| <<<Back 1 day (to 2014/06/05) | 2014/06/06 |
mvrhel_laptop | ok. so I have removed the dependency of the mupdfnet dll on the windows CRT dlls. time to make a new installer of for gsview | 04:49.56 |
superdealloc | Hello everyone. Anyone from Artifex around? | 08:36.54 |
kens | Yes | 08:37.02 |
superdealloc | Hi kens! | 08:37.20 |
| My company would like to know how much it would cost to sponsor a feature | 08:37.37 |
| This one in particular: http://bugs.ghostscript.com/show_bug.cgi?id=690626 | 08:37.41 |
kens | That's MuPDF, not my field. You would be best to contact either Miles Jones or Scott Sackett in the US (CEO and VP Sales respectively). You are talking to the eningeers here and we don't do money :-) | 08:38.35 |
| I should probably also say that we are very short handed at the moment... | 08:39.20 |
superdealloc | Good times! :-( | 08:40.46 |
kens | We recently took on another product and lots of our engineering resource is currently diverted to getting that straightened out. I'd suggest you contact Miles, or info@artifex.com | 08:41.44 |
| (I think that's a real mail address) | 08:42.04 |
| miles.jones at artifex dot com will get you to Miles | 08:42.27 |
superdealloc | Alright. I'll get on it. Thank you! | 08:43.33 |
kens | No problem | 08:43.39 |
| chrisl ping | 08:44.11 |
chrisl | kens: pong | 09:40.29 |
kens | Looking at a annotation problem and I think the problem is that the annotation is using an inappropriate font | 09:41.27 |
| Its a Type 0 with a single Type 11 descendant and the Encoding is given as 'Identity-H', the text however is UTF16 | 09:42.04 |
| Can you think of an easy way I can change the font encoding to test it ? | 09:42.24 |
| Its bug #694738 if you want to look at the file | 09:43.28 |
chrisl | kens: I'll have a look in a little while - in the middle of something just now | 09:48.11 |
kens | No problem, I'm off out shortly and will be back after lunch | 09:49.06 |
chrisl | kens: you might be able to change the encoding to Identity-UTF16-H ? | 09:54.38 |
| kens: there are also three definitions of a CMap "/Adobe-Identity-UCS" embedded in the file - is it possible we should be using one of those?? | 10:03.28 |
Robin_Watts | superdealloc: Hi | 11:36.46 |
superdealloc | Hey Robin_Watts | 11:36.56 |
| Feeling less jetlagged? :-) | 11:37.01 |
Robin_Watts | So, you're interested in getting color management pushed into mupdf? | 11:37.16 |
| superdealloc: Oddly, I'm not feeling to bad at all this time. | 11:37.29 |
superdealloc | That's gotta be a superpower of some sort. | 11:37.40 |
| Where are you based on? | 11:37.43 |
Robin_Watts | I'm in the UK. | 11:37.50 |
superdealloc | Yeah, I'm in Amsterdam. | 11:37.59 |
| I'd be floored if I were you... | 11:38.05 |
| So yeah | 11:38.08 |
| My company would like to sponsor color management on MuPDF. | 11:38.18 |
| I already got as far as plugging LCMS2 in | 11:38.29 |
| but it's a very naive implementation, as it used a fixed color profile. | 11:38.42 |
Robin_Watts | right. | 11:38.49 |
superdealloc | We wanted to contribute it first, but after trying to understand how GS does it (in it's 100s of lines of beauty) we simply thought sponsoring the people who know to be the best option. | 11:39.08 |
Robin_Watts | which company? (If you can say in public) | 11:39.16 |
superdealloc | Publitas | 11:39.19 |
| https://www.publitas.com/ | 11:39.29 |
Robin_Watts | We had planned to get mvrhel to do the color management in mupdf, but it keeps being pushed back with other stuff. | 11:39.58 |
superdealloc | We're evaluating MuPDF as a tool which will allow us to selectively export parts of a PDF (split it in JPG/SVG files for better text) | 11:40.06 |
Robin_Watts | he is our color expert, and did the gs work. | 11:40.21 |
superdealloc | Job well done on that front, btw. | 11:40.40 |
| GS has the best color reproduction out of everything we've evaluated. | 11:41.16 |
| Including PDFTron. | 11:41.19 |
Robin_Watts | He's based in Seattle, and miles is in San Raphael, so there is a timezone difference when communicating there. | 11:41.43 |
superdealloc | That's alright. | 11:42.08 |
Robin_Watts | At the moment all mupdf color handling is done by pushing to rgb and back - that needs to change. | 11:42.16 |
superdealloc | I've also emailed your CEO, at kens' recommendation | 11:42.18 |
Robin_Watts | we've made some changes in mupdf so that color management should be able to go in with a minimum of disruption though. | 11:43.16 |
superdealloc | Oh yeah | 11:43.45 |
| This is what I have | 11:43.57 |
| https://gist.github.com/andremedeiros/75979fff98f5a8af85f7 | 11:43.57 |
| Takes a bit more abstracting (ie. attaching the transformation to the actual fz_colorspace struct) | 11:44.12 |
| but it's close. | 11:44.16 |
| only thing is picking up the right color profile, which is impossibly hard in my experience :-( | 11:44.31 |
kens | chrisl sorry, I went out and thought I'd set my PC to sleep, but it didn't..... So I've been afk for the last couple of hours. | 11:48.36 |
| I did try changing the Identity-H to Identity-UTF16-H and that made no difference | 11:48.57 |
| The UCS2 CMap is in fact a ToUnicode CMap | 11:49.08 |
chrisl | kens: actually, I thought if we just skipped the opening two bytes, we'd get the right results | 11:51.18 |
kens | I coded that and we get the wrong result | 11:51.29 |
| Which is why I think the encoding is incorrect | 11:51.42 |
chrisl | Hmm..... | 11:52.19 |
kens | I'll hack something up to select the fallback CIDFont and see what I get after I finish lunch | 11:52.20 |
chrisl | Actually, I thought at the moment, we only used a single byte font | 11:53.43 |
kens | The code selects the font from the defaul appearance (DA) which selects /stanv_font | 11:54.20 |
| WHich is a type 0 font with a /Identity-H 'Encoding' | 11:54.48 |
| We do get the right number of glyphs out, so either the /.notdef is non-marking (which it isn't, we cansee it when the FEFF is drawn) or we are correctly consuming 2 bytes for each glyph | 11:56.02 |
| Hmm, chrisl using the DroidSansFallback gets the same incorrect result as using /stanv_font | 12:25.20 |
| If I send you a modified couple of files from the PDF interpreter can you take a quick poke and tell me what I'm doing wrong ? | 12:25.46 |
chrisl | kens: it's possible the file is wrong...... | 12:27.26 |
kens | I'm not sure, much of the text is 'ASCII', ie 0x00 and teh secondbyte is regular latin text | 12:27.54 |
| Acrobat (of course) displays it correctly..... | 12:28.31 |
chrisl | Is Acrobat actually using that information? | 12:29.59 |
kens | THere is no AP, so Acrobat must be using the text at least, even if its not using the actual font | 12:30.28 |
| Hmm, actually the DroidSansFallBack font does produce some different result to the specified font, there's some kanji in there | 12:31.12 |
| But the latin is the same | 12:31.34 |
chrisl | I'll take a look - but I feel absolutely awful, so it'll have to wait until I've had a lie down..... | 12:33.25 |
kens | NP I'll mail you the modified files, don't look till you feel OK | 12:33.44 |
chrisl | That could be quite some time - I have Paul's cold.... squared :-( | 12:34.13 |
kens | Umm, bad news :-( | 12:34.31 |
chrisl | Yeh - back in a while | 12:34.42 |
| kens2: I believe I know the solution to this acroform problem..... | 14:24.15 |
kens2 | Oh really ? | 14:24.25 |
chrisl | Yeh. So, using Identity-UTF16-H is the right thing to do, *but* there's an issue in FAPI that needs addressed, too | 14:25.00 |
kens2 | Ah, that might explain things! | 14:25.22 |
| I oculdn't get the DroidSansFallback to work 'properly' either | 14:25.35 |
chrisl | I'm only looking at the DroidSansFallBack case just now | 14:26.32 |
kens2 | Yeah tat's the one to go for | 14:26.42 |
| The doc uses /Identity, and I doubt that will work with Unicode | 14:27.07 |
chrisl | Yes, and I'm not sure the stanv_font will work with UTF16 - but I guess we could try it | 14:27.59 |
kens2 | I'll give it a bash if you fix the fallback :-) | 14:28.12 |
chrisl | Sure, but I'm not sure of getting it done today | 14:28.51 |
kens2 | Not a problem | 14:28.58 |
chrisl | The problem being, I don't have direct access to the CMap at the point the decision needs to be made :-( | 14:31.17 |
kens2 | Ah..... well there's no rush, this is an old bug, but it does affect other stuff too (synthesis of FreeText in particular) | 14:31.49 |
henrys | mvrhel_laptop: where is the gsview code? | 15:09.09 |
| git-wise | 15:09.17 |
| I guess win_desktop | 15:10.10 |
mvrhel_laptop | henrys: yes win_desktop | 15:40.39 |
henrys | chrisl: I did experiment with a setting of MAX_CCACHE_TEMP_BITMAP_BITS and that did produce paths for the fonts, but only looked at a few files. That should fix some the xpswrite issues. | 16:10.16 |
| chrisl: setting it to 0 | 16:10.31 |
chrisl | henrys: we maybe want to control it via a spec_op? | 16:10.58 |
| kens: fapi fix pushed - with your hacks to pdf_ops.ps you should see the correct glyphs | 16:11.46 |
kens | Oh cool, thanks chrisl I'll chekc it out now | 16:12.01 |
henrys | chrisl: I suppose, what does NOCACHE currently do? | 16:12.18 |
chrisl | henrys: I have no idea..... it will *probably* render to a bitmap, and throw away the bitmap | 16:12.47 |
kens | is staggered by the size of the change..... | 16:14.02 |
henrys | chrisl: I know it did not produce an outline but didnât look beyond to see exactly what it does. | 16:14.08 |
chrisl | henrys: I can look at the NOCACHE behaviour on Monday, and have FAPI only produce outlines for that case | 16:14.16 |
| kens: you're also getting new cups | 16:14.31 |
kens | Yeah I see that on reading the commits | 16:14.41 |
henrys | chrisl: yes I was wondering if we should hijack that option it doesnât seem to really have much purpose as it is. | 16:14.52 |
kens | You want me to compile that as well ? | 16:14.53 |
chrisl | kens: that would be good, thanks | 16:15.09 |
kens | Doing it now I think :-) Will check the nmake options after its finished building | 16:15.27 |
chrisl | henrys: I've never understood what it was for, tbh..... | 16:15.39 |
henrys | chrisl: the docs say debugging - so I imagine it would indicate a caching bug if the files works with NOCACHE and doesnât otherwise | 16:16.47 |
kens | chrisl WITH_CUPS=1 compiled OK | 16:17.08 |
chrisl | kens: thanks | 16:17.24 |
kens | seems to have the drivers present too | 16:17.42 |
chrisl | henrys: but setting MaxFontCache or MaxFontItem to zero should disable the cache..... | 16:18.40 |
kens | chrisl and indeed if I use the fallback font with Identity-UTF16-H it does work. THe font embedded in the file doesn't, but I think that's basically because its a font which never will, not when its mapped using Identity-H | 16:21.08 |
| I'll htink about that tomorrow.... | 16:21.42 |
chrisl | kens: can you check in Acrobat what font it's actually using? | 16:21.53 |
kens | Not really, because its an annotation | 16:22.08 |
| Err form | 16:22.18 |
| I can't edit it, and don't really know what its 'supposed' to look like | 16:22.38 |
| But we know that Acrobat always synthesises appearances anyway | 16:23.43 |
| I know for sure its not using the DA, because if I hack teh font encoding, it doesn't change in Acrobat | 16:24.08 |
chrisl | Erm, well, even Acrobat pre-flight says the stanv_font is "missing", so it's clearly not using that! | 16:26.00 |
kens | :-) | 16:26.07 |
| I'm not certain what they are doing is legal | 16:26.21 |
| They seem to be inheriting the font from the page resources, | 16:26.39 |
| And it *is* defined there. | 16:26.50 |
| But yes, I'm reasonably certain Acrobat isn't using it. | 16:27.05 |
henrys | chrisl: looking that the definition it is just shorthand for 0 setcachelimit. | 16:27.31 |
chrisl | henrys: yes, setcachelimit sets MaxFontCache IIRC - it seems pointless to have it as a special command line param, though..... | 16:28.22 |
kens | is still feeling tired, going to head off now | 16:28.28 |
| Goodnight all | 16:28.35 |
chrisl | goodnight kens | 16:28.42 |
| henrys: in the fapi code, I specifically try to produce a bitmap whenever possible (even when not caching) so we avoid (as much as possible) having freetype rendering directly next to ghostscript rendering in the middle of a string. | 16:30.36 |
henrys | chrisl: yes I think NOCACHE is not going to describe well what we want to do which is produce an outline for glyphs if possible. | 16:32.00 |
chrisl | henrys: what I was thinking was that if the maximum cache size is zero, then I can always produce outlines (as opposed to an exhausted cache size >0) | 16:33.12 |
henrys | chrisl: Iâm fine with that - but we do need a better parameter name so it is clear to the user we are trying to produce outlines vs. uncached bitmaps | 16:34.29 |
| chrisl: of course we will have problems where hinting is required for correct output but I think weâll live with that. | 16:35.35 |
chrisl | henrys: the outlines we get from freetype are "hinted" so things like the dynalab fonts will work okay. In other cases, hinting is generally overrated...... | 16:36.58 |
| henrys: I'm tempted to configure this as a spec op during initialisation, and let the device decide - if the device wants outlines, we just do "0 setcachelimit" | 16:37.12 |
henrys | chrisl: fine by me. | 16:37.38 |
chrisl | I'll at it next week | 16:37.45 |
henrys | chrisl: I would just take out NOCACHE, just more nonsense for the uninitiated to trip over | 16:38.36 |
| -c can be used with the setcachelimit operaror | 16:39.21 |
| for command line setting | 16:39.28 |
chrisl | henrys: exactly my thoughts | 16:41.46 |
Tommm | Heys guys just started using ghostscript through Ghost4J and I'm trying to convert a pdf into high quality images but I'm unsure how to set the dpi to 300 using the interpreter parameters. I'm following the display callback example here http://www.ghost4j.org/coreapisamples.html | 16:59.33 |
| Currently using those parameters the images look really look quality so I'm trying to figure out which parameters I need to add/change to get these results | 17:00.43 |
chrisl | Tommm: we're not really able to help much with Ghost4J - it's not something we had any hands in | 17:01.38 |
| Tommm: however, if you add another argument and make it "-r300" that'll have gs produce 300dpi output | 17:02.34 |
Tommm | Looks like that did the trick for a small pdf with only 4 pages but its causing my application to crash on a larger PDF | 17:05.50 |
chrisl | Tommm: Well, Ghostscript is generally fine with many page files, so the problem is either in Ghost4J or in your app, or in how either of those ends up driving Ghostsscript | 17:07.59 |
Tommm | Ya seems like there's an issue with Ghost4J since I'm using their example verbatim but its causing Java to run our of memory... Any recomendations how I can still use GhostScript from a Java environment | 17:11.44 |
chrisl | Erm, not really, it's not something we've ever contemplated looking into. I'd try to talk to the Ghost4J devs about it. | 17:13.37 |
| Okay, heading off - good night/day everyone! | 17:46.32 |
tkamppeter | Robin_Watts, hi | 18:57.10 |
Robin_Watts | tkamppeter: hi. | 19:02.07 |
| sorry, I completely forgot about your patch. let me look now. | 19:02.19 |
| tkamppeter: In future, it would be good if you could use a git format patch. | 19:04.59 |
| That way we'd have a commit message. | 19:05.04 |
noob | HI guys sorry this is a really dumb question but I'm having issues running ghostscript from the command prompt on WIndows7 using the gs command any help? | 19:46.18 |
| I've added the following to my PATH C:\Program Files\gs\gs9.14\bin\gswin64c.exe; but typing gs in the command prompt tells me its unrecognized | 19:50.56 |
| Robin_Watts: any idea what I'm doing wrong? | 20:00.42 |
Robin_Watts | noob: At the command line type: echo %PATH% | 20:11.20 |
| Is gs really in C:\Program Files\gs\gs9.14\bin ? | 20:11.59 |
| oh, sorry, I see the problem. | 20:12.12 |
| You want to add C:\Program Files\gs\gs9.14\bin to your path. | 20:12.30 |
| And you start gs using gswin64c.exe, not gs. | 20:12.43 |
noob | OK just changed it to reflect that and I made sure its present there when doing an echo but using gswin64c.exe still doesnt work for me... | 20:18.33 |
| Robin_watts: also just restarted the command prompt just to be safe | 20:20.47 |
| Doing this works though "C:/Program Files/gs/gs9.14/bin/gswin64c.exe" but i would like to not have to type that whole thing each time | 20:22.07 |
Robin_Watts | noob: Can you paste the results of echo %PATH% into pastebin.com and give me the URL please? | 20:49.47 |
noob | +Robin_Watts: Here you go http://pastebin.com/dceVbwkJ | 20:54.35 |
Robin_Watts | I'm not sure the spaces in the path are healthy. | 21:14.07 |
| "; C:\..." | 21:14.26 |
| You should have ";C:\..." | 21:14.40 |
noob | Awesome! that made the difference! Thank you so much! | 21:23.34 |
Robin_Watts | fab. | 22:22.56 |
| Forward 1 day (to 2014/06/07)>>> | |