| | <<<Back 1 day (to 2012/03/04) | 2012/03/05 |
| rico29 | hi all | 09:19.29 |
| | I have a problem with pdf transparency. I've read a lot of things on it on bugtracker, but I can't find if ghostscript can handle pdf with transparency | 09:20.13 |
| kens | Yes it can | 09:20.20 |
| rico29 | I need to convert a PDF with transparent thing in it, to a tiff file | 09:20.39 |
| | I'm pastebining the error | 09:20.51 |
| | http://pastebin.com/sdkgHziJ | 09:21.32 |
| | if anyone can take a look at it... would be very helpfull | 09:21.47 |
| kens | That's an archaic version of GS, current version is 9.05, first thing to try is a new version | 09:22.03 |
| rico29 | ok | 09:22.11 |
| kens | If it still fails, then open a bug report at http://bugs.ghostscript.com | 09:22.30 |
| | Don't forget to attach the PDF file so we can look at it ;-) | 09:22.41 |
| | The problem seems to be a font, and font handling changed very significantly with the 9.0 release. | 09:23.11 |
| | BTW you aren't really running GS as root are you ? | 09:23.42 |
| rico29 | kens: I'm running it as root for tests | 09:24.01 |
| kens | Well fair enough, its your machine :-) | 09:24.16 |
| rico29 | kens: I'll update it to last stable version. For now, the last centos repo version was ok to me | 09:25.40 |
| kens | OK that's good, glad it works | 09:29.48 |
| | chrisl what's the configure option to GS to not build in X ? | 09:30.02 |
| | And no X devices of course. | 09:30.13 |
| chrisl | Erm, I think it's --without-x | 09:32.36 |
| kens | Yes I think so, Mr Google agrees | 09:32.45 |
| chrisl | Why? | 09:33.27 |
| kens | SOmeone on stack overflow complaining they don't want to have to install X-Windows to use GS | 09:33.46 |
| | (actually their question is even less sensible) | 09:34.03 |
| chrisl | If X isn't installed (and the X development files aren't installed) configure should spot that, and disable the X devices automatically | 09:34.22 |
| kens | It probably does. | 09:34.44 |
| | There's no evidence they've actually tried to build GS. | 09:34.55 |
| rico29 | kens: looks better with last version of GS | 09:41.27 |
| | I don't get the same error as sooner | 09:41.41 |
| | I don't get any error at all in fact | 09:41.49 |
| kens | Sounds like it was a bug in that old version then | 09:43.11 |
| rico29 | ok | 09:43.49 |
| | I just get a thing that I didn't get before | 09:44.00 |
| | Now I have many tiff files : my file?tiff which is my PDF converted to tiff, but I also get a file.tiff.black?tif, file.tiff.Cyan.tif,file.tiff.Magenta.tif and file.tiff.Yellow.tif | 09:44.54 |
| kens | That's the tiffsep device | 09:45.15 |
| | try tiff24nc or tiff32nc | 09:45.21 |
| rico29 | I'm using : -sDEVICE=tiffg4 | 09:45.42 |
| kens | Then that should not produce any separated files. | 09:45.53 |
| | SouDelete the tiffs and try again I woudl say | 09:46.04 |
| rico29 | ok | 09:47.05 |
| | works better | 09:47.08 |
| kens | :-) | 09:47.27 |
| rico29 | but the output tiff file si very heavy (approx. 4x the size of PDF) | 09:47.42 |
| | with 24ncv | 09:47.47 |
| kens | ShG4 is monochrome, so should be smaller than colour. Check teh resolution you are using | 09:48.09 |
| rico29 | -sDEVICE=ShG4 ? | 09:48.48 |
| kens | SOrry, was a typo | 09:49.01 |
| | tiffg4 should be fine | 09:49.09 |
| rico29 | that's the one I was using sooner | 09:49.34 |
| kens | I suggest you try it again | 09:49.45 |
| rico29 | YEAH ! | 09:50.00 |
| | work great | 09:50.02 |
| | thanks a lot | 09:50.05 |
| kens | NP. | 09:50.09 |
| | tor8 is up early | 10:01.50 |
| tor8 | good morning to you too kens :) | 10:02.37 |
| kens | :0-) Morning tor8 | 10:06.40 |
| rico29 | has anyone already install the last version of image magick ? | 10:42.33 |
| kens | personally no | 10:42.57 |
| rico29 | ok | 10:49.12 |
| chrisl | rico29: I doubt you'll find any of the regulars here enthusiastic enough to install an ImageMagick other than what comes as standard with their Linux distribution. | 10:50.49 |
| rico29 | ok | 10:56.41 |
| Robin_Watts | tor8: ping | 11:08.17 |
| tor8 | morning robin | 11:08.35 |
| Robin_Watts | Morning | 11:08.39 |
| | Did you have a chance to look at my "split" patch ? | 11:08.51 |
| tor8 | (does your irc ping on "robin" as well as "Robin_Watts"?) | 11:08.53 |
| Robin_Watts | It does, I think. | 11:09.02 |
| | Yeah, I have it set to ping on "Robin", "Watts" and "Arm". | 11:10.02 |
| tor8 | cool, then I can be lazy with my shifts and underscores in the future :) | 11:10.23 |
| Robin_Watts | tor8: TAB :) | 11:10.37 |
| tor8 | mine pings on "tor" as well | 11:10.41 |
| | Robin_Watts: oh! | 11:10.48 |
| | new feature I didn't know of! | 11:10.58 |
| | I blame my touch typing habits :) | 11:11.10 |
| Robin_Watts | Today I plan to read up on forms ready for 5pm/ | 11:11.56 |
| tor8 | Robin_Watts: in general, but there are a few things I want to keep in the public header | 11:12.00 |
| Robin_Watts | tor8: Such as? Do you want you list them on here, or do an email and I'll pull 'em across ? | 11:12.34 |
| tor8 | I'm not decided on the fz_buffer and fz_stream reading functions | 11:12.54 |
| Robin_Watts | I feel that if we can hide them, we should. | 11:13.37 |
| tor8 | but I think it's better to err on the side of caution | 11:13.48 |
| Robin_Watts | We can always publish them later if we need to, but we can never make them private again. | 11:14.19 |
| tor8 | fz_find_device_colorspace is a sumatra function, to access the globals from the DLL barrier | 11:14.29 |
| | yeah, it's just a bit messy with git moving things across files (well, git blame that is) | 11:14.48 |
| Robin_Watts | And it's not really a hardship for people to have to include "fitz-internal.h" to get the unpublished ones. | 11:14.50 |
| tor8 | and merging diffs | 11:14.51 |
| Robin_Watts | tor8: Yes. | 11:14.59 |
| tor8 | my hope was that people in general shouldn't have to | 11:16.48 |
| | so how do the pdf apps get access to pixmap->samples in order to write out and draw on the screen? | 11:17.26 |
| Robin_Watts | tor8: Presumably they will be using fz_new_pixmap_with_data ? | 11:23.32 |
| tor8 | that's also internal :) | 11:23.49 |
| Robin_Watts | Ah, ok. | 11:23.59 |
| | I'd prefer an fz_pixmap_get_samples(pix) accessor function. | 11:24.31 |
| tor8 | the public set in mupdf.h is a nice set though | 11:25.09 |
| Robin_Watts | I'm just thinking that in future we may want to update the pixmap stuff to be able to cope with pixmaps with no alpha plane. | 11:26.20 |
| tor8 | still a few that should be shuffled around, pdf_load_outline should be public | 11:26.21 |
| Robin_Watts | tor8: Maybe the best way to proceed here is for you to pull my split patch and hack it about. | 11:27.00 |
| | Looking at mupdf.h, I dislike having pdf_xref_entry_s defined in public. | 11:27.48 |
| | I wonder why that was required. | 11:27.52 |
| tor8 | yeah. sebras is visiting me tonight so we can bang on it a bit then, these sorts of things go better with two opinions battling over an editor :) | 11:27.57 |
| | Robin_Watts: I wondered that too! | 11:28.06 |
| Robin_Watts | That sounds perfect. | 11:28.06 |
| tor8 | I think you may have put it there for pdfclean? | 11:28.30 |
| Robin_Watts | Yes, but then pdfclean reverted to using mudpf-internal.h | 11:28.49 |
| | because there is just TOO much there. | 11:29.07 |
| | brb. | 11:29.10 |
| tor8 | I think it should be possible to write pdfshow without using the internal mupdf header | 11:29.40 |
| | pdfclean I'm okay with needing internals for all the garbage collect and renumbering | 11:30.01 |
| Robin_Watts | I suspect we should maybe extract the gc etc into the library. | 11:32.14 |
| | If saving out is going to be important, then we may need that anyway. | 11:32.40 |
| tor8 | maybe the mupdf-internal.h that's included in a lot of these pdfapp tools may have hidden a few functions from making it into the public headers | 11:32.41 |
| | but as I said, I will bang on it with sebras tonight | 11:32.51 |
| Robin_Watts | Entirely possible. | 11:32.54 |
| | bool. | 11:32.56 |
| | cool. | 11:32.58 |
| | So, what jobs remain to be done before 1.0 ? | 11:33.05 |
| tor8 | did you manage to get the mt-example.c working? | 11:33.18 |
| Robin_Watts | I haven't looked. I will do so. | 11:33.29 |
| tor8 | I'm still not done with the text struct update :( | 11:33.31 |
| | procrastination is so much easier... | 11:33.42 |
| | I was thinking of dropping the fz_bbox struct altogether in favor of fz_rect | 11:34.04 |
| Robin_Watts | So: 1) Verify mt-example.c now works - if not, fix it. 2) text struct update. | 11:34.19 |
| | tor8: bbox is int ? | 11:34.26 |
| | I think having an int type is important. | 11:34.45 |
| tor8 | yeah. we may want to move it internally to dev_draw but hide it from public view | 11:34.57 |
| Robin_Watts | Passing rects in to (say) pixmap functions is going to be a bad thing. | 11:35.09 |
| | Thinking of banded rendering (or partial updates), then a bbox is what we want. | 11:35.22 |
| | From an external API point of view, bbox seems more important than rect to me. | 11:36.02 |
| tor8 | floats can contain integer values of the ranges we want | 11:36.22 |
| Robin_Watts | (When we do form filling, for example, the app wants to know the integer pixels it should cover) | 11:36.40 |
| | Yes, but rounding is important. | 11:36.57 |
| | So, I'd rather we rounded and passed ints out rather than relying on the app to get the rounding right. | 11:37.17 |
| tor8 | the dance we do every time we use rects with fz_round_rect(fz_transform_rect()) is ugly | 11:37.52 |
| Robin_Watts | I like the idea that fz_bboxes represent 'output space' coords in general, and fz_rects represent 'document space' ones (at least as far as any seen outside the API). | 11:38.03 |
| | Yes, but it's ugliness that's hidden within the API, rather than ugliness forced out onto the app writer. | 11:38.28 |
| tor8 | not strictly true. see mudraw.c where we calculate the bbox to draw into. | 11:39.31 |
| | we measure the page (which returns a fz_rect in document space), transform it, then round to a fz_bbox | 11:39.54 |
| | but that explanation (fz_rects are in document space, fz_bboxes are in integer device space) could go a long way | 11:40.19 |
| Robin_Watts | So, what bit of what I said is that not consistent with ? | 11:40.28 |
| tor8 | the bbox device should take a fz_rect though | 11:40.31 |
| | "hidden within the API" is what I said was false | 11:40.46 |
| Robin_Watts | Ah, right. | 11:40.53 |
| | 'mostly hidden within the api' :) | 11:41.13 |
| tor8 | important qualifier :) | 11:41.58 |
| Robin_Watts | mostly harmless. | 11:42.08 |
| tor8 | but to make it consistent, the bbox device (which calculates in document space) should take rects instead | 11:42.17 |
| Robin_Watts | that seems reasonable to me (without too much thought) | 11:42.41 |
| | OK, so the jobs I have before release are: 1) Verify mt-example.c, 2) text structs, 3) finish split. | 11:45.37 |
| | where 3) may fork more tidying tasks. | 11:45.53 |
| | 1 is with me, 2 with you, 3 with you/sebras (for now at least, but might get kicked back to me later) | 11:46.34 |
| tor8 | link destinations could do with another pass; I think they're in the wrong document space at the moment (not rotated by page /Rotate, shifted by the mediabox offset, and flipped to be y-is-down) | 11:49.55 |
| | nothing crucial, compared to 1, 2, and 3 and sorting out remaining crashes and bugs | 11:50.26 |
| Robin_Watts | 4) link destinations then. | 11:51.01 |
| | Are there known 'remaining crashes and bugs' ? | 11:51.23 |
| | or is that just a catch all for the inevitable ? | 11:51.40 |
| tor8 | bugzilla knows best, I'm worried about the growing number on there that I haven't been paying attention to lately | 11:51.57 |
| kens | Hehe, welcome to my world.... | 11:52.18 |
| ManDay | Hello guys, *is* PDF postscript? | 12:31.25 |
| | man groff_tmac says the .PSPIC macro can include postscript images in the document, I've got a PDF here which I want to include as a picture, do I need to convert it to .ps? | 12:32.14 |
| | Can gs(1) do that? | 12:32.27 |
| sebras | ManDay: PDF is not the same as postscript, but I don't know gs well enought to help you with a suitable command-line. pdf2ps may be helpful though. | 12:32.49 |
| Robin_Watts | ManDay: PDF is not postscript, but it is related. gs can convert PDF to postscript for you. | 12:35.33 |
| | gs -sDEVICE=ps2write -o out.ps in.pdf | 12:35.44 |
| | There may be options etc that you want, but the above command should have a stab. | 12:36.01 |
| sebras | Robin_Watts tor8: once we have settled the split I'll take another look at the documentation. | 12:40.39 |
| Robin_Watts | oh, yes, 5) Docs. | 12:40.56 |
| ManDay | Thanks Robin_Watts :) | 12:44.27 |
| | Isn't pdf a wrapper arround eps which is a wrapper arround ps? | 12:44.43 |
| | forgive my ignorance :P | 12:44.55 |
| sebras | ManDay: No PDF is a different format, which has some ideas taken from postscript, but is not as general, AIUI. | 12:53.07 |
| kens | PDF uses (and extends) the graphics model from PostScript, PostScript is a Turing-complete programming language, PDF is simpl a file format. | 12:58.58 |
| ManDay | So PDF is to PS something like a syntax-conform language extension? | 13:00.57 |
| kens | Umm, no not really. | 13:01.09 |
| | You can write things like 'for' and such in PostScript, you can't in PDF. | 13:01.27 |
| ManDay | Ah | 13:01.44 |
| kens | PDF uses the PostScript graphics model for drawing but extends it in several ways (transparency, path and text modes) | 13:01.56 |
| | PDF includes metadata and user-interaction PostScript deos not. | 13:02.32 |
| | and so on | 13:02.40 |
| | But because the graphics models are similar its reasonably possible to convert between them. | 13:03.03 |
| ManDay | and on the other hand PDF is not turing complete | 13:03.10 |
| | I understood that PDF only extended ps first | 13:03.19 |
| | i think i got it now | 13:03.25 |
| kens | No, its not a programming language at all, its a 'Page *Description*' | 13:03.31 |
| ManDay | thanks :) | 13:05.49 |
| kens | No problem. | 13:05.55 |
| Robin_Watts | PDF spec overload. Lunchtime. | 14:06.05 |
| kens | Well, I've managed to reduce the customer file fro 4Mb to 830Kb but reducing it further makes the problem go away. | 14:50.04 |
| | It looks like its not pdfwrite at afult though. | 14:50.15 |
| | The colour space being handed in for the image has an empty name for the Separation. | 14:50.33 |
| chrisl | Huh? So how come it works sometimes? | 14:52.38 |
| kens | Well it must still be memory corruption, just not in pdfwrite. | 14:52.58 |
| | Or not obviously so | 14:53.08 |
| | But you're welcome to prove me a liar.... | 14:53.32 |
| chrisl | Oh, I see. What happens if you run it with -dNOGC? | 14:53.42 |
| kens | Probably it works, I'll try it | 14:53.55 |
| | Yes that works. | 14:54.49 |
| | But probably only because memory moves around | 14:55.03 |
| chrisl | Hmm, yes. | 14:56.33 |
| kens | I htink the next thing is to stop when the colour space is defined and then see when it changes. | 14:56.41 |
| | THe data that is | 14:56.46 |
| | Presumably I can set a data breakpoint in ddd | 14:57.01 |
| chrisl | It might not help if a gc cycle runs - the name could move | 14:57.39 |
| kens | Looks like it may not be possible anyway, the memory containing the colour space is different every time | 14:58.16 |
| chrisl | You have to reset watch points each run for them to work properly, anyway | 14:59.19 |
| kens | Its not so much that, more how do I know which colour space is going to be the one to watch. | 14:59.56 |
| chrisl | How many separation spaces are there? | 15:00.22 |
| kens | One. | 15:00.28 |
| | Invoked a number of times. | 15:00.38 |
| chrisl | How many images? | 15:01.01 |
| kens | I'm not sure. But at least 4 | 15:01.12 |
| chrisl | Ugh :-( | 15:01.22 |
| | Can you ID the image from dimensions or something? | 15:01.56 |
| kens | Actually I take that back, its 4 executions of the named colour space with a /Separation colour | 15:02.00 |
| | In the case that fails its an image, the earlier ones I haven't checked | 15:02.17 |
| | I'm breaking on a known number of executions :-) | 15:02.27 |
| chrisl | So, I'd start by assuming the color space is intact at the beginning of the image in question, and work from there | 15:03.50 |
| kens | I'm not sure I can work backwards like that, at least with my minimal grasp of ddd | 15:04.19 |
| ManDay | What's wrong with "gs -sOutputFile=test.pdf -sDEVICE=pdfwrite /usr/share/doc/groff-1.21/meref.ps"? | 15:05.31 |
| chrisl | It shouldn't be hard to find the call to the image (or whatever) operator for the image in question | 15:05.31 |
| kens | Looks like 19 executions of setseparationcolorspace | 15:05.40 |
| | ManDay, what error do you get ? | 15:06.03 |
| | Hmm ddd says 17 hits | 15:07.51 |
| ManDay | kens: After I run that cmd I end up with a gs-prompt saying >>showpage, press <return> to continue<< | 15:08.49 |
| | So then I press Ctrl C | 15:08.54 |
| | Then, the PDF is broken according to poppler | 15:09.03 |
| kens | Well don't do that. Press return like it says | 15:09.07 |
| | Then type 'quit' | 15:09.20 |
| | Alternatively set -dBATCH -dNOPAUSE | 15:09.30 |
| ManDay | JOh | 15:09.41 |
| kens | You are killing the application before its written the PDF file | 15:09.45 |
| ManDay | I see | 15:09.54 |
| | Yes, that works, ty! | 15:10.12 |
| kens | You're welcome | 15:10.35 |
| Robin_Watts | ManDay: If you use -o rather than -sOutputFile= that implies -dBATCH and -dNOPAUSE, I think. | 15:19.17 |
| kens | chrisl so it looks like hte setseparationspace is at fault, tis defintiely creating an empty name. | 15:20.27 |
| | Now to figure out why. | 15:20.37 |
| ManDay | Robin_Watts: I'll try that next time! | 15:23.48 |
| chrisl | kens: is it reusing a color space - corrupted between set space calls? | 15:26.09 |
| kens | chrisl it may well be. | 15:26.21 |
| | The colour space is a persistent object | 15:26.30 |
| | I think ddd is deluding me :-( | 15:29.57 |
| | Hmm, or maybe not. | 15:31.31 |
| | I may be doing something bad in the colour code. | 15:31.51 |
| | I call names_ref with teh same pointer for source and destination, that doesn't look like its a good diea at all | 15:32.20 |
| | Converting a tring into a name object | 15:33.02 |
| | Think I'll look at this in VC to see if I can understand it better. | 15:33.48 |
| chrisl | Are you transferring ownership of the pointer between the calls? | 15:34.45 |
| kens | I don't think so | 15:35.58 |
| | Look at zcolor.c:3389 | 15:36.18 |
| | code=name_from_string(...) | 15:36.30 |
| | Its been a long time since I did this, but that looks dubious | 15:37.19 |
| chrisl | Hmm, I don't see any code like that in zcolor.c...... | 15:37.55 |
| kens | In setseparationspace. | 15:38.04 |
| | VC says line 3425 | 15:38.15 |
| | WHich disagrees with ddd but I trust it better | 15:38.24 |
| | code = name_from_string(imemory, &sname, &sname); | 15:38.38 |
| chrisl | Ah, yes, I see it | 15:39.25 |
| kens | Hmm very few calls to that routine | 15:40.05 |
| chrisl | Yeh, and the one in ztype.c uses the same pointer for source and dest | 15:40.40 |
| kens | So it does. | 15:40.55 |
| | Well that's where the problem occurs though | 15:41.02 |
| | If you set a breakpoint in setseparationspace with a count of 17 you should be able to see it too | 15:41.22 |
| | err, maybe count of 16 | 15:41.32 |
| | Cos ddd is an 'ignore' count | 15:41.39 |
| chrisl | That'll be for your cutdown file, though, won't it? | 15:41.54 |
| kens | Yes, but that only throws away stuff after the error. THough I cna send it to you if you like. | 15:42.15 |
| | TH efull file should be the same | 15:42.20 |
| chrisl | The eventual function seems to be names_ref() and the comment says: "The return may overlap the characters of the string!" | 15:44.18 |
| kens | I'm not sure what that is saying though. | 15:44.43 |
| | I'm debugging through it now though | 15:44.56 |
| | So it creates a hash and then looks for the existence of a name with that hash. | 15:45.22 |
| | I wonder if hte tbale is corrupted. | 15:45.29 |
| chrisl | Oh, for goodness sake - a function just to de-reference a pointer...... | 15:45.57 |
| kens | Which one ? | 15:46.18 |
| chrisl | names_memory() | 15:46.25 |
| kens | Oh, haven't seen that one yet | 15:46.37 |
| | name index appears to be 514 | 15:46.52 |
| | and the name is correct | 15:47.15 |
| | But name_index_ptr_inline returns an empty data string | 15:48.03 |
| | So I think the table may well be corrupt | 15:48.10 |
| | Or maybe not, I'm not sure what the make_name thing does | 15:49.11 |
| | Oh its a macro, wonderful | 15:49.26 |
| | My head hurts :-( | 15:50.02 |
| chrisl | It *seems* to find the name in the table, and then something weird happens...... | 15:51.33 |
| kens | Yes, that's what I'm seeing | 15:51.41 |
| | It all looks OK then it returns a NULL | 15:51.49 |
| chrisl | Yeh..... | 15:52.02 |
| kens | of course make_name just calls maske_tasv which is another macro | 15:53.21 |
| | Well it all looks OK, so now I'm baffled. | 15:56.49 |
| | Coffe break I think | 15:56.56 |
| chrisl | kens: I think you're right, I think the name table is corrupt: nt->sub[1].names->names[19].pvalue = NULL | 15:59.09 |
| kens | :-( | 15:59.23 |
| | Well I guess I need to see when that happens | 15:59.31 |
| chrisl | but nt->sub[1].strings->strings[19] is a string object with "PANTONE 877 C" in it | 16:00.01 |
| kens | sigh... | 16:00.18 |
| sybariten | hey | 16:04.37 |
| ghostbot | hey, sybariten | 16:04.37 |
| sybariten | anyone using xnview with vector viewing on Win ? | 16:04.52 |
| kens | I think ray_laptop does, but he's not online right now | 16:05.14 |
| | Not really our area of expertise | 16:06.05 |
| sybariten | ok, no... | 16:10.41 |
| | but hey, guess what | 16:10.45 |
| | i just looked a little closer, and sure enough there _is_ information and settings for changing the default viewing resolution when using xnview + gs | 16:11.17 |
| | i had a default value of 72 dpi being used, and most files come up as, well, 5 cm images on a typical computer setup... my work will actually be easier now after having changed this setting | 16:12.32 |
| henrys | I see foxit has an android port now. | 16:33.08 |
| kens | Its a popular platform | 16:33.20 |
| ManDay | Eh, how come I've never heard of gv | 16:36.13 |
| Robin_Watts | Coo. The latest XFA doc is only 7 weeks old. | 16:36.13 |
| ManDay | wait, is that... motif... ? | 16:37.17 |
| chrisl | ManDay: Motif is a GUI toolkit for X | 16:37.57 |
| kens | gv might use Mortif, but I thought it was X | 16:38.14 |
| chrisl | kens: I now can't reproduce the empty separation name on Linux | 16:38.39 |
| ManDay | kens: eh? "It was X"? | 16:39.17 |
| kens | chrisl I still can :-) | 16:39.29 |
| chrisl | Thing is, I haven't changed anything..... | 16:40.27 |
| Robin_Watts | kens: Have you ever tried "gdb -tui" ? | 16:41.58 |
| kens | no, what does it do ? | 16:44.43 |
| | using ddd anyway | 16:44.53 |
| chrisl | it's a curses interface to gdb | 16:44.57 |
| Robin_Watts | wot chrisl said. | 16:45.08 |
| kens | will stick with ddd | 16:45.22 |
| chrisl | IIRC, you still need to know the gdb command line to drive it | 16:45.29 |
| Robin_Watts | yes. | 16:45.58 |
| chrisl | Didn't really see the point | 16:46.11 |
| Robin_Watts | Seeing an updated program listing while stepping seems a big help to me. | 16:46.34 |
| | Likewise you can see registers and stuff. | 16:46.43 |
| chrisl | I get that, and better, with DDD | 16:46.48 |
| Robin_Watts | I must revisit ddd. | 16:47.25 |
| henrys | emacs is the time honored way to use gdb on any platform ;-) | 16:47.36 |
| Robin_Watts | makes tea so as to be fortified for the JS/Forms discussion to come. | 16:48.13 |
| henrys | make it strong | 16:48.25 |
| chrisl | I'm not bigging up ddd, it's clunky and agricultural - but it still seems to me to be the best compromise for a gui front end for gdb | 16:48.40 |
| kens | And better than that, I know minimally how to use it :-) | 16:53.33 |
| henrys | the gui situation is unfortunate gdb has all the hooks to do something as full featured as Visual ... | 16:55.17 |
| chrisl | The problem is, it's virtually impossible to use both the GUI hooks and the gdb command line at the same time - so with any gui using the hooks you're always going to lose access to gdb newer or more advanced features | 16:57.00 |
| Robin_Watts | chrisl: really? | 16:57.48 |
| chrisl | about what? | 16:58.02 |
| henrys | I don't know xcode looks pretty good to me. | 16:58.04 |
| chrisl | henrys: can you expand macros in xcode? | 16:58.25 |
| Robin_Watts | I could cope with a button to swap between gui and console mode. | 16:58.46 |
| henrys | not sure but I can in emacs ;-) | 16:58.54 |
| chrisl | Robin_Watts: you have restart gdb to swap.... | 16:59.13 |
| henrys | anyway we should start the meeting. | 16:59.34 |
| Robin_Watts | oh. that's pants. | 16:59.38 |
| | OK. | 16:59.44 |
| henrys | for paulgardiner the goal here is to have something at least as featured as google chrome's built in pdf we believe this is technology from foxit. | 17:00.20 |
| paulgardiner | henrys: do they advertise exactly what features they support? | 17:01.15 |
| henrys | paulgardiner:not as far as I knowl | 17:01.30 |
| | to that end I've been playing with it. | 17:01.40 |
| paulgardiner | There are two quite different types of form, and it would help a lot to be able to concentrate on one. | 17:01.52 |
| | .. at first, at least. | 17:02.12 |
| henrys | for one I notice it does not save form data but it does print it. That said I think we do need to save form data. | 17:02.13 |
| | I think we can dispense with XFA entirely no? tor8? Robin_Watts? | 17:02.31 |
| Robin_Watts | There are 4 different ways to save form data, as far as I can make out. | 17:02.35 |
| | I've been reading the specs etc (and have been discussing stuff with paulgardiner). | 17:03.04 |
| | First off, we have "AcroForms" | 17:03.14 |
| henrys | right and in my reading I thought that would suffice. | 17:03.37 |
| Robin_Watts | AcroForms gives us forms with a static field structure. | 17:03.57 |
| henrys | is tor8 here? | 17:04.04 |
| Robin_Watts | Then we have AcroForms + Javascript, which gives us dynamic field creation. | 17:04.18 |
| | Then we have the newer XFA based stuff which is uber dynamic and needs (from what paulgardiner was reading) dynamic reflow of the underlying document etc. | 17:05.18 |
| henrys | when you say dynamic I think of dynamic layout which is XFA stuff. | 17:05.24 |
| Robin_Watts | Right, no. I'm trying to distinguish between those levels. | 17:05.49 |
| | AcroForms by themselves specify a fixed set of fields. | 17:06.09 |
| henrys | And I was suggesting we restrict the project to Acroforms and js | 17:06.15 |
| paulgardiner | The sort of dynamic creation achieved with AcroForms plus Javascript doesn't require reflow of the document, I believe | 17:06.24 |
| Robin_Watts | Add Javascript into that mix, and you can add new fields dynamically, but you don't need to reflow. | 17:06.32 |
| | (please add "AIUI" to the end of every line I type :) ) | 17:06.54 |
| | henrys: That sounds a very attractive idea, but... we ought perhaps to check whether that's enough. | 17:07.16 |
| | paulgardiner pointed out on the phone earlier that Adobes docs point everyone at using LiveCycle Designer now, and that makes XFA. | 17:07.55 |
| | So we might find that we spend a lot of time implementing AcroForms only to find that all new forms are coming out in XFA format. | 17:08.25 |
| | That being said, it's quite possible that the underlying work we'll need to do to do AcroForms won't be wasted when we come to do XFA, but... | 17:08.55 |
| paulgardiner | XFA distinguishes betwen static and dynamic subtypes, and the static type doesn't require reflow of the document. | 17:09.10 |
| Robin_Watts | Even within just supporting AcroForms, there are other things to consider. | 17:09.40 |
| | Do we want to support import/export of data? Or just support submission of it? | 17:09.54 |
| henrys | XFA looks extremely ambitious to me, and I suspect we can emulate what is in chrome with Acroforms (I haven't verified) that. | 17:09.55 |
| | I'd like to hear what tor8 has to say. | 17:10.06 |
| Robin_Watts | There is FDF (basically cut down PDF) (can be used for importing/exporting) | 17:11.11 |
| | There is XFDF ("what, no one liked FDF? Damn. Let's redo it in XML.") | 17:11.42 |
| | There is PDF (export the completed form as a PDF file) | 17:12.04 |
| | and then finally there is XFA. | 17:12.11 |
| | I suspect the most common use case is for people to download a PDF file, with no external dependencies, and then either 1) save that file out as PDF, or 2) submit the results using a GET or a POST HTTP request. | 17:13.17 |
| henrys | Robin_Watts:I've just been playing with Acrobat to see how completed forms are saved in PDF's and that looks straightforward enough. An object is just appended. | 17:13.30 |
| | I haven't studied the other ways yet. | 17:14.03 |
| Robin_Watts | That cuts FDF/XFDF/XFA and everything other than saving PDF out of the equation, which makes it much more tractable. | 17:14.04 |
| | henrys: Right. | 17:14.19 |
| henrys | we likely should postpone until tor8 is here. | 17:14.19 |
| paulgardiner | henrys: Is it to support particular form features that is important, or will there be a set of existing documents that we need to support? | 17:14.50 |
| tor8 | sorry about the delay | 17:15.11 |
| henrys | tor8:have a look at the logs. | 17:15.22 |
| | paulgardiner:no we don't have an existing set of documents. I could see if our testing company has form testing, that might be useful. | 17:16.05 |
| | we use quality logic for most testing. | 17:16.18 |
| | s/most/a lot of/ | 17:16.30 |
| Robin_Watts | Searching for "PDF Form" on QualityLogics site doesn't produce anything. | 17:17.06 |
| paulgardiner | Sorry, I meant is a customer likely to have a set of existing forms that they want to work, or is it more likely that they care about features, and will be willing to construct the forms? | 17:17.13 |
| Robin_Watts | Depends on the customer, I suspect. Some will see "Form Filling" as a tickbox item. | 17:17.40 |
| | Others may be wanting to roll out our tool into an existing infrastructure. | 17:18.08 |
| tor8 | agrees with Robin on most points he's already stated | 17:18.12 |
| Robin_Watts | Others will want to build the infrastructure around our tool. | 17:18.24 |
| tor8 | the project doesn't have a concrete target, we're building this more to be something useful than against a given specification | 17:18.42 |
| henrys | paulgardiner:some of this should be private and I'll email you something. | 17:18.57 |
| tor8 | so what people actually use it for is what we should aim for. this whole XFA thing scares me, but isn't a lot of that covered by what flash and Air and all the other "let's make a framework for writing applications" stuff is about? | 17:19.46 |
| paulgardiner | Oh ok. Thanks. | 17:19.50 |
| mvrhel | henrys: is alexcher around today do you know? | 17:20.07 |
| henrys | paulgardiner:just sent you an email that should help. | 17:22.01 |
| | mvrhel:haven't seen him - SMS him. | 17:22.28 |
| paulgardiner | henrys: ta | 17:22.34 |
| mvrhel | henrys: ok | 17:22.43 |
| Robin_Watts | henrys: Did you copy it to me? I've not heard of anything on this subject that should be private, so I'd be interested to read it too. | 17:23.03 |
| henrys | done we did discuss it at the meeting. | 17:24.04 |
| Robin_Watts | oh, right, that. thanks. | 17:24.34 |
| henrys | and you didn't have a puzzle this time so I assumed you were listening ;-0 | 17:24.49 |
| | sent to tor8 also. | 17:25.04 |
| paulgardiner | It would be great to be able to neglect XFA. I guess it shouldn't be hard to determine if we need handle it. | 17:25.05 |
| Robin_Watts | tor8: You said you agreed with me on "most" things - was there something specific you disagreed on ? | 17:25.30 |
| tor8 | whether or not XFA is worth supporting :) if it does turn out we need big fat reflowing I'm not sure we can/should proceed. | 17:26.22 |
| Robin_Watts | My opinion on that was "I really hope we can get away with not supporting it." | 17:27.01 |
| | I think we can support the non-reflowing kind. Reflowing is more of a problem :) | 17:27.16 |
| | but I hope we can afford not to, at least initially. | 17:27.29 |
| paulgardiner | XFA doesn't necessarily imply reflowing. It has a static form (apparently determines when you save out your freshly created form). | 17:27.31 |
| henrys | tor8:I wanted to dispense with XFA entirely and just do Acroforms | 17:27.52 |
| Robin_Watts | Do we know of any example XFA forms? | 17:28.18 |
| henrys | at least for now. | 17:28.20 |
| Robin_Watts | If our competition doesn't handle them, then we can feel more secure in ignoring it. | 17:28.49 |
| henrys | we could create some XFA document and see what chrome does with it. | 17:28.58 |
| Robin_Watts | I don't know if we need LiveCycle Designer to be able to create such things, or whether we can hand hack them with any degree of success. | 17:29.38 |
| paulgardiner | Wording carefully: could we create an XFA form and see what other readers handle it? | 17:29.41 |
| | ah, you already said that. | 17:30.04 |
| Robin_Watts | I think it's tor8's turn to suggest the same thing, and then we'll all have done it :) | 17:30.19 |
| henrys | paulgardiner:seems like a next crucial step to me. | 17:30.22 |
| Robin_Watts | Reading through the PDF doc on forms today, I found myself making rough notes on the kind of interface we'd need to allow form handling. | 17:32.46 |
| | (for AcroForms at least). | 17:32.50 |
| henrys | we do have to do AcroForms anyway right - and I can't imagine AcroForms would somehow prevent us from doing XFA. | 17:33.35 |
| Robin_Watts | Something to enumerate fields in a document, 'activation' methods, selection etc. | 17:33.53 |
| paulgardiner | I've found a pdf document entitled "Creating a Radio Check Box XFA Sample Form", and it has a Radio Check Box in it. I'm hoping it's self referential | 17:33.57 |
| henrys | paulgardiner:ah great. | 17:34.41 |
| Robin_Watts | And it struck me that that interface is hopefully the kind of interface that exists within javascript. | 17:34.47 |
| paulgardiner | http://www.windjack.com/DownLoads/RadioCheckBox_XFA.pdf | 17:34.51 |
| henrys | works great in chrome sigh | 17:35.28 |
| Robin_Watts | Well... the clear button doesn't. | 17:35.46 |
| henrys | let's look at the pdf... | 17:36.41 |
| paulgardiner | Clear works in FF | 17:36.45 |
| henrys | I do see an acroform in here as well. | 17:37.54 |
| tor8 | looking inside that pdf the annots for the forms look like the usual stuff | 17:38.11 |
| paulgardiner | Here's another: http://smarte-forms.com/wp/wp-content/uploads/2011/03/birth-certificate-application-eform-tt-sample.pdf | 17:39.29 |
| | At the top of the page "SAMPLE PDF/XFA e-Form" | 17:40.01 |
| Robin_Watts | tor8: There *is* XFA code in that file. | 17:40.05 |
| | but I suspect that there is enough AcroForms stuff in there to explain what we're seeing in Chrome. | 17:40.31 |
| henrys | so I guess someone needs to figure out if we implemented only acroforms what what would be lacking in these fiels. | 17:40.33 |
| Robin_Watts | FF uses Acrobat. | 17:40.37 |
| tor8 | C:\DOCUME~1\thomp\LOCALS~1\Temp\_150k136cu5e3ebls.log :) | 17:40.58 |
| henrys | I only scheduled 45 minutes for this meeting I assume you guys can carry on without me and I can catch up later. | 17:41.21 |
| Robin_Watts | In fact, let me just cut out the xfa crap and see what happens when I reload it. | 17:41.23 |
| henrys | paulgardiner:are you okay with invoicing and everything is all that working, anything I need to do with Miles? | 17:42.12 |
| paulgardiner | Could it be common to include both forms in a document? | 17:42.30 |
| Robin_Watts | OK, with all the xfa crap cut out, Chrome still behaves exactly the same. | 17:43.36 |
| paulgardiner | Henrys: I'm not aware of any problems. I haven't invoiced for anything yet. I was going to wait until I'd finished Tor's requests. | 17:44.45 |
| henrys | okay I have to run now and will catch up later. | 17:45.21 |
| paulgardiner | Robin_Watts: Can you do the opposite test, and knock out the AcroForm? | 17:46.27 |
| Robin_Watts | In a mo. Just knocking out the XFA from the second document. | 17:46.48 |
| henrys | bbiaw | 17:48.19 |
| Robin_Watts | OK, second document also seems to work as a basic form without the XFA stuff. | 17:48.58 |
| paulgardiner | I must install Chrome. What's the best way? I have some memory of there being two different places to get it. | 17:49.03 |
| Robin_Watts | paulgardiner: Go to google, click the advert for chrome :) | 17:49.30 |
| | You want chrome, not chromium. | 17:49.38 |
| | The Male/Female tickboxes don't work right in my cutdown version. | 17:50.14 |
| | but they do in the full version. | 17:50.31 |
| | I'm not sure how to remove the Acroforms stuff without destroying the XFA. | 17:51.41 |
| tor8 | Robin_Watts: from my limited understanding, XFA form documents come in two flavors | 17:52.36 |
| paulgardiner | Robin_Watts: you noticed a difference in behaviour without the XFA in Chrome? | 17:52.37 |
| tor8 | one: static pdf backdrop with XFA forms in the foreground | 17:52.48 |
| | two: no backdrop, everything done with XFA forms (using xml and css styling) | 17:53.08 |
| Robin_Watts | tor8: That's not my reading. | 17:53.30 |
| paulgardiner | I think that's true. The most recent doc I looked at said something similar | 17:54.03 |
| Robin_Watts | One is a static PDF backdrop with XFA forms in the foreground - but if I remove the XFA from the file, I get exactly the same behaviour - in chrome at least. | 17:54.07 |
| | And if I remove the /Fields reference, so leaving just the XFA - I get exactly the same behaviour still! | 17:55.12 |
| paulgardiner | I have a suspicion that the reflowing around the form happens only in pure XFA based forms, not when XFA is contained in PDF. | 17:55.53 |
| Robin_Watts | With file 2, if I remove the xfa, I still get most of the output. which leads me to believe that it's a similar thing to 1. | 17:56.43 |
| paulgardiner | XFA Spec V 3.3 talks about two possibilities: PDF contains XFA and XFA contains PDF. | 17:56.53 |
| sebras | Robin_Watts: you never settled whether saving form contents is worthwhile to pursue. the only (acro)form PDFs I've seen were tax office forms, so they are destined for the printer as soon as I stop typing... | 17:57.05 |
| tor8 | Robin_Watts: if I cut out the /Annots link from the page object, the button goes away and the checkboxes don't act like checkboxes anymore | 17:57.08 |
| Robin_Watts | sebras: Indeed. | 17:57.16 |
| | I strongly suspect that AcroForms, plus printing is all we need for a worthwhile step forwards. | 17:57.36 |
| | saving would be an added bonus. | 17:57.44 |
| tor8 | Robin_Watts: look at the javascript in object 29 | 17:58.47 |
| sebras | Robin_Watts: and _if_ saving is interesting, then it may be smart to save the data the data in an external file. | 17:58.57 |
| Robin_Watts | tor8: Which file ? | 17:59.23 |
| tor8 | RadioCheckBox_XFA.pdf | 17:59.34 |
| Robin_Watts | Right. That forms a version independent ADBE.SYSINFO string | 18:00.38 |
| | Sorry... That's figuring out what version of Acrobat to tell you to update to. | 18:01.10 |
| tor8 | lots of javascript testing whether XFA is available and pops up some alerts telling you to update | 18:01.18 |
| Robin_Watts | None of which I see Chrome running for me. | 18:01.45 |
| | The thing I don't understand with Radio, is that I've removed BOTH the Fields and the XFA... and I still see it behaving so I can select the items. | 18:02.22 |
| tor8 | Robin_Watts: remove the /Annots in the page object (object 32) | 18:02.53 |
| Robin_Watts | Ah! | 18:03.09 |
| | OK, so I think we can safely say that Radio does not require XFA to perform in Chrome the way it is. | 18:03.56 |
| | (i.e. Radio gives us no evidence for Chrome supporting XFA) | 18:04.24 |
| | And the pdfcleaned birth.pdf works differently to the unpdfcleaned one anyway. | 18:06.13 |
| | But removing XFA from the unpdfcleaned one, it still behaves the same. | 18:07.26 |
| tor8 | the forms support for these XFA samples in chrome seems dodgy at best | 18:08.19 |
| | I suspect it's picking up underlying acroform widgets and failing hard at the button scripts | 18:08.49 |
| Robin_Watts | http://www.google.com/support/forum/p/Chrome/thread?tid=325f0de367340434&hl=en | 18:09.01 |
| | So Chrome does NOT support saving of the answers from forms. | 18:09.13 |
| | "Chrome's built-in PDF viewer has limited support for PDF forms, and does not support JavaScript, which is what Acrobat uses behind the scenes when you use one of the built-in calculation options." | 18:10.52 |
| | I don't believe that :( | 18:10.58 |
| tor8 | they probably wrote that to cover their behind when it fails :) | 18:11.18 |
| Robin_Watts | That was from a forum - probably an uninformed luser. | 18:11.45 |
| | From reading these forums, it sounds like the Chrome form filling was dodgy - a year ago. | 18:12.44 |
| sebras | Robin_Watts: I just ran strings on tor8's chrome binaries. XFA seems nowhere to be seen... | 18:37.15 |
| marcosw_ | Is anyone else having trouble with bugs.ghostscript.com? | 18:46.23 |
| mvrhel | henrys: you around? | 18:49.15 |
| paulgardiner | "Web Browser that has full support for XFA Forms. Google Chrome for example does not support XFA forms (as per my recent findings, if I did not miss anything)." | 18:50.27 |
| | http://forums.adobe.com/message/4160932 | 18:50.45 |
| mvrhel | henrys: the fix for proper setting/resetting of the icc profiles from the user params has been committed so now your stuff should work. | 18:51.41 |
| Robin_Watts | paulgardiner: OK. I've created my own PDF form. | 18:52.06 |
| mvrhel | last issue was with the transparency mask that has a pattern drawn in it that also included a transparency mask. all the icc profile swapping was getting mixed up in that case | 18:52.19 |
| paulgardiner | XFA? | 18:52.22 |
| Robin_Watts | Adobe LiveCycle Designer comes as part of Acrobat Pro. | 18:52.34 |
| paulgardiner | Right. | 18:52.43 |
| Robin_Watts | No, this is supposed to be a PDF form. | 18:52.45 |
| | But I can export it in different formats I think. | 18:52.55 |
| paulgardiner | Sorry. I meant a PDF form created via LCD. | 18:53.44 |
| marcosw_ | anyone mind if I reboot casper? Currently sebras and Robin_Watts are logged on. | 18:55.07 |
| Robin_Watts | Go for it. | 18:55.16 |
| | Do we have time to organise bets on how long it will be down for? :) | 18:55.40 |
| aditsu | hi, how can I rotate pages in a pdf? I tried for example: -c "<</Orientation 2>> setpagedevice" but got Error: /undefined in app2.pdf | 20:39.43 |
| | oh, I had to add -f before the input file name, but now it doesn't have any effect | 20:47.54 |
| | ugh, I'll try with pdftk | 20:57.35 |
| | done | 21:01.08 |
| henrys | mvrhel:did you ever hear back from alexcher? I guess a bunch of us are looking for him. | 22:40.58 |
| | tor8:ping | 22:42.36 |
| tor8 | henrys: yes? | 22:43.38 |
| henrys | do you remember something about printing pages before gettin the entire pdf as being part of the reason we lost out? Raph is in discussion with Scott and he raised that issue. | 22:45.29 |
| | And apparently Raph had a long meeting with miles enumerating all the issues if we can't recover that I think we should talk to Raph directly. | 22:46.05 |
| Robin_Watts | about us not being able to display/print pages before the whole lot was downloaded? | 22:46.25 |
| | mupdf can't make use of linearized format PDF currently, but it might not be so bad to add. | 22:46.52 |
| tor8 | no memories | 22:47.02 |
| Robin_Watts | I have to go fetch Helen from the station. | 22:47.09 |
| henrys | I think I should send an email to raph anyway at this point, it would be a shame to do this project and find out we are missing some piece we could have done ... | 22:47.12 |
| | tor8:thanks I'll talk to Miles or Raph | 22:47.36 |
| | tor8:so I'll send him an email and copy you and the others in - unless you want to stay dark on the whole issue. | 22:51.53 |
| Robin_Watts | I'd love to be involved. | 23:34.49 |
| | (nosiness is my default position) | 23:35.02 |
| henrys | If we actually need to compete with foxit we've a lot to do. XFA for one. | 23:36.11 |
| | see foxit phantom | 23:36.49 |
| | Forward 1 day (to 2012/03/06)>>> | |
IRC Logs