| <<<Back 1 day (to 2011/12/19) | 2011/12/20 |
mvrhel_laptop | henrys: you there? | 02:35.52 |
henrys | hi mvrel_laptop | 02:39.05 |
mvrhel_laptop | hi. I might be late tomorrow to the meeting | 02:39.17 |
| we are going to do a white house tour early in the morning ( 9am EDT) | 02:39.37 |
henrys | okay I'm not expecting any great excitement holidays coming and all | 02:39.48 |
mvrhel_laptop | so I should be back in time, but it is possible that we dont | 02:39.54 |
| ok | 02:39.56 |
Robin_Watts | tor8: You got any strong opinions on 692753? | 11:31.42 |
tor8 | Robin_Watts: it comes up often enough | 11:32.19 |
| didn't someone also ask for a multi-tool binary like busybox? | 11:32.40 |
| that would certainly solve the namespace problems :) | 11:32.52 |
Robin_Watts | tor8: how does that help with the namespace? | 11:33.07 |
tor8 | there's only one binary instead of half a dozen | 11:33.18 |
Robin_Watts | with a multi-tool binary, you still need the same number of names though. | 11:33.25 |
| cos it uses the name under which it was invoked to determine what function to perform, right ? | 11:33.56 |
tor8 | hmm. right. the argv0 checks. I was thinking the first argument could be the tool name as well | 11:33.58 |
| like: mupdftool clean foo | 11:34.13 |
| I like short names, that's my only gripe against the bug report | 11:34.42 |
Robin_Watts | I think a systematic change to use 'mupdf' rather than pdf would be a reasonable thing to do, and doing it before 1.0 makes sense. | 11:35.04 |
| You can always link pdfdraw to mupdfdraw :) | 11:36.13 |
tor8 | yeah, I can set up symlinks to fix my habitual typing of pdfshow and pdfdraw | 11:38.23 |
Robin_Watts | http://picsel.com/news/picsel-uk-joins-the-getjar-gold-program-smartoffice-available-as-a-free-do/ | 13:40.52 |
| For android users. | 13:41.00 |
kens | Oh God. My latest failing file has a type 3 font whose BuildChar calls a CharProc which uses another (different) type 3 font..... | 13:48.12 |
kens | goes to fecth a coffee in disgust | 13:51.13 |
DrZimmerman | hello everyone | 14:16.01 |
| is there a way to change pdf page orientation? | 14:16.14 |
| i'm googling now since hours, tried compiling pdftk but this is somehow broken, soon i'm really getting desperate | 14:16.55 |
| can it be really so damn hard to just change page orientation of a pdf? | 14:17.12 |
Robin_Watts | DrZimmerman: Can you be more specific about what you're trying to do ? | 14:17.16 |
| You have a PDF file, and you want to change it's orientation? Just a single PDF file, or a whole set of them? Are you creating the PDF files from some particular piece of software? | 14:18.03 |
DrZimmerman | just change page orientation from landscape to portrait from a A4 pdf | 14:18.05 |
| i've created the pdf from scanned documents using a bash script | 14:18.24 |
Robin_Watts | And what's the goal? To get a PDF that you can give to other people that will automatically be displayed properly in their viewers? | 14:18.38 |
DrZimmerman | but the orientation is wrong | 14:18.39 |
| yes | 14:18.45 |
Robin_Watts | So what tool are you using to create the PDF file from scanned documents/ | 14:19.10 |
| ? | 14:19.12 |
| I don't know of an automated tool myself, but it's not exactly hard. Maybe one of the others here will know. | 14:20.06 |
| I'm being called for lunch. bbs, sorry. | 14:20.18 |
DrZimmerman | http://nopaste.info/66e82fd9b4.html | 14:20.31 |
chrisl | DrZimmerman: you could look at pdftk - (from the manpage) "Rotate the first PDF page to 90 degrees clockwise" : pdftk in.pdf cat 1E 2-end output out.pdf | 14:23.00 |
DrZimmerman | as i've said above, i tried to compile pdftk, but it's broken | 14:24.05 |
chrisl | Oh. sorry, I just got back. What platform are you on? | 14:24.35 |
DrZimmerman | gentoo linux | 14:24.46 |
| in some bug reports some ppl said "pdftk is known to have compilation problems" | 14:25.09 |
| usually i can work around such things but this time i didn't figured out what the problem is | 14:25.25 |
chrisl | Hmm, I thought pdftk was java..... | 14:25.35 |
DrZimmerman | yes | 14:25.39 |
| but it's compiled with gcj | 14:25.45 |
| gnu java compiler | 14:25.52 |
chrisl | Yep, I'd expect the same tools to be used on Debian/Ubuntu to compile it | 14:26.22 |
DrZimmerman | probably | 14:27.13 |
| i'll try it with an older version of pdftk, maybe this will compile | 14:27.31 |
chrisl | So, you could run it through Ghostscript using pdfwrite, and try using the -dAutoRotatePages=/All command line option | 14:28.04 |
DrZimmerman | ghostscript = ?! for me :-) | 14:28.28 |
chrisl | Pardon? | 14:28.43 |
DrZimmerman | i tried to help myself with the man page but it doesn't tell anything about all the endless options gs has | 14:28.48 |
| simply put, i don't really know how to use gs | 14:29.12 |
chrisl | okay, so you'll want something like gs -dAutoRotatePages=/All -sDEVICE=pdfwrite -o out.pdf in.pdf | 14:29.42 |
| Whether that works (or works reliably) I think will depend on pdfwrite's heuristic for page orientation work for your specific files | 14:30.53 |
DrZimmerman | didn't worked | 14:32.45 |
chrisl | Okay, what is the current orientation of the pages? | 14:34.06 |
DrZimmerman | landscape | 14:34.58 |
| should change to portrait | 14:35.02 |
chrisl | So, you could try: gs -c "<</Orientation 0>> setpagedevice" -sDEVICE=pdfwrite -o out.pdf in.pdf | 14:35.34 |
DrZimmerman | i've also tried it with gs -dAutoRotatePages=/None -c "<</Orientation 0>> setpagedevice" -sDEVICE=pdfwrite -o out.pdf in.pdf | 14:35.38 |
| lol | 14:35.40 |
| :-) | 14:35.43 |
| also didn't helped | 14:35.50 |
| maybe i should remove the autorotate | 14:36.08 |
| didn't change anything | 14:37.06 |
| neither with my, nor with your version | 14:37.18 |
chrisl | Oh, I can't remember if order is important, maybe: gs -sDEVICE=pdfwrite -o out.pdf -c "<</Orientation 0>> setpagedevice" -f in.pdf | 14:37.19 |
DrZimmerman | ok i'll try this one | 14:37.35 |
| didn't worked | 14:38.26 |
| damn, i can't believe this, how complicated can it be to turn a pdf counter clockwise 90degrees? | 14:39.27 |
chrisl | With the right tools, it's very easy | 14:39.56 |
DrZimmerman | i know | 14:40.02 |
| adobe acrobat, but i'm linux user | 14:40.16 |
chrisl | pdftk - it's a unix-centric toolkit | 14:40.32 |
DrZimmerman | when it would compiled | 14:40.44 |
| the older version is also broken | 14:40.51 |
| but it breaks at another point | 14:41.14 |
| gcjh says to use --help | 14:41.21 |
| inside the compilaten | 14:41.27 |
chrisl | Have you tried giving explicit page dimensions to tiff2pdf? | 14:41.32 |
DrZimmerman | inside the compilation and then it stops | 14:41.35 |
| yes i've tried with A4 | 14:42.05 |
| it makes an A4 pdf, but the images are still landscaped and cut | 14:42.20 |
chrisl | No, I mean the -w width -l length options | 14:42.38 |
DrZimmerman | hmm no | 14:43.02 |
| anyway i'm running out of time, i should've sent this pdf hours ago, i just send it like this, i don't care anymore | 14:43.29 |
chrisl | To be honest, it sounds to me like your tiff images are the wrong orientation - what about rotating the image before creating the PDF? | 14:43.48 |
DrZimmerman | i think i've found the problem, insice the scanning script there's a pnmrotate command... | 14:49.43 |
| removed it, trying it again | 14:49.48 |
| yay! it worked | 14:51.36 |
chrisl | So, there's why it was so difficult to rotate your PDF - it was right all the time! | 14:52.15 |
DrZimmerman | chrisl, thanks, without your advice to change the tiff orientation i never would've gone to the section in the scanning script where it makes all the pnm to tiff converting magic :-) | 14:53.01 |
| yes | 14:53.02 |
chrisl | np. pnm is lossless, so you could leave the scanning script alone, and rotate it back again in your script without losing quality | 14:54.48 |
DrZimmerman | well, i've just removed the pnmrotate command inside the script, problem solved :-) | 15:06.00 |
| thank you very much! | 15:06.08 |
Robin_Watts_ | DrZimmerman: Fab. | 15:26.37 |
henrys | the puzzles arrived I've made it through cast baroq, onto level 5 | 15:37.17 |
kens | congratulations :-) | 15:37.29 |
Robin_Watts_ | henrys: Which ones did you get? | 15:38.09 |
| baroq is the 'prettiest', I feel. | 15:38.33 |
henrys | elk, quartet and eqa | 15:38.53 |
| equa | 15:38.56 |
Robin_Watts_ | I have elk, chain, baroq and H+H. | 15:39.33 |
| and I believe I have enigma coming for Xmas. | 15:40.02 |
chrisl | Robin_Watts_: can you remember, when you did the lcms2 integration: has lcms2 removed the ability to use third party memory management? | 15:43.13 |
Robin_Watts_ | chrisl: No. | 15:43.21 |
chrisl | You can't remember? | 15:43.31 |
Robin_Watts_ | I can remember. | 15:43.38 |
| It has not removed that ability. | 15:43.44 |
| It just does it in a slightly odd way, if memory serves. | 15:44.00 |
| See gs/base/gsicc_lcm2.c presumably. | 15:44.39 |
| See gs/base/gsicc_lcms2.c presumably. | 15:44.47 |
chrisl | Okay, the details aren't important, I just noticed there was a compiler option missing from lcms2.mak compared to lcms.mak | 15:44.56 |
Robin_Watts_ | Yeah, there is this 'PlugIn' thing that lcms2 uses to redirect memory stuff. | 15:45.31 |
chrisl | Super thanks | 15:45.54 |
Robin_Watts_ | Well... now I'm confused. | 15:46.36 |
| There is a line in mupdf, that says: | 15:46.47 |
| state->chain->rp = state->chain->wp - state->cinfo.src->bytes_in_buffer; | 15:47.18 |
| Sometimes, in some cleanup conditions, that can cause a SEGV (when state->cinfo.src is NULL) | 15:47.46 |
| So I've prefix that line (and just that line) with if (state->cinfo.src) | 15:48.03 |
| and that changes the behaviour of the code when I memsqueeze it. | 15:48.24 |
henrys | mvrhel_laptop:you made it afterall. | 16:58.09 |
mvrhel_laptop | yes. just got back | 16:58.21 |
henrys | I've never toured the inside of the white house. How was it? | 16:58.46 |
mvrhel_laptop | It was nice. Especially this time of year with the christmas decorations | 16:59.08 |
| christmas trees all over the place | 16:59.19 |
| kids choirs singing | 16:59.31 |
| not too many people | 16:59.42 |
henrys | well there all looking for jobs ;-) | 17:00.06 |
| they're | 17:00.15 |
kens | Meeting time ? | 17:00.27 |
mvrhel_laptop | :) well there were plenty of people employed in the security business there | 17:00.39 |
henrys | yes | 17:00.40 |
| meeting time | 17:00.47 |
mvrhel_laptop | one guy was so loaded down with weapons I don't know how he could walk | 17:00.57 |
henrys | is tor8 about? | 17:01.07 |
| I didn't really have general stuff just things from marcos, ray and tor none of whom are here AFAICT | 17:02.11 |
| anybody else? | 17:02.27 |
kens | Nothing from me | 17:02.41 |
| Didn't you want to talk about support ? | 17:02.50 |
henrys | not without marcos | 17:03.00 |
kens | OK | 17:03.05 |
mvrhel_laptop | I don't have anything but I was wondering if marcos had a chance to run the planar stuff with the null device | 17:03.11 |
Robin_Watts_ | I have a nagging feeling that I thought of something to discuss, but it's vanished from my memory. | 17:03.12 |
henrys | chrisl:what is the qa status of the URW fonts? | 17:03.45 |
| do you know if marcosw went through them? | 17:04.13 |
chrisl | henrys: I don't know, leaving it to him, mostly | 17:04.33 |
henrys | for the logs I just notice that we aren't getting 24 hour turnaround sometimes and I wonder if we need a new scheme. | 17:05.22 |
chrisl | henrys: I am going to compile a list of the issues listed by t1lint to get back to URW with - a couple of them look potentially real | 17:05.25 |
henrys | alexcher do you have anything? | 17:05.50 |
alexcher | Do we have any 32-bit embedded customers? How critical is VM usage? | 17:06.07 |
henrys | alexcher:why? | 17:06.41 |
alexcher | I want to increase ref size on 32-bit systems. | 17:06.45 |
chrisl | alexcher: as we discussed as the meeting? | 17:07.23 |
Robin_Watts_ | This is to raise the 64K limit on strings ? | 17:07.41 |
henrys | I think we should assume we might have a 32 bit embedded customer - is it a dramatic hit? | 17:07.44 |
alexcher | From 8 to 12 bytes. | 17:08.09 |
henrys | an embedded customer will be rasterizing - so the ps or pdf interpreter should not be the bottleneck right? | 17:08.32 |
kens | Its 4 bytes per object, and there can be a *lot* of objects | 17:08.56 |
Robin_Watts_ | A stupid question from someone who doesn't know the code in question; is there scope for adding a 'long string' type? | 17:09.13 |
alexcher | Yes, to increase the max sizes of arrays, strings, and dictionaries. | 17:09.38 |
kens | Robin_Watts_ : Its the internal representation | 17:09.42 |
chrisl | Robin_Watts_: we'd still need the larger "size" field | 17:09.43 |
kens | What chrisl said | 17:09.53 |
Robin_Watts_ | Unless the new type implicitly means "64K + whatever length is signalled" | 17:10.06 |
| It limits us to 128K of course :) | 17:10.17 |
alexcher | Robin_Watts: a new type is fine but requires to change all operators to support it. | 17:10.27 |
Robin_Watts_ | indeed. | 17:10.33 |
kens | Remember ther is no limit in PDF now | 17:10.36 |
chrisl | There is already a "bytestring" thingy which can be larger, but it's very limited, as far as I can tell..... | 17:10.42 |
kens | alexcher, could you make the change and test the VM usage of a um,ber of jobs to see how bad it is ? | 17:11.40 |
tor8 | henrys: yes? | 17:12.01 |
alexcher | kens: there's no difference on 64-bit platforms. | 17:12.08 |
chrisl | To be honest, if there is a noticeable impact on the executable size, I bet we could recoup a chunk of it by removing all the hoops we currently go through to work around the limit...... | 17:12.08 |
henrys | alexcher:do you have some sense of what the worst case slowdown would be in a real world jobs - maybe this is not such a big problem. | 17:12.22 |
kens | alexcher yes but on 32-bit platforms I mean | 17:12.24 |
| I don't think there's a performance hit, but a VM usage hit | 17:12.43 |
| We will use more memory | 17:12.50 |
henrys | hi tor8 let's start the documentation projects I'll cross the t's with sebras today. | 17:13.16 |
alexcher | kens: yes, I'll check the performance difference on 32-bit Linux. | 17:13.17 |
tor8 | henrys: okay! | 17:13.33 |
chrisl | henrys: One other thing: it occurred to me that if we're doing a staff meeting at the end of Feb, we should probably discuss whether to do 9.05 at the beginning of the month, or leave until early March | 17:13.42 |
kens | alexcher I still think its worth makgin the change, but it would be nice to have an idea of how much it will cost us. | 17:13.58 |
henrys | we want to document mupdf as a pdf tool and as a general purpose graphics library. Are all the architectural issues settled, best to write this once. | 17:14.27 |
Robin_Watts_ | henrys: I had some thoughts on function level docs, but no clear idea of how to approach the high level docs. | 17:15.17 |
henrys | chrisl:well if we have something stable beginning of February we'll release else march and we're late. | 17:15.43 |
chrisl | henrys: okay | 17:16.13 |
Robin_Watts_ | Perhaps tor8/sebras/I should prepare docs for some small section of the code, and bounce that around for a while until we have something we are all happy with ? | 17:16.15 |
tor8 | Robin_Watts_: agreed. | 17:16.38 |
sebras | Robin_Watts_ tor8: and have the apps/* be well documented to illustrate how to use mupdf. that seems to be a common question. | 17:16.47 |
henrys | well we are paying sebras to do it to avoid incurring lost engineering time, as long as we keep that in mind that should be fine. | 17:17.17 |
chrisl | tor8: do you think this would sufficient pointer to the mupdf iOS port: http://www.ghostscript.com/index-tmp.html ? | 17:18.03 |
tor8 | sebras and I thought it would be a good idea to turn the tools into thoroughly well documented examples | 17:18.09 |
Robin_Watts_ | right, I'm not suggesting we get in his way, just we don't want to get to the end of the process, and then say "Ah, but you haven't mentioned which functions can throw exceptions and which don't" or something. | 17:18.10 |
henrys | the tricky part of the docs to me is how to communicate how mupdf/fitz can be used as a pdf graphics library, and I'm not quite convinced tor8 has everything in place for that in the code, might be wrong though. | 17:18.46 |
henrys | suspects writing about it will result in new ideas for the code. | 17:19.50 |
tor8 | henrys: yeah, this exercise will probably be good to shake out confusing and bad parts of the interface | 17:20.45 |
sebras | tor8: will this also mean that the headers will be split into internal/external parts? | 17:21.28 |
henrys | tor8:also I wanted to check how much code sebras has contributed. Miles feels heavy contributors should sign copyright agreements, I don't feel strongly about it but not one of the battles I'd like to pick with him. | 17:22.11 |
| sebras is that a problem? | 17:22.55 |
| didn't know you were here. | 17:23.07 |
sebras | henrys: if a copyright assignment is required I'm happy to sign one. tor8 has mentioned briefly before but we never got around to having me sign one. | 17:23.09 |
| henrys: I got alerted by tor8. :) | 17:23.22 |
tor8 | yeah... miles never managed to send the paperwork he wants filled in. | 17:23.36 |
henrys | okay I'll have Miles send something... | 17:23.54 |
| so shall we call the meeting done? | 17:26.07 |
alexcher | chrisl: It is a good time to switch to OpenJpeg as a default library. | 17:26.12 |
kens | henrys, thanks, I shall offsky then. | 17:26.48 |
| goodnight all | 17:26.58 |
chrisl | alexcher: it's next on my list of things to do - I meant to do it today, but I've been buried in autmake cra*p. Should get it done tomorrow | 17:27.04 |
alexcher | Does any customer use gs without PDF interpreter and other optional features? | 17:27.57 |
henrys | I don't know of any send mail to ray. | 17:29.13 |
tor8 | chrisl: yeah, that's fine. I should add a note on the front page of mupdf.com too. | 17:29.57 |
| chrisl: I only put it on the 'news' page so far | 17:30.10 |
chrisl | tor8: thanks. | 17:30.31 |
mvrhel_laptop | Robin_Watts: you still here? | 17:34.08 |
| oops Robin_Watts_ | 17:34.22 |
Robin_Watts_ | tor8: sebras: For function level documentation, I'd like every function prototype (in the header) to be preceeded by a description of each param, a statement saying whether the function can throw an exception or not. | 17:34.46 |
| mvrhel_laptop: I am. | 17:34.50 |
chrisl | tor8: I've reassigned the bug to you, you can close it when you've added the info to mupdf.com's front page (or before, as you see fit!) | 17:34.53 |
tor8 | chrisl: okay. | 17:35.11 |
Robin_Watts_ | Also, the descriptions should make it plain what happens to the ownership of buffers/references passed in to functions both normally, and in the case of the function throwing an exception. | 17:35.24 |
| Does that seem fair ? | 17:35.29 |
mvrhel_laptop | so I am still a bit confused how to add in a flag for 'icc' in the list of debug flags | 17:35.55 |
Robin_Watts_ | mvrhel_laptop: OK, let me pull up the projects. | 17:36.13 |
| -s | 17:36.18 |
| So... gdbflags.h | 17:36.40 |
mvrhel_laptop | yes I am there | 17:37.02 |
Robin_Watts_ | The 'C' flag is currently unused, so lets hook on that. | 17:37.06 |
mvrhel_laptop | oh | 17:37.14 |
sebras | Robin_Watts_: that seems reasonable. we'll just have to agree on a format. I'm not sure whether there is any desire to post-process the header files to generate html documentation or similar..? | 17:37.15 |
Robin_Watts_ | Replace "UNUSED('C')" | 17:37.25 |
| with: | 17:37.36 |
mvrhel_laptop | I was confused in that I thought there was someplace to define "icc" | 17:37.50 |
Robin_Watts_ | FLAG(icc, 'C', 0, "Frobnicate the icc blizzle") | 17:37.58 |
| and Robert's your fathers brother. | 17:38.18 |
| sebras: Well, that sounds like a nice idea to me - not crucial, but nice. | 17:38.41 |
mvrhel_laptop | had to look that expression up | 17:39.32 |
Robin_Watts_ | "Bob's your uncle" - that's not used in the US? Sorry. | 17:39.53 |
mvrhel_laptop | and in fact Robert is my father's brother | 17:39.54 |
sebras | Robin_Watts_: alright, I'll have a discussion with tor8 about what he wants me to do as well... | 17:39.58 |
Robin_Watts_ | Ha! | 17:39.59 |
mvrhel_laptop | ok. so then all the debug statements for ICC that I want to add I can use 'C' and I can set the precedence to have them turn on with the color stuff someplace | 17:41.17 |
Robin_Watts_ | mvrhel_laptop: You want C to be implied by one of the other flags? | 17:41.36 |
| which one ? | 17:41.38 |
| Or you want C to imply one of the others ? | 17:41.56 |
mvrhel_laptop | hmmm I thought there was a color one | 17:42.57 |
| ok. c is general color | 17:43.41 |
| not halftone | 17:43.44 |
Robin_Watts_ | Oh, well, fix that description then. | 17:43.55 |
mvrhel_laptop | I will | 17:44.00 |
Robin_Watts_ | You want 'c' to imply 'C'? | 17:44.04 |
mvrhel_laptop | hold on | 17:44.20 |
| we already have a c and a C in the code | 17:44.39 |
Robin_Watts_ | We do ? | 17:44.49 |
mvrhel_laptop | C seems to be more detailed color mapping information | 17:45.26 |
Robin_Watts_ | Rats. I'll have to fix that (or you can). | 17:45.42 |
mvrhel_laptop | yes. this is all messed up | 17:45.49 |
| I may rework c and C | 17:45.55 |
Robin_Watts_ | I made that table by looking at the docs and grepping as best I could. | 17:45.59 |
mvrhel_laptop | no I understand | 17:46.05 |
| but we are limited to single characters correct? | 17:46.18 |
Robin_Watts_ | Currently, but that's trivial to change. | 17:46.29 |
| (or rather, I can make it so you can have options where you only have a long option. | 17:46.56 |
| Instead of using if_debug('C') you can use if_debug(gs_debug_flag_icc) | 17:47.48 |
mvrhel_laptop | oh | 17:48.01 |
| and how would gs_debug_flag_icc be defined? | 17:48.15 |
Robin_Watts_ | FLAG(icc, ...) :) | 17:48.32 |
mvrhel_laptop | and where | 17:48.32 |
Robin_Watts_ | It's magic. | 17:48.41 |
| You just change the one file. | 17:48.55 |
mvrhel_laptop | so hold on | 17:49.01 |
| sorry I am a bit slow on this | 17:49.14 |
Robin_Watts_ | No worries. | 17:49.17 |
| Everything you've ever learned about how in C you should only ever include a header once? Forget that :) | 17:49.37 |
| The gdbflags.h header is included multiple times with different definitions for FLAG, so we build many tables out of just one definition. | 17:50.18 |
mvrhel_laptop | so where/how is gs_debug_flag_icc defined in the above case then and how is that related to the icc in FLAG(icc, ...) and what would the other parameters be for this | 17:51.01 |
| I am missing something | 17:51.28 |
Robin_Watts_ | If you put a line FLAG(fred, ....) you'll get gs_debug_flag_fred defined. | 17:51.36 |
mvrhel_laptop | oh | 17:51.42 |
| and so I could have FLAG(icc, 0, 0, "ICC stuff") | 17:52.15 |
Robin_Watts_ | No. | 17:52.26 |
mvrhel_laptop | and the use gs_debug_flag-cc | 17:52.28 |
| _icc | 17:52.30 |
| hehe | 17:52.33 |
| ok | 17:52.35 |
Robin_Watts_ | FLAG(icc, 1, 0, "ICCstuff") | 17:52.51 |
| would be OK, if you replaced UNUSED(1) with that. | 17:52.58 |
| gs_debug_flag_icc would be defined to be 1. | 17:53.17 |
| 0 is special. | 17:53.20 |
mvrhel_laptop | ok. I see | 17:53.39 |
Robin_Watts_ | (so effectively you would have a short flag, it's just one you can't easily use) | 17:53.52 |
mvrhel_laptop | and to make gs_debug_flag_icc work when someone specifies 'c' what do I do? | 17:54.50 |
Robin_Watts_ | so you want gs_debug_flag_icc to be implied by 'c'. | 17:55.10 |
| So: FLAG(icc, 1, 'c', "ICC stuff") | 17:55.22 |
mvrhel_laptop | ok | 17:55.29 |
| simple enough. even a child could do it | 17:55.38 |
| not sure what that says about me.... | 17:56.04 |
Robin_Watts_ | I should have documented it in the header a bit better :( | 17:58.13 |
| Also, you're hampered by your hard won experience saying that every table can only be used once, I suspect. | 17:59.06 |
mvrhel_laptop | yes. that is still confusing to me | 17:59.37 |
| do we still only have 127 entries though? | 18:00.05 |
Robin_Watts_ | Currently, yes. | 18:00.19 |
| but that's easy to increase. | 18:00.25 |
| I just change one #define and add some more 'UNUSED' entries. | 18:00.40 |
| There were enough spare entries that I didn't bother. | 18:00.50 |
| If you want to see how the sausages are made, see gdebug.h line 60 | 18:01.31 |
| and gsmisc.c lines 1079 and 1088 | 18:01.58 |
| Well, txtwrite seems to like SEGVing for me :( | 20:06.57 |
| A vanilla language_switch build on casper (with the txtwrite device added) SEGVs for me. | 20:13.26 |
henrys | Robin_Watts:I know that pdfwrite has problems on the language switch also. language switch is just not in good shape. | 20:27.10 |
Robin_Watts_ | http://ghostscript.com/txtwrite.html | 20:35.28 |
| That's using gs rather than language_switch | 20:36.26 |
| Rays message to support about overprinting... I thought overprinting *wasn't* supported when color management was in use? | 20:39.19 |
| Or is that customer using an old enough version of gs that they don't have color management ? | 20:39.35 |
henrys | they are pre color management | 20:42.23 |
Robin_Watts_ | ok, that makes sense then, thanks. | 20:42.37 |
henrys | so the raw mode is supposed to have these embedded nulls? | 20:42.54 |
| I guess a question for kens | 20:43.09 |
Robin_Watts_ | Unless my perl wrapper is screwed up... | 20:43.22 |
henrys | with my test each character is prepended with 0x00 | 20:45.55 |
Robin_Watts_ | Do you get the same thing when calling txtwrite direct ? | 20:47.11 |
| I get the nulls with my test too. | 20:48.02 |
henrys | I'll check - also for raw mode it returned txtwrite.cgi to my browser do we want a different extension? | 20:48.12 |
Robin_Watts_ | Don't understand that. Let me fiddle. | 20:48.41 |
| Right. I get the nulls in raw mode calling the device direct. | 20:49.04 |
henrys | otherwise the page looks good. | 20:49.26 |
Robin_Watts_ | I had it showing the page in the browser earlier - that's what I was shooting for. what have I broken? | 20:53.33 |
henrys | well for utf-8 or unicode it does show in the browser but for raw I download a file called txtwrite.cgi - presumable the browser doesn't know what to do with the nulls. | 20:55.12 |
Robin_Watts_ | Maybe that's it. | 20:55.21 |
| I'm being called for dinner. ttyl. | 20:55.33 |
henrys | okay | 20:55.42 |
kens | For the logs; in txtwrite raw really does mean raw, so if you have an n-byte CID where teh upper bytes are 0, then you will get embedded NULLs. | 21:44.46 |
| Without seeing the file I can't be sure if that's the case, tehre could always be a bug. | 21:45.00 |
| But normally I would expect the default mode of UTF-8 to be in use. | 21:45.17 |
henrys | kens:I converted doc/ghostpdl.pdf from the tree - I didn't look at the pdf. | 23:15.40 |
mvrhel_laptop | henrys: the CRD cache is no longer created now. Next I am going to work on getting the proof profile hooked up and working for the one customer. This should not be too difficult. It is going to get stuck in the device like the other profiles and basically be used anytime that we link up with the device profile | 23:30.18 |
henrys | great news that should be a lot less -ZC output to wade through also. | 23:31.23 |
| kens:did a quick look I just see type 1 fonts in ghostpdl.doc | 23:41.13 |
mvrhel_laptop | henrys: yes. I also need to do the fix to get the icc stuff added in with debug statements | 23:45.07 |
| Forward 1 day (to 2011/12/21)>>> | |