| | <<<Back 1 day (to 2012/03/21) | 2012/03/22 |
| arthurf | tor, sebras: I just opened two bugs which are MuPDF related. 692941 might be worth taking a peek at before 1.0 goes live - I have it ranked as minor - but still it might help VS 2005 Team Edition developers. | 02:15.10 |
| | 692942 is a minor Ghostscript web link - need to revise a link to the third party libraries. | 02:16.30 |
| | Should also include Robin_Watts, but I know he is away. | 02:32.55 |
| mvrhel | oy: did you see my reply to you about the OutputIntent? | 06:51.13 |
| | Did that all make sense | 06:51.23 |
| kens | chrisl installing Acrobat X I decided to read the installation guide. First instruction 'restart your computer' ! Its bad enough having to restart all the tiem when installing software, now Adobe want me to do it before I begin..... | 08:06.24 |
| chrisl | kens: I decided not to bother after Michael said he'd phoned Adobe and demanded his money back for Acro X! | 08:11.53 |
| kens | chrisl I need something to validate and prduce PDF/A-2 files :-( | 08:12.13 |
| | Otyherwise I wouldn't have done it | 08:12.36 |
| | 294MB for a full installation. Selecting ful Asian Language Support in PDF viewing doesn't install full Asian language support for PDF creation, you have to do that separately.... | 08:13.19 |
| chrisl | kens: yeh, I really want to get rid of the cr*ppy Vista64 VM I have Acro 9 on, but if Acro X is even further feature hampered, I can't do that :-( | 08:14.08 |
| kens | I'll let you know. I'm more worried that the installation may have removed my Acrobat 9.... | 08:14.35 |
| | GAH! It didn't ask me which group to install into in the start menu, and now I can't find it! | 08:16.11 |
| | bet its installed the stupid 'helper' startup app as well, that'll be first against the wall. | 08:17.28 |
| | At least Acrobat 9 still works. | 08:17.55 |
| | Oh god, the reason I cna't find the 'preflight' is because its not visible by default. You have to go to the 'view' menu and actually add it. And the new task pane sucks. Uses up loads of screen real estate and adds nothign of benefit. Looks like some designer's bright idea. | 08:29.15 |
| | It does look like all the features of 9 are present. Text editing is still worse than 7 so I'll be keeping that ;-) | 08:33.50 |
| chrisl | I think I'll leave it a bit longer, and see how you get on over a little time...... | 08:35.53 |
| kens | Well I'd say there's no rush. If you need somethign checking on X I can always do it | 08:36.16 |
| kens | tries to remember how to produce an A-2 file. | 08:36.35 |
| | I only wrote the code a well ago, how hard can it be ? | 08:36.51 |
| chrisl | kens: the second Brother problem appears to have been another compression filter issue, and tkamppeter found what appears to have been a very similar issue testing an HP printer. | 08:56.46 |
| tkamppeter | chrisl, kens, I have issued cups-filters 1.0.8 now which takes off mono/color image, CCITT glyph, and font compression for all PS printers. I keep only page compression. I hope this gives a higher grade of compatibility. We must keep in mind that Poppler delivers a PostScript that works on all PS printers, so this should be possible with GS, too. | 09:00.39 |
| chrisl | tkamppeter: what Poppler does is of no concern to us: we produce valid Postscript Level 2, and any printer that advertises Postscript Level 2 or 3 compatibility that *doesn't* work with it is broken, and it should be the printer manufacturer that deals with the problem. | 09:02.45 |
| kens | chrisl tkamppeter I'[m willing to look at the compression filters, but as chrisl sys, these are bugs in the printers themselves, the files are not terribly complicated, and so the decompression *should* work. There are bound to be files which could be created by applications which woudl also fall foul of these problems | 09:05.19 |
| | Given that this is only for printing, and the printers are connected by high speed comms these days, the lack of compression is probably not vitally important. | 09:09.19 |
| chrisl | That's what I said to Till on this subject. The remaining issue the HP 4050 which is giving an invalidaccess error on filter..... | 09:11.41 |
| kens | That one is more worrying. | 09:11.57 |
| | No progress on it ? | 09:12.02 |
| chrisl | Not really, I got a bit confused yesterday, and put debugging stuff I meant for the Brother into the HP test job :-( | 09:12.34 |
| kens | I can understand that. | 09:12.52 |
| chrisl | I'm waiting to hear if I can use the back channel for debugging (it looks like using the "nc" command to send data to the printer also receives at least *some* back channel data, probably depending on the printer) | 09:13.41 |
| kens | That would help a lot if possible | 09:13.56 |
| chrisl | The only official thing I can find that should cause an invalidaccess on filter is trying to apply a decode filter to a writable file, or an encode filter to a readable file - neither should be relevant here | 09:15.07 |
| kens | Exactly, I'm struggling to undestand how an invalidaccess could occur at all. | 09:15.41 |
| | Not using SubFileDecode could be a lot harder to fix. | 09:15.51 |
| chrisl | I don't think we can without completely re-implementing ps2write. | 09:16.24 |
| | 0 | 09:16.26 |
| kens | Not really an option | 09:16.54 |
| | There might be a way round it, I'd have to think | 09:17.13 |
| chrisl | I think there would be a way round it for something like the image data, but I can't see another way to handle content streams..... | 09:18.20 |
| | Here's the full error: http://pastebin.com/62drhAft | 09:18.28 |
| kens | Might be able to do content streams a different way, like I said, it woudl take some thinking about, and I don't want to start without knowing what the real problem is. | 09:19.19 |
| | It *can't* be a general HP problem, I'm sure we'd have heard already by now. | 09:19.36 |
| chrisl | The other problem is, I can hack around with debugging code, and fiddling with the PS until it works, but that's not a guarantee that I've found the core of the problem - we could end up doing a hell of a lot work, only to find it still doesn't work...... | 09:20.57 |
| kens | I don't suppose nayone has taken it up with HP ? | 09:21.18 |
| chrisl | No, as Till indicated above, "this is Ghostscript's fault, because it used to 'work'......." :-( | 09:22.13 |
| kens | Yeah right.... | 09:22.24 |
| | Until they find the output from some other application doesn't work either. Oh well. | 09:22.39 |
| chrisl | I found a similar thread on Quark's forum about a font problem with the XPress 8.x PS ouput failing on certain HP printers: totally valid (if poor!) Postscript, works on plenty of other devices/RIPs, but it's a Quark problem, not HP - because earlier XPress versions worked fine. | 09:24.45 |
| kens | sigh... | 09:25.35 |
| chrisl | Do you know, under what circumstances we'd emit multiple "content streams" for each page? | 09:26.24 |
| | this is ps2write of course | 09:26.35 |
| kens | No I don't I've wondered that before. | 09:26.35 |
| | ps2write does the same as pdfwrite | 09:26.46 |
| | I have wondered if its a length thing | 09:27.01 |
| | I've never been bothered to look, because its all valid. | 09:27.18 |
| | If its important I can find out of course. | 09:27.49 |
| chrisl | Do we preserve Type 1 forms? | 09:27.49 |
| kens | No, defintiely not | 09:27.57 |
| | They are inserted multiple times | 09:28.10 |
| | There's an enhancement open to not do that. | 09:28.19 |
| chrisl | Same for forms...... | 09:28.36 |
| kens | Forms generally | 09:28.54 |
| chrisl | Okay, so Type 3 glyphs, and transparency groups *should* be the only time we emit multiple content streams | 09:29.05 |
| kens | ? Type 3 glyphs aren;'t in a content stream. | 09:29.35 |
| chrisl | The marking operations are, aren't they? | 09:29.48 |
| kens | But they are emitted as an aside and stored in a CharProc | 09:30.05 |
| | Transparency is flattened to an image as well. | 09:30.25 |
| chrisl | In ps2write, yes, I was thinking of the greater world of pdfwrite *and* ps2write | 09:31.01 |
| kens | In general they work the same. | 09:31.14 |
| | THough the transparency is of course different, yes. | 09:31.24 |
| | In pdfwrite that does end up in a separate content stream. Images can too. | 09:31.52 |
| | Thoguh that's not exactly a content stream | 09:32.13 |
| | Note that we *do* emit forms for transparency in pdfwrite (but obviously not ps2write) | 09:32.44 |
| chrisl | So the marking operations for the Type 3 font are in a stream/endstream, so they do use the subfiledecode method | 09:33.58 |
| kens | Yes, I was assuming that was the area of the problem | 09:34.28 |
| | And considering if it would be possible to do the type 3 CharProcs differently to avoid the SubFielDecode | 09:34.55 |
| | But without knowing waht the problem area is, I can't be sure. | 09:35.20 |
| chrisl | The HP problem doesn't have Type 3 font(s) in it...... | 09:35.29 |
| kens | Can't be thath then. I was going suggest we do a ps3write which preserves CIDFonts, but that just opens a new can of worms. | 09:36.01 |
| chrisl | Unless I can establish there is something "special" about the SubFiledecode call that's throwing the error, we'd be looking at eliminating *all* uses of subfiledecode in the ps2write output <shudders> | 09:37.14 |
| kens | Yeah that's going to be a problem. | 09:37.26 |
| | I just came across this comment in opdfread.ps: % currentfile exch () /SubFileDecode filter % id obj file | 09:39.08 |
| | % We would like to use the code above, | 09:39.08 |
| | % but HP LaserJet 1320 continues parsing after the byte count exceeds. | 09:39.08 |
| | pop currentfile 0 (endstream) /SubFileDecode filter % id obj file | 09:39.08 |
| | So it looks like there's already a known problem with HP devices and SubFileDecode | 09:39.36 |
| | Also: | 09:40.25 |
| | % HACK BEG | 09:40.25 |
| | % This hack provides a compatibility to HP LaserJet 1320, | 09:40.25 |
| | % which sometimes closes the underlying stream when EOD | 09:40.25 |
| | % is reached in the ASCII85Decode filter. | 09:40.25 |
| | Re-implementing the stream code without using SubFileDecode looks like it would be a lot of effort. | 09:41.07 |
| | Probably easier to write a new ps2write device. | 09:41.18 |
| chrisl | I guess we *might* be able do something like define "stream" as "{" and "endstream" as "} exec" | 09:41.35 |
| kens | Maybe. As I said I'd want to look at the offending PostScript and have a chance to think about it. | 09:42.03 |
| chrisl | Well, there are only four calls to filter with the parameters in the error, so even without back channel output available, it probably won't be hard to fathom which is causing the error. | 09:43.51 |
| kens | Fair enough | 09:44.05 |
| chrisl | However, that also means there's nothing really special going on with any of them, so...... :-( | 09:44.48 |
| kens | Not looking good then. WHat a pile of cr*p | 09:45.08 |
| chrisl | Given that I appear to have hundreds of customer jobs that do "currentfile .... /SubFileDecode filter", how the hell can HP have got away with shipping it broken.......? | 09:50.00 |
| kens | Ah, I was sure it must be possible to find 'in the wild' PostScript files that use SubFIleDecode | 09:50.43 |
| | It must be somethign more sublte. | 09:51.01 |
| | I wonder if the ASCIIDecode has 'closed the underlying strea' as per teh comments in opdfread above. That might give an 'invlaidaccess' error. | 09:51.38 |
| | You might need to look at what's actually going on in the stream. | 09:52.10 |
| chrisl | But the underlying stream is "currentfile", how do you close currentfile? | 09:53.44 |
| kens | Well I was wondering if the stream used a decode on the current (SubFileDecode) file, and that the new decode closed the underlying stream (the SubFileDecode) rather than closing currentfile | 09:54.43 |
| | That seems to be what the comment is implying | 09:54.58 |
| chrisl | I guess that would be possible. Once I get a reply about using the back channel, I'll narrow down the specific call that's causing the error, and take it from there | 09:55.57 |
| kens | Well its somethign to look out for. | 09:56.21 |
| | OK I *think* I've made a PDF/A-2 file.... | 09:57.14 |
| chrisl | Yay! ;-) | 09:57.30 |
| kens | Need to check it, and now Acrobat X wants me to register, for which I have to have an Adobe ID (to *register* ?) | 09:57.54 |
| | Oh good, it appears I do already have one, because I can't use my email address :-( | 09:58.52 |
| chrisl | kens: BTW, did you notice spamcop had an outage yesterday? | 09:59.20 |
| kens | And I used my standard throw-away password obvioulsy because it works :-) | 09:59.22 |
| | chrisl, certainly did! | 09:59.30 |
| | Email started arrriving late last night | 09:59.40 |
| chrisl | It was a DNS problem, and it looks like some mail might have gone missing (bounced), just so you're aware...... | 10:00.04 |
| kens | ooh, 'answer a few simple questions and get a complimentary gift from Adobe'.... | 10:00.08 |
| | chrisl Ah, thanks for that. Mostly it looks OK though. And if it bounced presumably people will at least get a bounce mail | 10:00.34 |
| chrisl | Hopefully, yes. TBH, the spamcop people aren't being very helpful - I'm afraid it seems to have really gone downhill in the last 18 months or so..... | 10:01.34 |
| kens | I was wondering if that would happen when IronPort took over. I may have to switch to using one of my ISP accounts instead. | 10:02.10 |
| | Well I can't see anything that Adobe are offering that I would actually want. So I don't think I'll bother with teh customer intrusion survey | 10:02.47 |
| | Well the preflight says I've managed to make a PDF/A-2 golfer OK. Now for somethgin a little more challenging. | 10:04.11 |
| | Hmm, problems with CIDSet. I thought they had changed that for PDF/A-2 :-( | 10:05.42 |
| chrisl | You mean they actually tell what it means, now? | 10:06.30 |
| kens | They spell it out clearly. | 10:06.43 |
| | Unlike the other documentation. | 10:06.53 |
| chrisl | Well, that's an improvement..... | 10:06.57 |
| kens | It looks like our current usage (compatible with PDF/A-1 validators) isn't correct for PDF/A-2 | 10:07.12 |
| | Not really a problem, but it tells me what to fix. | 10:07.26 |
| | On the plus side, the rest of it seems to be OK (so far) and that's tested a fair few things. | 10:07.51 |
| | I guess today's challenge is to figure out what's wrong with CIDSet production and fix it. | 10:08.19 |
| chrisl | I was thinking, if wasn't for google having this fetish for you only having "one account to rule them all", I might switch to gmail..... | 10:08.32 |
| kens | I think I need some coffee before I try to read that part of the spec and code again though. | 10:08.42 |
| | chrisl I'm steering away from GMail since it insists on tracking you everywhere you go. | 10:09.02 |
| | I only log in once a day to clear out spam. | 10:09.16 |
| | Fortunately for me, I've always pulled my mail off using POP3 anyway | 10:09.36 |
| chrisl | Do they track all accesses, or just webmail ones, do you know? | 10:10.17 |
| kens | I believe (but don't quote me) that they store a cookie or something while you are logged into GMail, and then all Google services track you | 10:10.55 |
| chrisl | Right, so not tracking POP3 or IMAP connections, then | 10:11.26 |
| kens | BSame is also true for Google Docs and so on | 10:11.29 |
| | Oh yea, using IMAP or POP3 (as I do) means you aren't logged in to Google services, so none fo them track you. | 10:11.54 |
| | THey only track you while you are logged in (allegedly) | 10:12.03 |
| | Anyway, coffee break. | 10:12.35 |
| chrisl | Still, as I said, google make it *really* unpleasant trying to hold two separate accounts, so it's not really an option..... | 10:12.36 |
| tor8 | chrisl: as you say, they made it really annoying to hold two separate accounts, so I wouldn't bother with gmail if you haven't already bought into it | 10:14.37 |
| | I've taken to running two separate browsers, one for work gmail and one for my private gmail. very annoying. | 10:15.01 |
| chrisl | tor8: yeh, I don't want to go down that route. I also had problems with google code and a couple of things, pain in the a*se! | 10:16.30 |
| tor8 | as for the tracking, I have a facebook, twitter and google tracking-blocker extension on my browser. it's scary how many tracking cookies that thing deletes! | 10:16.38 |
| kens | Oh I have a generla purpose tracking cookie blocker too :-) | 10:17.06 |
| tor8 | I use "disconnect" for chrome, but since I use chrome I'm sure google tracks everything I type anyway. | 10:17.48 |
| | not that it matters much, they already have my email. but not being tracked by facebook even when I don't have a facebook account feels nice and smug ;) | 10:18.15 |
| kens | I probabl;y says s in teh T&C's somewhere :-) | 10:18.22 |
| tor8 | privacy is a lost cause anyway. they just voted through the data retention laws here :( | 10:19.18 |
| kens | I'm sure GCHQ has been spying on me for decades :-) | 10:19.55 |
| paulgardiner | Can you remove a gmail account once created? I seemed to be forced to created one as part of setting up my Android phone. | 10:20.14 |
| tor8 | paulgardiner: set your age to < 13 and they'll automatically delete it for you ;) | 10:20.45 |
| paulgardiner | :-) | 10:20.58 |
| tor8 | thanks to COPPA or whatever that crazy "think of the children" law is called | 10:21.17 |
| |Frederik | https://accounts.google.com | 10:27.41 |
| ysangkok_ | hi. i was trying to crop a PDF and it successfully crops all pages but dies at the end with "Error: /undefined in /UniGB-UTF16-H" | 10:28.35 |
| | is there a way that i can make it ignore errors like that, maybe just skip the command/statement? | 10:29.06 |
| kens | No, because its a fatal error | 10:29.41 |
| | But you don't say what exactly you are doing | 10:30.00 |
| ysangkok_ | gs -o cropped.pdf -sDEVICE=pdfwrite -c "[/CropBox [24 72 559 794] /PAGES pdfmark" -f in.pdf | 10:32.41 |
| chrisl | ysangkok_: what OS are you using? | 10:33.06 |
| kens | Well it sounnds like your in.pdf file doesn't include a necessary CMap. | 10:33.06 |
| ysangkok_ | adobe acrobat reader alerts a warning about missing chinese support on that last page too | 10:33.08 |
| | this is debian stable | 10:33.20 |
| chrisl | OKay, have you installed the package that includes the CMaps files? I can't remember what it's called...... | 10:33.50 |
| kens | UniGC-UTF16-H doesn't look standard | 10:34.26 |
| ysangkok_ | i don't have rights to install packages... but i was wondering if it is possible to skip the last page? | 10:34.59 |
| chrisl | -dLastPage=X will stop at page X | 10:35.37 |
| kens | Use -dFirstPage and -dLastPage | 10:35.44 |
| ysangkok_ | okay thanks | 10:36.17 |
| | i dont think i have any cmaps packages installed | 10:36.24 |
| | cause they are in the apt cache but not in the dpkg | 10:36.35 |
| chrisl | I guess that CMap might be a typo, anyway - probably should be UniGB-UTF16-H | 10:36.39 |
| kens | Err yes | 10:36.59 |
| chrisl | Oh, it is that...... | 10:37.01 |
| | Well, that is standard in our CMap collection, then | 10:37.19 |
| kens | GS version ? | 10:37.57 |
| ysangkok_ | 8.71 :/ | 10:38.17 |
| kens | Ancient, try 9.05 | 10:38.28 |
| ysangkok_ | yeah this is debian stable after all :P | 10:38.38 |
| | but i'll try doing it at home with admin rights and stuff... the 300 mb disk quota probably wont let be build gs anyway | 10:39.10 |
| chrisl | ysangkok_: if you want, the package you need installed is something like: "gs-cjk-resource" | 10:39.11 |
| ysangkok_ | okay thanks a lot | 10:39.40 |
| chrisl | UniGB-UTF16-H was included in 8.71 | 10:40.17 |
| kens | Could be a bug though | 10:40.40 |
| | Or a distro thing | 10:40.52 |
| chrisl | Debian used to split the CMaps out because of the license confusion, and put it in their non-free repositories | 10:41.38 |
| | IIRC, the current testing, or unstable or whatever the heck it is now has that package in the normal repos, but still a separate package for some reason.... | 10:42.20 |
| kens | To save a couple of hundred Kb of disk :-) | 10:43.01 |
| ysangkok_ | :) anyway ill just try at home. | 10:43.27 |
| | you were very helpful. i never received replies so quickly before on IRC. have a nice day, bye | 10:43.29 |
| chrisl | ysangkok_: glad to help | 10:43.41 |
| | kens: like I say, originally it was the license issue, I think it's remained in a separate package so far just through inertia | 10:44.15 |
| kens | Well, not really our problem :-) | 10:44.28 |
| chrisl | At least with Debian, it's a fairly simple answer, and there was a reason behind the original decision - this crap with Gentoo dropping the GS CMaps in favour of the poppler ones, which are in a totally incompatible directory layout - and missing *our* CMap, that's just stupid :-( | 10:47.03 |
| robin_watts_mac | Namaste! | 10:47.34 |
| kens | Hi robin_watts_mac | 10:47.41 |
| | I guess you re in a hotel somewhere ? | 10:47.48 |
| robin_watts_mac | Yeah, in Kathmandu | 10:48.44 |
| chrisl | It's going to be interesting when a Gentoo user tries to substitute a TTF for a CIDFont, and we need our CMap to deal with the TTF...... | 10:48.54 |
| kens | chrisl well we cna send them back to the maintainer :-) | 10:49.07 |
| | 'Not our problem John' | 10:49.14 |
| chrisl | Yeh, that *always* works...... :-( | 10:49.26 |
| | robin_watts_mac: shouldn't you be sight-seeing or something?? | 10:49.42 |
| kens | Been I expect | 10:49.53 |
| robin_watts_mac | Only just arrived. Knackered. | 10:50.21 |
| chrisl | dreads to think what trashy magazines Helen would have read on the plane........ | 10:51.44 |
| robin_watts_mac | "Look" and "Now", I believe. | 11:05.07 |
| | and "Reveal", apparently. And yes, they were trashy. | 11:05.19 |
| kens | Isn't that a Children's televeision show ? :-) | 11:05.31 |
| robin_watts_mac | Same IQ level required, yes. | 11:05.43 |
| chrisl | That'll be "'plane cabin IQ level....." | 11:06.58 |
| robin_watts_mac | tor8: Malcs ARM pdf is my casper home dir (with a shorter name) I think. | 11:15.20 |
| tor8 | robin_watts_mac: arm.pdf ? | 11:16.01 |
| robin_watts_mac | yes. | 11:16.12 |
| tor8 | thanks! | 11:16.17 |
| robin_watts_mac | I had a thought about how to match ghostscripts fill adjust stuff with the faster scanline algorithm on the plane. | 11:16.55 |
| | Need to think it through for a bit to make sure it wasn't just my brain drying out at 35000 feet. | 11:17.25 |
| kens | Oh great. Acrobat Distiller X doesn't produce PDF/A-2 files. | 13:39.48 |
| | How sad is that | 13:39.59 |
| | And Acrobat prefligth won't convert the broken PDF/A-2 to a working PDF/A-2. | 13:42.48 |
| | Oh, and they've updated the prefligth checker so that it doens't like out CIDSet for PDF/A-1 files either, brilliant. | 13:44.05 |
| | Ah no, that may be me :-) | 13:44.21 |
| | chrisl ping | 13:58.28 |
| chrisl | sorry kens, parents were here | 14:17.48 |
| kens | No problem | 14:17.53 |
| | Is there a quick way to find out how many glyphs are in a TT font ? | 14:18.04 |
| | I'm running out of ways to permute hte CIDSet, I can't figure out what Acrobat is complaining about | 14:18.20 |
| chrisl | Well, I would use the Freetype tools, but I think one of the MS tools gives you that info. | 14:18.57 |
| kens | ttfdump ? | 14:19.08 |
| | It measn extracting the font but that's OK | 14:19.16 |
| chrisl | Also, I think we store the number of the glyphs in the font dictionary - something like NumGlyphs | 14:19.28 |
| kens | num_glyphs, but I want to check what's actually in teh font | 14:19.42 |
| chrisl | That *should* be what's actually in the font, I think | 14:20.06 |
| kens | Well then I'm stumped. | 14:20.16 |
| | Acrobat complains about all the CIDSet entries, and I dont; knwo why | 14:20.32 |
| chrisl | Did you manage to get Distiller X to emit a PDF/A-2 file/ | 14:21.08 |
| kens | I've tried the number of glyphs used (a la PDF/A-1), the actual number of glyphs i n the font, and 256 (because it uses a OneByte-Identity-H CMAP) | 14:21.13 |
| | chrisl, it won't. No way around it | 14:21.25 |
| | Nor will the Acrobat preflight convert my file to a PDF/A-2 file. | 14:21.48 |
| | Becasue the CIDSet is wrong.... | 14:22.01 |
| chrisl | But if the CIDset was right it would a valid A-2 file, and wouldn't need converting - doh! | 14:22.35 |
| kens | Exactly, pile of rubbish.... | 14:22.48 |
| tor8 | kens: mupdfextract and ttfdump any help? | 14:23.57 |
| kens | tor8 to be honest I'm clutching at straws. | 14:24.12 |
| chrisl | Well, the problem is, you can break in Freetype and find the number of glyphs in there, but that will have come from the GS font, which probably uses the same value as the PS dictionary. | 14:24.26 |
| kens | Well, I think its correct anyway, because it worked previously for PDF/A-1 | 14:24.48 |
| tor8 | kens: http://git.ghostscript.com/?p=user/tor/snippets.git;a=blob;f=ttfdump.c will dump the headers and list the number of glyphs | 14:24.55 |
| kens | THe PDF/A-2 spec is not much more enlightening about what it regards as the 'font program' | 14:25.09 |
| | Anyone got a copy of the ISO PDF spec ? | 14:25.18 |
| | THE PDF/A-2 spec references that. | 14:25.33 |
| tor8 | kens: pdfref32000-1.pdf in /home/tor on casper | 14:26.22 |
| chrisl | I thought the ISO version was just 1.7 with the ISO boiler plate..... | 14:26.23 |
| kens | Well it could be.... | 14:26.35 |
| tor8 | chrisl: the iso spec has renumbered all the chapters, IIRC | 14:26.43 |
| kens | I don't have a copy of the ISO version to check. | 14:26.44 |
| robin_watts_mac | paulgardiner has the ISO version I believe. | 14:27.32 |
| tor8 | http://www.adobe.com/content/dam/Adobe/en/devnet/acrobat/pdfs/PDF32000_2008.pdf | 14:28.17 |
| kens | What, you don't have to pay the exorbitant ISO prices to get a copy ? :-O | 14:28.45 |
| tor8 | it's probably just the latest draft as is usually the case with these things | 14:29.01 |
| kens | Might be good enough for my purposes | 14:29.13 |
| chrisl | tor8: so, basically, it looks like it's the 1.7 spec, with renumbered chapters, and a less "nice" layout..... | 14:29.16 |
| tor8 | "This document is an ISO approved copy of the ISO 32000-1 Standards document. By agreement with ISO, Adobe Systems Incorporated is allowed to offer this version of the ISO standard as a free PDF file on their own Web site. It is not an official ISO document but the technical content is identical; the page and section numbers are also preserved." | 14:29.40 |
| kens | Its got the table i need, and its no help at all. | 14:30.04 |
| | :-( | 14:30.09 |
| | It just says hte same as table 5.21 in teh 1.7 spec | 14:30.55 |
| chrisl | You could get Marcos to ask Adobe for clarification - not sure if that would be covered by our Acro support contract, though | 14:32.01 |
| kens | I would doubt it.... | 14:32.12 |
| | Off to Mr Google I suppose | 14:32.21 |
| chrisl | Does Adobe have a tool available for creating these files? | 14:32.32 |
| kens | Not that I can find. | 14:32.40 |
| chrisl | Er, bit order? | 14:38.08 |
| kens | I don't think so. | 14:38.20 |
| | It all looks correct to me. | 14:38.31 |
| chrisl | I'm just looking at other people's CIDSet bugs, and what the solutions were | 14:39.05 |
| kens | Me too. | 14:39.13 |
| | But they all seem to be PDF/A-1 and we produce valid CIDSet for PDF/A-1 | 14:39.28 |
| chrisl | Well, we produce a CIDSet which the A-1 validators accept..... | 14:39.58 |
| kens | Including Acroabt. | 14:40.06 |
| chrisl | Still, it does seem to be less than an exact science, at the moment | 14:40.32 |
| kens | Yeah, very much so. | 14:40.51 |
| | I think I'll have to give up until someone can explain to me what's wrong with what we produce, or show me an example to work backwards from | 14:41.15 |
| chrisl | Well, as far as I can tell, it should be a bit set for every glyph index with a valid glyph in the embedded font file | 14:43.15 |
| kens | Yes, and that's what I'm writing. | 14:47.31 |
| chrisl | Have you tried setting all the bits in the stream? | 14:48.17 |
| kens | All except the bits that are beyond the 'num_glphs' in the font, eys | 14:48.39 |
| | that is, the padding bits | 14:48.52 |
| chrisl | I just wondered if the pre-flight was getting confused by the slightly different handling of notdef in TTF | 14:49.38 |
| kens | Its possible, though if you recall I had to explicitly add the /.notdef (CID=GID=0) in teh CIDSet | 14:50.07 |
| | Becasue we never use the /.notdef, but it has to be present. | 14:50.28 |
| chrisl | Hmm, yeh. Oh well, as you say only so much you can do. | 14:51.32 |
| kens | I've tried all teh combination sthat maek sense, and all the ones I can think of, sensible or tno. | 14:51.57 |
| | not | 14:52.00 |
| | Just ran the same file to to PDF/A-1 and it validates OK, so they've definitely changed it. | 14:52.44 |
| chrisl | Eh? That makes no sense..... | 14:53.09 |
| kens | What can I say. If I create a PDF/A-1 vliad CIDSet in a PDF/A-2 file, it is flagged as invliad. | 14:53.40 |
| | Of course, teh PDF/A-1 validity never made sense if you recall. | 14:54.12 |
| chrisl | My feeling is they've tightened up the validator for A-2, but left the A-1 validator as it was..... | 14:54.18 |
| kens | That could be | 14:54.24 |
| | But the validation of A-1 never made nay sense. It was supposed to have a bit set for every glyph in the font, but in fact you must only set the bits for the glyphs which the PDF *uses* | 14:55.01 |
| | Which is not what the spec says | 14:55.08 |
| chrisl | You're not allowed subset fonts in PDF/A are you? | 14:56.10 |
| kens | Yes you are | 14:56.16 |
| chrisl | Oh, so the validator is probably assuming that if you subset, it will only contain the glyphs used in the job - | 14:57.14 |
| kens | But that's not what the spec says. | 14:57.35 |
| chrisl | No, like I say, probably an invalid assumption by whoever wrote the validation code | 14:58.01 |
| kens | Its says 'regardless of whether the PDF file uses these glyphs' or words to that effect | 14:58.08 |
| | Well I'm going to commit what I have, which sets every bit in CIDSet for all the glyphs in the font. I think that's the right thing to do. | 14:58.52 |
| | Interesting | 15:03.03 |
| | pdf-tools valdiator says its OK | 15:03.15 |
| | :-) | 15:03.19 |
| | Good enough for me | 15:03.43 |
| chrisl | Of course, it would have made sense for them to release the PDF/A-2 spec as a valid PDF/A-2 file....... but they didn't | 15:05.18 |
| kens | Of course not, that would be sensible. | 15:05.28 |
| | Morning Michael | 15:13.19 |
| mvrhel | Morning kens | 15:36.50 |
| henrys | kens:wow A-2 complete nice! | 15:48.33 |
| kens | Yes, but... | 15:48.48 |
| | see teh CIDSet caveats | 15:48.53 |
| henrys | kens:why can't marcosw report that to adobe? maybe they'll explain what's going on. | 15:53.14 |
| kens | Well I don't mind if someone wants to try that. | 15:53.35 |
| henrys | might as well see what they say right? | 15:55.47 |
| kens | Sure, but I bet the answer will be no more illuminating than what's in the spec :-) | 15:56.06 |
| henrys | well if marcosw sees this and feels ambitious he can go ahead... nice to scratch a project off the todo list. | 15:58.56 |
| kens | Yes, certainly is. | 15:59.10 |
| ManDay | Urgh, I knew there was something I didn't want to upgrade... | 16:00.47 |
| | Did you guys by any chance rectify the issue of Freetype? | 16:01.08 |
| kens | A patch has been sent upstream and adopted | 16:01.19 |
| ManDay | That's great, thank you! | 16:01.31 |
| chrisl | ManDay: are you using Gentoo? | 16:01.39 |
| ManDay | chrisl: Yes | 16:01.44 |
| chrisl | I think they've pulled in the patches, so freetype in the gentoo repos should work - let me see if I can find the bug.... | 16:02.13 |
| | ManDay: https://bugs.gentoo.org/show_bug.cgi?id=408513 | 16:02.44 |
| ManDay | chrisl: Yes, it works! Thanls | 16:10.15 |
| | *thanks | 16:10.19 |
| chrisl | ManDay: np | 16:10.36 |
| henrys | sorry, kicked the cable on henrysx6 - it's back. | 16:53.14 |
| Guest98390 | Anyone know how to get pdf2ps to output postscript? I've tried "-dDuplex" and "-dTumble" but it doesn't work | 16:55.33 |
| | oops, I meant to say output duplex in the postscript | 16:55.53 |
| kens | If this is Ghostscript, you cna't | 16:56.01 |
| Guest98390 | Anyone know how to get pdf2ps to output duplex postscript? I've tried "-dDuplex" and "-dTumble" but it doesn't work | 16:56.07 |
| kens | The outptu is device-independent | 16:56.11 |
| Guest98390 | ghost script doesn't support duplex? | 16:56.14 |
| kens | pswrite/ps2write output is device-independent | 16:56.51 |
| Guest98390 | that's what I was worried about | 16:57.24 |
| kens | You cna easily add a line of code to the output at the start to turn it on | 16:57.53 |
| Guest98390 | is there a way I can supply it with a PPD or add custom postscript ? | 16:57.54 |
| oy | mvrhel: have now understood, thanks | 16:58.41 |
| mvrhel | oy: ok good | 16:58.53 |
| oy | :-) | 16:59.02 |
| Guest98390 | kens, can I do that as a command line argument to pdf2ps, or do I have to modify the output file manually? | 16:59.31 |
| kens | You have to modify the output | 16:59.43 |
| oy | mvrhel: one added small sentence would helpful, the -sOutputICCProfile will turn a -dUsePDFX3Profile into a proofing profile | 17:02.20 |
| | or the like | 17:02.24 |
| mvrhel | oy: ok. I will get that clarified | 17:02.38 |
| oy | good | 17:02.48 |
| mvrhel | brb coffee | 17:05.53 |
| kens | Time to go, goodnight all | 17:10.37 |
| henrys | chrisl:Len's been quiet unless I missed something. | 17:14.33 |
| chrisl | henrys: yes, almost suspiciously so - I wanted to check with Ray whether they'd been bugging him by phone........ | 17:18.16 |
| ray_laptop | chrisl: I saw your comment. Len has not been calling me. | 17:37.12 |
| chrisl | ray_laptop: well, I guess that's good...... | 17:37.48 |
| ray_laptop | henrys: chrisl: AFAICT Len is "owning" the problem and attempting to find a way to debug on the target -- but the tools aren't very good. | 17:38.03 |
| chrisl | ray_laptop: there's really no reason why the tools are so bad - correctly setup, using gdb remotely ought to be just as functional as using it locally | 17:39.05 |
| ray_laptop | chrisl: BTW, Scott says that the other company you've been helping with the DroidSansFallback for CJK fonts is looking hopeful | 17:39.15 |
| chrisl | ray_laptop: Oh, that's good. I was surprised they were asking about DroidSansFallback, I thought they were looking for other substitutes - I need to revisit that RSN..... | 17:40.15 |
| ray_laptop | another printer company !! They're interested in PS and PDF | 17:40.19 |
| chrisl | Do we know what level of integration work they might need from us? | 17:40.57 |
| ray_laptop | chrisl: thay may be looking for other fonts -- Scott was just commenting on the fact that they were still interested | 17:41.00 |
| | chrisl: well, at least they don't need ASIC integration. AFAIK we would be delivering the raster | 17:42.04 |
| chrisl | ray_laptop: well, that would certainly make things easier | 17:42.38 |
| ray_laptop | but I haven't any details on the printer -- like does it have a hard disk and what OS | 17:42.44 |
| | but it's a high end printer (>$10K) so it may have a hard drive | 17:43.31 |
| | chrisl: the one complication is that it has a 3-level per pixel mode, so it'll need the 2-bit thresholding like I did for cust 532. But at least that is not new | 17:46.36 |
| chrisl | ray_laptop: as long as we can meet the performance requirements by post-process halftoning, it's not a big deal, I'd have thought. | 17:48.07 |
| ray_laptop | chrisl: and the memory footprint. The advantage to what I did for cust 532 is that the buffer is only 2-bit, not 8-bit per component. This is even more critical in a CMYK machine | 17:49.45 |
| chrisl | ray_laptop: I'd hope we weren't *too* memory constrained on a $10k printer! | 17:52.35 |
| henrys | ray_laptop, chrisl:I find it staggering gdb isn't working perfectly fine on the printer. | 17:52.38 |
| chrisl | henrys: I assume their running the gdb server on the printer, and connecting to it via network/serial whatever - I seem to remember that needs a bit of configuration to get it fully functional (it's along time since I looked at it) | 17:55.19 |
| | s/their/they're | 17:55.30 |
| ray_laptop | henrys: I agree, but they rely so much on the simulator that they haven't invested the time in the build/toolchain issues needed. Also sometimes the memory footprint in the target is an issue | 17:55.43 |
| | they use gdb on a serial port (from windows hyperterminal) | 17:56.13 |
| chrisl | ray_laptop: so they run gdb directly on the target, not server/client? | 17:57.04 |
| ray_laptop | and since it is a serial port, capturing gobs of data (like pdf14_mark_fill_rectangle or -ZA is a problem) | 17:57.13 |
| | chrisl: correct -- at least the time or two I watched them. | 17:57.44 |
| chrisl | ray_laptop: ah, that is not great - running gdb in client/server mode, you can run a stripped binary on the target, but still have full debugging through the client. | 17:59.12 |
| | I need to head off - g'night all..... | 18:01.59 |
| marcosw_ | henrys: great news, your http://git.ghostscript.com/?p=ghostpdl.git;a=commit;h=dcf401ee8172aaff5363b2fc86e166f73225e429 fixed http://bugs.ghostscript.com/show_bug.cgi?id=692816 | 18:25.02 |
| henrys | marcosw_:cool | 18:52.07 |
| mvrhel | hmm. working through the shading code to support the devn color type which is used so that we can handle more colors with tiffsep, and I do not see how our current shading code could have worked with more than 8 colorants | 18:52.24 |
| | this is likely the hairiest part of this whole project | 18:52.54 |
| | marcows_ it is likely going to be at least 2 more weeks before I get this wrapped up | 18:56.15 |
| | good grief there are so many macros in gx_fill_trapezoid_as_lc I can hardly follow how the colors are computed | 19:01.24 |
| | or passed along | 19:01.37 |
| | bbiaw | 19:28.43 |
| | Forward 1 day (to 2012/03/23)>>> | |
IRC Logs