | kens: so when a glyph is missing in the Encoding, what's wrong with using the .notdef glyph ? | 16:06.20 |
mvrhel | henrys: I am still working right now on the tiffsep changes | 16:06.35 |
kens | ray_laptop : Its *in* the Encoding, but there is no CharString. | 16:06.35 |
henrys | tor8:mupdf 1.0? | 16:06.37 |
mvrhel | I am in the shading code. I have one final issue in that it appears | 16:06.54 |
kens | ray_laptop : at the moment, when the glyph is missing, we set a width of 0. To make this file work I need to set a width = the width of the /.noptdef. | 16:07.13 |
mvrhel | then I still need to make sure all is fine with patterns and also do a bit of work in transparency | 16:07.16 |
ray_laptop | kens: I see -- but I agree that it is a bad PS file. Probably won't run on any real PS printers | 16:07.21 |
kens | ray_laptop : but as I said, if I do that then two of oru tets files show regressions. | 16:07.34 |
henrys | mvrhel:okay | 16:07.41 |
kens | ray_laptop : I also discovered that if we get an error with glyphshow (in pdfwrite) due to the glyph being missing in the font, we don't raise an error.... | 16:08.03 |
ray_laptop | kens: who created the PS file (not GS I hope) | 16:08.11 |
henrys | that's all I had this week. Does anyone have more for the meeting? | 16:08.16 |
kens | ray_laptop : provenance is 'difficult' Till claims it came from Okular, but its been through a PDF to PostScript conversion (NOT ours) | 16:08.35 |
| So I don't knwo where the problem was introduced. | 16:08.52 |
| My guess is the original PDF was bad. | 16:09.01 |
ray_laptop | kens: and tkamppeter doesn't have the original PDF ? | 16:09.31 |
kens | I haven't asked. Its not really relevant, because I'd still haver the problem with the PostScript.... | 16:10.02 |
chrisl | ray_laptop: unfortunately, I've seen Adobe produced PS that does something similar (missing CharString), so...... | 16:10.41 |
kens | Till says it was produced by Okular, but the PS file says : | 16:11.29 |
| 'Produced by poppler pdftops version: 0.18.4 (http://poppler.freedesktop.org)' | 16:11.29 |
henrys | mvrhel:I saw a video add on youtube - QNX + html5 for cars, I didn't realize they were a big player in the car business, I knew about RIM... | 16:11.36 |
mvrhel | oh wow | 16:12.00 |
henrys | I guess QNX made the video. | 16:12.00 |
ray_laptop | kens: great! so we can just say it's a poppler bug :-) | 16:12.09 |
chrisl | kens: Okular will be using poppler/cairo as a rendering engine/back end | 16:12.13 |
kens | All I'm saying is I don't know who to blame. | 16:13.16 |
chrisl | kens: did you say it's also missing a MissingWidth entry? | 16:13.55 |
kens | There is no MissingWidth, its a type 1 font | 16:14.15 |
tkamppeter | kens, Okular uses Poppler to handle PDF. | 16:14.19 |
chrisl | kens: oh, sorry, forgot it was PS... doh! | 16:14.46 |
kens | tkamppeter the point is that I don't know if the PDF was wrong, or the PDF->PS conversion (though I suspect the original PDF) | 16:14.53 |
| I'm not giving up on it yet. | 16:15.15 |
| But the PostScript is sub-optimal at the very least, and either Okular or Cairo should fix it (like that's going to happen....) | 16:15.44 |
henrys | Adjourn whe you guys are ready I have a fire to put out. I'll be back in a bit. | 16:15.49 |
ray_laptop | kens: but if it's an invalid PS, poppler is still at fault -- even if the PDF was bad, it can either throw an error, or do the work to generate a well defined PS file | 16:16.15 |
| henrys: OK. | 16:16.32 |
kens | ray_laptop : Its not clear that its 'invliad' so much as bad practice. | 16:16.36 |
tkamppeter | kens, the original PDF is attached to the Ubuntu bug report, comment #1, "Test case", https://bugs.launchpad.net/gs-gpl/+bug/960989/+attachment/2908787/+files/UNI-ZM1_Wniosek_zm_danych.pdf | 16:16.49 |
kens | There is no glyph named /space in the font, so we use the /.notdef. Relying on the behaviour is bad, and means that the fgile could never be converted ot a PDF/A file (you may not use /.notdef in a conforming file) but that doesn't make it invalid | 16:17.40 |
| Looks to me like the original PDF was badly made. | 16:18.34 |
chrisl | kens: I think, "officially" it does make it invalid: each entry in the Encoding should match a key in the CharStrings dict - but back in the real world..... | 16:18.51 |
kens | By AFPL Ghostscript 8.53! Lol | 16:18.51 |
| Someone really needs to update their copy of Ghostscript..... | 16:19.14 |
| Especially since the 'Application' is PDF Creator 0.8.1 | 16:19.43 |
| I woudl not be surprised to find that the subset font is a bug in the *very* old copy of Ghostscript they are using. | 16:20.21 |
| But then, its likely that the original emission was PostScript, so maybe not. | 16:20.50 |
| SO it look slike we have a PostScript file which was converted to PDF by Ghostscritp, tehn converted back to PostScript by poppler being converted back to a PDF by Ghostscript. | 16:21.17 |
| Is it any wonder this doesn't work ? | 16:21.25 |
ray_laptop | kens: I don't think GS ever had a bug like that, but I may be wrong. More likely that the original PS had an embedded font that was broken and GS just put it in. | 16:22.31 |
kens | Till, I can' tupdate the bug report, you might like to put some of this in there. If possible it would be nice to get the PostScript from the original application *before* it was converted into the PDF file by Ghostscript 8.53 | 16:22.37 |
| ray_laptop : yes, its possible, again, I can't tell, keep pulling on that thread.... | 16:22.56 |
| Anyway, as I said, I'm going to carry on looking at it, tomorrow..... | 16:23.20 |
| My head hurts too much today | 16:23.35 |
ray_laptop | kens: reading the PLRM, I still maintain that the PS is broken: 1. It uses the character code as an index into the current fontâs Encoding array to obtain a character name. | 16:24.02 |
| 2. It invokes the glyph description by name. For a Type 1 font, it looks up the name in the fontâs CharStrings dictionary to obtain an encoded glyph description (called a charstring), which it executes | 16:24.04 |
kens | Yes, but the whoel point of a /.notdef is so that you can use it if a glyph is missing. | 16:24.28 |
ray_laptop | kens: pg 329 of PLRM-3rd ed. | 16:24.31 |
kens | The problem as I see it is *relying* on the behaviour of a /.notdef. In effect whenever you sue a /.notdef glyph you are in an error condition. | 16:25.39 |
ray_laptop | kens: OOPS. pg. 330 says: If an encoding maps to a character name that does not exist in the font, the name .notdef is substituted. | 16:26.02 |
kens | :-) | 16:26.09 |
| Its actually the 2 PDF files that show regression that I'm working on. | 16:26.48 |
| When I discovered that our code unilaterally assigned a width of 0 to the glyph (in the /Widths array) when the glyph was missing I figured we should actually sue the /.notdef width | 16:27.29 |
| Since the /.notdef will be substituted this makes the /Widths and character description agreee, as they are supposed to. | 16:27.49 |
| But like I said, the two files I have here cause spacing problems if I do that. | 16:28.14 |
ray_laptop | kens: well, that's implied by substituting the .notdef glyph | 16:28.17 |
chrisl | It does say the behavior when resorting to the notdef is device dependent - so we are "correct", in a way! | 16:28.22 |
kens | I like using the /.notdef width better. | 16:28.35 |
| If I could only figure out the behaviour I'm seeing with these PDF files. | 16:28.51 |
| I suspect they are actually incorrect. | 16:29.00 |
ray_laptop | kens: where did the two PDF's in the test suite come from ? | 16:29.10 |
kens | I suspect they have a Width of 0 in teh Widths array, even though they are using the /.notdef, which has an actual width. But I will have to dig more to find out. | 16:29.38 |
| Weirdly, if I remove enough text, the problem goes away..... | 16:29.53 |
ray_laptop | kens: if they have a Widths array, that's supposed to supersede the glyph default width, right ? | 16:30.32 |
kens | Yes. | 16:30.37 |
| But.... | 16:30.43 |
| If we substitute with a /.notde, which width shoudl we use ? The one from teh glyph (which is missing) or the one from teh substitute :-) | 16:31.07 |
| But thsi is speculation anyway | 16:31.17 |
| I need to do more debugging | 16:31.25 |
ray_laptop | kens: OK. But I think if we substitute .notdef for a glyph we should still use the Widths entry corresponding to the original glyph. | 16:33.05 |
kens | ray_laptop : when I find otu what's actually happening I'll figure out what Acrobat is doing too, then I'll know how to fix it. | 16:34.27 |
| If I ever figure it out | 16:34.35 |
ray_laptop | kens: right. | 16:34.39 |
marcosw | sorry I missed the meeting. ANything for me? | 16:35.18 |
ray_laptop | meeting ended a while ago marcosw, but you didn't miss anything much. | 16:35.18 |
kens | WOnderful. THe mail just in with attached PDF files, one of them is created by Adobe InDesign, and it has genuine errors according to GS | 16:35.23 |
ray_laptop | marcosw: and nobody mentioned you | 16:35.35 |
chrisl | marcosw: thanks for sorting out the AIX VLP thingy - it got me what I needed. | 16:36.06 |
marcosw | chrisl: okay, I'll let the machine expire. | 16:36.23 |
kens | Time to go, night all. | 16:36.37 |
ray_laptop | chrisl: I forgot to tell you -- I don't think I ever had the VLP id thingy -- marcosw is the only one that ever used it AFAIK | 16:37.08 |
marcosw | ray_laptop: I think you set up access for me, in 2007 :-) | 16:37.32 |
ray_laptop | it's really hard for me to remember to follow up on things after my laptop crashes. It usually takes about 20 minutes to reboot after a crash, and sometimes it takes two or three reboots before everything works again. | 16:38.30 |
chrisl | ray_laptop, marcosw: Hmm, I can ask IBM who our nominated representative is | 16:38.45 |
ray_laptop | one time it took 2 hours because it had to recover the file system | 16:38.54 |
marcosw | chrisl: will do | 16:39.29 |
ray_laptop | chrisl: IBM keeps calling me, so I think I am the 'official' rep. I do have a contact, but I'll go look at my 2007 email and see if I do have the code from them | 16:39.40 |
chrisl | ray_laptop: okay thanks - the thing we're after is the "company's unique token", apparently | 16:40.24 |
ray_laptop | chrisl: can I say the token here, or should I put it on a private chat ? | 16:41.49 |
chrisl | ray_laptop: private, i think | 16:42.07 |
ray_laptop | chrisl: see the private chat | 16:42.49 |
chrisl | ray_laptop: nope, gives an invalid error - it should be 16 characters, according to the site. | 16:43.52 |
Salil | ray_laptop: chrisl: -dDEBUG doesn't print anything in debug console. Do I need to specify different compile setting? I will try manual debug though. Thanks for all the help. | 16:44.48 |
ray_laptop | chrisl: let me try the "I forgot"... | 16:44.58 |
chrisl | ray_laptop: thanks. | 16:45.12 |
ray_laptop | Salil: -dDEBUG doesn't depend on a compile mode. | 16:45.21 |
chrisl | Salil: the output should go to stderr (I think) | 16:45.58 |
ray_laptop | Salil: if you have a debug build, you can breakpoint if gs_log_error in base/gsmisc.c then make it conditional on err==-15 | 16:46.35 |
marcosw | chrisl and ray_laptop: I just finished a chat with IBM. Ray is the IBM PartnerWorld contact and they are emailing him the token. | 16:50.40 |
chrisl | marcosw: cool, thanks | 16:51.03 |
ray_laptop | chrisl: marcosw: OK I've found my id and password and am signed in | 16:52.29 |
| but I still don't know what the 'token' thingy is ? | 16:52.51 |
| and when I sign in, I don't see anything pertaining to the VLP program. Just a bunch of hype AFAIK (other people selling software) | 16:56.45 |
| chrisl: OK. I have the email from IBM -- I'll forward it to you. Looks like you need to "register" first (pick you own username, password, use your email) then "associate" it with the company ID token that's in the email. Then the email me and I have to approve you to be added. What a run around :-( | 17:03.06 |
| been so long since I did it for marcos that I didn't remember :-( | 17:03.32 |
chrisl | ray_laptop: Thanks. I'm already registered, just didn't around to the next part before.... | 17:03.55 |
ray_laptop | chrisl: OK. you can skip the first part then and just go to the 'associate' part. | 17:04.56 |
| Go tohttp://www.ibm.com/PartnerWorld | 17:05.59 |
| From the right-hand navigator on the home page click on "join PartnerWorld" | 17:06.00 |
| Next page scroll down to the section titled "add yourself to your company membership" Click on the Link to Add yourself to a location profile | 17:06.02 |
| Enter your registration Userid/password (the one you already created) | 17:06.03 |
chrisl | I'm at the "Emlpoyee self association" page now..... | 17:06.40 |
ray_laptop | chrisl: see the private tab | 17:07.30 |
Salil | ray_laptop: It seems gs_log_error is not getting called at all. I can debug gs_call_interp in inter.c but not sure which objects to look at :( | 17:08.02 |
chrisl | ray_laptop: yeh, I'm past the token bit, now - filling in my remaining details (why didn't they ask this stuff on the original registration?!?) | 17:08.19 |
ray_laptop | chrisl: it's IBM they don't have to be user friendly ;-) | 17:08.50 |
| Salil: are you sure you have a debug build -- gs_log_error is not called from a release build | 17:09.38 |
chrisl | ray_laptop: so, apparently I've finished the employee association stage - we'll see what happens next. | 17:09.46 |
ray_laptop | chrisl: OK. Good luck. | 17:09.56 |
| marcosw: thanks for getting them to divulge the ID token. I cc'ed you and henrys as well. | 17:10.28 |
chrisl | Of course, once this process is complete I'll have no excuse to punt AIX issues back to marcosw :-( | 17:11.21 |
marcosw | ray_laptop: probably violated the IBM PartnerWorld TOS by revealing the token to non-authorized persons :-) | 17:11.24 |
chrisl | IBM's catch 22 equivalent? | 17:11.56 |
ray_laptop | two guys us dark suits, white shirts and skinny black ties just pulled up in front of my house. Oh oh -- they have pocket protectors with about 6 pens in that I see under their jackets | 17:13.24 |
chrisl | Well, it's been nice knowing you.... ;-) | 17:14.03 |
marcosw | chrisl: I don't think you want to admit to having known ray_laptop | 17:19.00 |
ray_laptop | chrisl: marcosw is right -- it is a global organization | 17:19.24 |
Salil | ray_laptop: I think not as I am doing only "make framework" Do you think "make framework debug" will create proper object files? | 17:20.09 |
chrisl | marcosw: who is this ray_laptop of which you speak? | 17:20.58 |
Salil | ray_laptop: otherwise I will just make normally without building framework but for arm7 | 17:21.08 |
chrisl | Salil: I don't think there was a target for making a debug framework, so you might just have to edit the compiler name or CFLAGS to add debug symbols | 17:22.53 |
ray_laptop | chrisl: actually debug symbols isn't as important as having /DDEBUG | 17:25.18 |
| DEBUG #defined | 17:25.37 |
chrisl | ray_laptop, Salil: yes, of course, for the debugging output - but the point remains: add the required flags "manually", as I don't see a debug framework target in there. | 17:28.03 |
chrisl | has to go..... | 17:28.38 |
henrys | marcosw:will you be processing the incoming mupdf issue? usually robin_watts_mac would have done it. | 17:30.03 |
marcosw | Yes, doing it now. | 17:30.18 |
henrys | great | 17:30.36 |
marcosw | I cut down the file and now mupdf is processing it correctly but Ghostscript still doesn't. Which is kind of surprising since neither can process the original file. | 17:30.57 |
|Frederik | 3/part | 17:32.35 |
mvrhel | finally got the gradients working with the devn high level color stuff for tiff sep | 17:57.32 |
| now on to patterns and transparency | 17:57.53 |
| so ray_laptop: in working on the gradient stuff I don't believe our code would have worked with more than 8 colorants in a shading | 17:58.49 |
| it will now with this fix | 17:58.59 |
ray_laptop | mvrhel: great -- and I'm not surprised that it never got noticed | 17:59.21 |
mvrhel | yes. I doubt we will run into such conditions | 17:59.41 |
Salil | ray_laptop: I did compile using -DDEBUG flag but still I don't get any log. Debug library is bigger than release so surely it has proper debug symbols. Its failing in interp function. Also interp doesn't seem to call gs_log_error but then I am just going through this code for first time. | 18:23.13 |
Jenkens | Hello Folks. I'm having a bit of an issue with GS 9.04 that I'm not having luck sorting out. | 18:35.37 |
| We run bioPDF which uses GS to convert the files from PS to PDF and when we have windows fonts like MS Sans Serif Adobe (and other readers/editors) identify this as TTXXXXXXXXXX (usually something random) | 18:36.21 |
| because of this my editors can't handle highlight/editing that text since it's not an identified font (though it looks fine).. | 18:36.41 |
| am I doing something wrong or missed something? | 18:37.21 |
mvrhel | ah my favorite. a crash in release but not debug build | 19:26.54 |
| valgrind seems to have found the issue for me | 19:49.44 |
henrys | Jenkins:any reason you don't work with biopdf on your problem? From what you've said it exactly clear what the problem is. Are you saying fonts are not detected as fonts in say Adobe Reader or Apple Preview. I guess I don't understand what "my editors" means exactly. | 20:41.56 |
| s/it exactly/it isn't exactly/ | 20:42.16 |
| tor8:ping | 20:44.30 |
Jenkens | henrys: I am working with them on it, but it seems to be GS related nearest they or I can tell. | 20:48.08 |
| "my editors" = adobe acrobat or nitropdf | 20:48.36 |
| they show as "embedded subsets" with weird names and markup tools like strikethrough and highlight do not work real well on them like they do something identified as say, Times-Roman. | 20:49.36 |
henrys | FWIW yes biopdf uses gs without payment, if they payed you could report a bug to them and it would be treated as a customer bug here and we would fix it promptly. | 20:50.06 |
| if it is a bug | 20:50.44 |
Jenkens | that's what I'm trying to figure out | 20:50.50 |
| I find it hard to believe nobody has printed something through ghostscript in MS San Serif and not experienced this | 20:51.05 |
henrys | post it up on bugs.ghostscript.com and the pdf font gurus will tell you what's going on. | 20:51.47 |
Salil | henrys: I compiled using DDEBUG flag but still there are no error logs using gs_error_log. It fails in interp in inerp.c with error code -15. As I compile the library through make rather than XCode I can't set breakpoints and debug normally. | 20:51.55 |
henrys | hmm, you are using the ghostpdl/xcode/GhostPDL.xcodeproj and creating the debug build, it worked for me some time ago but I haven't fooled with it recently. | 20:54.11 |
| run your program with -ZI if it doesn't produce voluminous output it isn't a debug build. | 20:55.29 |
Salil | I am not using ghostpdl/xcode/GhostPDL.xcodeproj. Let me check it out. Can it convert from PDF to PCL? or HTML to PCL? | 20:58.57 |
Jenkens | henrys: bug added.. i guess we'll see what's said | 20:59.24 |
henrys | Salil:yes it is just a ghostscript build for xcode. | 21:00.27 |
Salil | henrys: http://www.ghostscript.com/GhostPCL.html this is it right? | 21:01.19 |
henrys | Salil:what command line are you using with ghostscript - there's lots of different "kinds" of pcl. Usually it is different for each printer, that's why there are so many printer devices. | 21:01.45 |
| well part of the reason. | 21:02.01 |
| Salil that release should work or git to checkout the current code. | 21:02.41 |
| that should lead you to http://www.ghostscript.com/download/ where you'd get ghostpcl 9.05 | 21:03.30 |
Salil | henrys:great. I was using hpdj340 on command line which worked. I am trying to integrate with iOS. I got Ghostscript and compiled it. I wasn't aware of the GhostPDL | 21:03.35 |
| henrys: Yeah getting that right now. | 21:04.15 |
| henrys: that builds a command line tool or a library? | 21:04.58 |
henrys | command line tool AFAIK | 21:05.36 |
Salil | henrys: I need to use it as library but probably I can modify it to be used that way. Or integrate in my project for the functionality. | 21:06.14 |
henrys | there is a sample makefile target for building a dylib in unix-dll.mak but I don't think that is integrated into the xcode project. | 21:08.00 |
| make sodebug from the gs directory is what you want for that. | 21:09.04 |
| bbiab | 21:11.38 |
tor8 | henrys: sorry, something personal came up I had to take care of. I'm planning on pushing the mupdf 1.0 release candidate tomorrow (was hoping to do it today, but alas...) | 22:01.03 |
sebras | tor8: what to do about link_dests? | 22:01.34 |
tor8 | sebras: let's leave them as is for the release candidate (with a marker saying "due for change") | 22:02.03 |
sebras | tor8: right. is the marker there right now? | 22:02.21 |
| tor8: should I do the honours? | 22:02.31 |
tor8 | if not I will push it along with the version number bumping | 22:02.45 |
sebras | link_dests is still missing docs. | 22:04.19 |
henrys | np tor8 | 22:05.02 |
| so the docs are complete sebras? | 22:05.30 |
sebras | henrys: no, not fully. | 22:05.53 |
| PDF links are not documentet (albeit started on, hence the XXXes). | 22:06.11 |
| henrys: also the text extraction interface is not documented as far as I know. | 22:07.12 |
henrys | tor8:Robin_Watts would have picked up on this bug so you'll have to jump in http://bugs.ghostscript.com/show_bug.cgi?id=692949 | 22:08.06 |
sebras | henrys: but a large portion of the generic interface is, along with a quick intro and two examples of how to do single and multi-threaded execution. | 22:08.12 |
tor8 | henrys: I'll look into that one, but at a glance it would appear to be a freetype limitation | 22:09.21 |
| Forward 1 day (to 2012/03/28)>>> | |