| <<<Back 1 day (to 2013/09/01) | 2013/09/02 |
svip | I know this channel is not about Adobe Reader, but can someone tell me why it renders Helvetica as dots? | 07:31.30 |
| I thought the PDF standard meant that if Helvetica was chosen, it would revert to some other sans-serif font. | 07:31.52 |
| (All other PDF readers I've tried works fine.) | 07:32.02 |
kens | svip most likely its using a custom encoding where the character codes do not map to glyphs in the font, so you get a 'notdef' | 07:37.07 |
| But I'd have to see the file to be sure. Its likely that the font is a *CIDFont*, not a 'Font', and as such you cannot use a Multiple Master replacement | 07:37.41 |
svip | kens: Well, when I start Adobe Reader up with the PDF, it says 'It cannot find the font 'Helvetica'.' and then shows dots. | 07:39.33 |
| And it's true, I don't have Helvetica on my computer. | 07:39.49 |
kens | If it was plain Helvetica you would get 'something'. | 07:40.07 |
svip | The font is not embedded either, because our customers complained about the file size. | 07:40.23 |
kens | Not unusual, but dumb customer move. | 07:40.42 |
| *always* embed fonts | 07:40.54 |
svip | I would too. | 07:40.58 |
| But they are the ones that pay us. | 07:41.09 |
kens | If you want me to look at the file you'll haveto send me a copy or give me a URL | 07:41.12 |
| You oculd also look at document properties (File->Properties) on teh fonts tab, and see what it says there | 07:41.56 |
svip | kens: It says CID. | 07:43.46 |
kens | THere you are then, its a CIDFont, not a Font. You can't substitute a Multiple Master for a CIDFont, so you get the Bullet fotn instead. | 07:44.12 |
| Not embedding a CIDFont is pretty much disastrous | 07:44.25 |
| Ghostscript uses the DroidSans font as a fallback which works, sometimes. | 07:44.51 |
| But there is absolutely no specified fallback mechanism for missing CIDFonts | 07:45.15 |
| You either need to embed the CIDFont, or remake teh PDF file so that it uses a Font instead of a CIDFont | 07:45.37 |
svip | How do I do the latter thing? | 07:46.17 |
| And - if I may ask a dumb question - what is the difference between CIDFonts and Fonts? | 07:46.29 |
kens | I can't tell you that, it depends too much on how you are making teh PDF file. | 07:46.39 |
svip | kens: Hmm... the export tool does not allow this functionality. It can only use CIDFonts. | 07:47.51 |
kens | svip I can't answer that indepth, its too complictaed, but essentially a Font is limited to a maximum usable space of 255 glyphs, with each glyph bein encoded by a single byte character code. A CIDFont may contain any number of glyphs and they are encoded with 1+ byte character codes, the highest number I've seen is 5 bytes, but there is no theoretical limit | 07:47.58 |
svip | OK. | 07:48.17 |
kens | So your choice is; 1 use a differnt workflow to create the PDF, 2 embed the CIDFOnt, 3 live with teh fact that its broken. | 07:48.52 |
svip | Seems like. | 07:49.31 |
kens | fingers crossed chrisl | 08:47.04 |
| Sneaing releases out on US Public Holidays seems like a neat idea :-) | 08:47.36 |
chrisl | kens: we'll see if it works...... | 08:57.32 |
kens | :) | 08:57.40 |
| Hmm, that's amusing. If you set interpolate to true for a imagemask, and a 1-bit image, Acrobat interpolates the mask, but not the image.... | 09:56.57 |
chrisl | Well, it is device dependent...... | 09:59.03 |
kens | I know, but different behaviour on 1-bit images and imagemasks is just weird | 09:59.40 |
chrisl | It sounds like an Acrobat "feature" to me! | 10:00.29 |
kens | Indeed, I wonder what it does with colour images, but I'm not bothered enough to construct an example | 10:01.00 |
Robin_Watts | kens: gs handles masks differently to images. | 10:03.09 |
kens | OMG more device weirdness in GS. If we get a type 1 image, we call _begin_image, which calls begin_typed_image, which calls _begin_image | 10:03.22 |
| Robin_Watts : I knwo | 10:03.27 |
Robin_Watts | which is a pain when you get a masked image cos gs splits it internally to a mask and an image. | 10:03.36 |
| kens: That sounds like the splitting I was just talking about. | 10:03.53 |
| tor7: hey. | 10:03.56 |
kens | Robin_Watts : it handles imagemasks (which are *just* a mask) differently | 10:04.00 |
tor7 | Robin_Watts: morning(-ish)! | 10:04.16 |
Robin_Watts | tor7: I was thinking about having a go at writing some code to spit out jni bindings for mupdf. | 10:25.09 |
tor7 | Robin_Watts: autogenerated? | 10:25.59 |
Robin_Watts | tor7: substantially, yes. | 10:26.12 |
tor7 | have a look at http://www.codenix.com/~tolua/tolua++.html if you haven't seen it | 10:26.51 |
| it's a code generator for lua bindings that looks pretty useful | 10:27.05 |
Robin_Watts | My vague plan (as far as I've got) would be to have an input file that describes each java class, with a list of members/native functions | 10:27.27 |
tor7 | also, I was thinking of creating a set of common wrapping functions for the common fz_devices so that normal users don't need to use the device set up api | 10:27.43 |
Robin_Watts | tor7: ok, but we'd want to offer the fully general stuff too, right? | 10:28.08 |
tor7 | Robin_Watts: I'd start by compiling a list of which structs and functions should be exposed. | 10:28.12 |
Robin_Watts | tor7: That's something that I hope would drop out of it ,yes. | 10:28.38 |
tor7 | yes, I think we should expose most of the API. if we could do it by auto generating lots of files, that'd be best for maintenance (and let us do more than just JNI) | 10:28.43 |
| I think we'd still need to write some of the bindings manually, for the core context stuff and setup | 10:29.24 |
Robin_Watts | tor7: quite possibly. | 10:29.36 |
| I'd figured that if I could get it to spit out the boring 90% of the code, writing the extra 10% by hand would be more managable. | 10:30.10 |
tor7 | fz_pixmap_from_page, fz_text_from_page, fz_display_list_from_page, fz_pixmap_from_display_list, etc were the functions I was thinking about writing | 10:30.18 |
Robin_Watts | tor7: makes sense. | 10:30.33 |
tor7 | Robin_Watts: yeah. you'll definitely want help if we want to expose more than a dozen functions. | 10:30.35 |
Robin_Watts | so I'll prod at this for a bit. | 10:31.05 |
| tor7: oh, I think all the commits on robin/master are ready to go now. | 10:31.36 |
tor7 | I'll cook up these functions and make our tools use them. It's that or implementing shadings in the opengl device, or implementing path rendering using trapezoids instead of the NV_path_rendering extension. | 10:31.58 |
| Robin_Watts: fab! I'll take a look then. there are a few on tor/master as well (but ignore the path one, it's still broken) | 10:32.23 |
Robin_Watts | will look at tor/master now. | 10:32.33 |
| tor7: The path one in tor/master is broken. | 10:32.58 |
| cluster test it, and you'll see lots of errors. | 10:33.18 |
tor7 | Robin_Watts: yeah, I know it is. I saw it when verifying some xps fixes in gxps when comparing to muxps on that branch. | 10:33.48 |
Robin_Watts | with the version one... | 10:34.46 |
| the version target updates the include/mupdf/fitz/version.h file ? | 10:35.14 |
tor7 | Robin_Watts: when doing banded rendering, if we fz_throw halfway through, do we remove the file? | 10:35.29 |
| Robin_Watts: yeah, "make version" updates the version.h file. not sure if or how we should automate that. | 10:35.48 |
Robin_Watts | tor7: no, the file is closed correctly, it's just not complete. | 10:35.49 |
| same as all the other methods, I believe. | 10:36.04 |
tor7 | Robin_Watts: s/do/should/ :) | 10:36.07 |
Robin_Watts | to7: We don't remove partial (unbanded) pngs etc in the event of an error. I'm tempted to leave it as is. | 10:36.44 |
| cos at least you can see how much you got :) | 10:36.48 |
tor7 | Robin_Watts: I puth the #ifndef FZ_VERSION in case we want to pass it on the command line as -D flags | 10:36.56 |
Robin_Watts | yeah, I saw that. | 10:37.05 |
| The top 3 commits in tor/master look fine. | 10:38.02 |
tor7 | Robin_Watts: I think the HSUBPIX and VSUBPIX defines can be deleted in "Be smarter when quantising" | 10:39.35 |
Robin_Watts | and QUANT | 10:40.27 |
| removed and repushed | 10:41.55 |
tor7 | Robin_Watts: they all look fine now | 10:45.21 |
Robin_Watts | Ta. | 10:45.32 |
tor7 | still no way to bypass the glyph cache though. | 10:45.37 |
| but that's not hugely important at the moment | 10:45.57 |
Robin_Watts | tor7: The render_glyph_pixmap functions avoid the glyph cache. | 10:46.11 |
| and the subpixel calculations have been pulled out into separate functions so that you can safely layer your own caching on top. | 10:47.04 |
tor7 | Robin_Watts: one thing we could do here is move the fz_subpixel_adjust to the callers instead | 10:47.14 |
| right! I see now. | 10:47.20 |
Robin_Watts | tor7: Having the subpixel adjust inside render_glyph seems nicer to me, cos it's all in one place. | 10:47.54 |
| but for render_glyph_pixmap, if you're doing caching, then it needs to be done before the call. | 10:48.14 |
tor7 | sure. | 10:48.31 |
Robin_Watts | so at the moment, it'd end up being done twice (no harm, except a slight waste of time). | 10:48.33 |
| we could pull it out later if we want to. | 10:48.41 |
tor7 | I'll make a mental note to do so when I merge the opengl stuff | 10:49.01 |
| now I need to get some lunch | 10:49.06 |
Robin_Watts | looks like ray broke gs. | 10:49.14 |
| oh, no, angstroms is full. | 10:49.26 |
| chrisl: do you have access to angstroms to fix it? | 10:49.40 |
chrisl | Robin_Watts: I don't think so, no | 10:49.55 |
Robin_Watts | and beards. | 10:50.00 |
| My mate at UndoDB just gave me a 30 day beta test license for the new ARM version. | 10:50.41 |
| just after I finish doing ARM stuff :( | 10:50.50 |
chrisl | Robin_Watts: where are the real hostnames for the nodes stored? | 10:59.03 |
Robin_Watts | chrisl: The cluster doesn't know where/what the nodes are. | 10:59.31 |
| The nodes connect into the cluster and say "Hi, I'm angstroms, got anything for me to do?" | 10:59.52 |
chrisl | I thought marcos kept a list somewhere on casper | 11:00.11 |
Robin_Watts | oh, well, possibly, but nowhere I know of, sorry. | 11:00.27 |
chrisl | Oh, well even if I had a login on the nodes, if we don't know the hostname......... | 11:01.04 |
Robin_Watts | chrisl: It might be a transient fault. Let me try readding that job onto the end of the queue. | 11:01.14 |
sanoj1 | chrisl: what was the URL for the font tarballs? http://ghostscript.com/doc/current/Fonts.htm refers bogusly to ftp://mirror.cs.wisc.edu/pub/mirrors/ghost/fonts/ | 11:06.50 |
| thanks for adding the explicit font licensing to LICENSE file | 11:07.32 |
sanoj1 | wanted to check if a tiny LICENSE file was added at font download dir too | 11:08.03 |
chrisl | sanoj1: the fonts are here: http://downloads.ghostscript.com/public/fonts/ - the archives there don't yet have the license info in them | 11:08.24 |
sanoj1 | thanks | 11:08.33 |
| chrisl: do you intend to add a LICENSE file or similar there too? I imagine simplest is to simply add such file separately from the zip files (assuming those zip files come straight from URW++) | 11:09.37 |
chrisl | sanoj1: yes, that's the plan. | 11:10.03 |
sanoj1 | excellent :-) | 11:10.12 |
| chrisl: CIDFont/ArtifexBullet contains a binary chunk - do true sources for that file exist somewhere? | 11:13.50 |
chrisl | sanoj1: no, it's just a CID font, you can open it in fontforge | 11:15.15 |
sanoj1 | ah, ok | 11:15.50 |
| chrisl: I susepct the (1-05) in lib/ghostpdf.ppd refers to font version and need updating | 11:22.23 |
| (1.05) | 11:22.35 |
| git grep -l \\b1\\\.05\\b | 11:23.08 |
chrisl | sanoj1: could be, I've no idea who maintains that, though..... | 11:23.55 |
sanoj1 | ...and in contrib/pcl3/ppd/fonts.ppd as well, but I guess contrib area is less of a concern to you | 11:24.25 |
chrisl | I guess I can update them - all this crap dotted around is a pain :-( | 11:25.18 |
sanoj1 | chrisl: as I understand it, Utopia is no longer part of Ghostscript - but is referenced at multiple places | 11:42.02 |
kens | has never heard of it | 11:42.12 |
sanoj1 | ...I seem to recall more fonts dropped | 11:42.22 |
kens | Hmm, Utopia appears to be a font from Tor | 11:43.10 |
tor7 | kens: Utopia is one of the fonts that Adobe donated to the X consortium in the early 90's | 11:43.30 |
kens | RIght I was just reading the licence | 11:43.40 |
chrisl | sanoj1: AFAIK, only the URW fonts have been shipped with Ghostscript, before settling on those, fonts were in a separate package | 11:43.55 |
tor7 | it has nothing to do with postscript, other than being from Adobe and being freely distributable | 11:43.58 |
kens | tor7 yes, I can't see any references to Utopia apart form where we say its a free font, I may have missed some | 11:44.31 |
sanoj1 | CharterBT, IBM Courier, and Utopia are all mentioned in History2.htm and are listed e.g. in Resource/Init/Fontmap.GS - together with Hershey also no longer included | 11:46.16 |
kens | Well, history mentiosn all sorts of stuff that is... historical.... | 11:46.38 |
sanoj1 | Resource/Init/Fontmap.GS | 11:46.44 |
kens | The demo fontmap contains that sort of thing too yes, its a demo | 11:47.01 |
sanoj1 | git grep -li hershey | 11:47.23 |
chrisl | sanoj1: but they were, AFAIK, never shipped with GS, and all will still work. Several fonts are referenced in Fontmap.GS that we don't and have never shipped..... | 11:47.38 |
sanoj1 | ok - just mentioning in case it made sense to cleanup | 11:48.05 |
| ...or in case you in fact assume some fonts being available via X11 (meaning that our packaging for Debian and other distros would need stronger ties to those fonts) | 11:49.14 |
| explicitly stripping "noise" might reveal if that was the case, by causing failures to your testsuite ;-) | 11:50.37 |
kens | All the various example Fontmap files contain example references to Utopia (and other fonts), if the font file is not present this causes a 'font not found' error. Ghostscript needs Helvetica as a minimum. | 11:51.53 |
| THe occurences in history*.htm reference ways to use Utopia (and friedns) by converting to other formats | 11:52.26 |
| Use.htm contains a copyu of the Sun Fotnmap example, which is the only reference there. | 11:52.47 |
| The font list in the PPD file is wrong. | 11:53.13 |
| (ghostpdf.ppd) | 11:53.50 |
| But it will not cause any problems | 11:54.09 |
sanoj1 | ok - thanks for checking! | 11:54.39 |
kens | We have specific substitutions in place (gs_fonts.ps) for all these fonts. | 11:54.49 |
| THe other occurences are white listing font embedding for known open-source fonts, even when the font flags forbid it. | 11:55.24 |
henrys | chrisl, Robin_Watts : I offered your git help to shelley but if I'm here I'll help I know you are busy with other stuff and getting ready for the trek. | 13:56.32 |
Robin_Watts | henrys: No, I'm happy to help him. | 13:56.47 |
chrisl | Sure thing. | 13:57.12 |
kens | spent most of Sunday updating laptop | 13:57.59 |
chrisl | I should probably do that, too..... | 13:58.28 |
Robin_Watts | Weather in chicago is HOT. | 13:59.02 |
kens | Oh had forgotten to look | 13:59.11 |
Robin_Watts | You may all have to live with the sight of me in shorts. | 13:59.17 |
henrys | Robin_Watts: awesome I was going to try out swimming in the lake. | 14:00.14 |
kens | crikey, i gets hotter as meeting approaches | 14:00.31 |
Robin_Watts | henrys: Are you staying there for the show? | 14:00.31 |
henrys | yup the whole thing | 14:00.45 |
Robin_Watts | henrys: Well, you have time then. It's a long walk back from the lake to the airport hotel in a wet wetsuit :) | 14:01.26 |
henrys | there's a beach and I assume lockers. | 14:02.39 |
| http://www.cpdbeaches.com/beaches/ohio-street-beach/ | 14:03.05 |
| btw today's a US holiday I'll be in and out. | 14:12.21 |
Robin_Watts | pictures henrys trying to get changed into a wetsuit in a locket. | 14:13.30 |
| locker. | 14:13.32 |
| For the non-facebookians amoungst you: http://www.wss.co.uk/photos/namibia2013/ | 14:14.44 |
kens | thanks Robin_Watts | 14:15.39 |
sebras | Robin_Watts: yey! I'll have a look. :) | 14:17.21 |
kens | yikes lions.... | 14:18.16 |
tor7 | and giant beetles! | 14:18.38 |
kens | who is the lady with the large bird ? | 14:20.44 |
henrys | Robin_Watts:the wetsuit goes over the swimsuit so I can put that on anywhere, but honestly I'm not sure about logistics, these guys do it http://www.chicagotriclub.com/inside.aspx?uid=110 so it's not a new problem to be solved. | 14:23.20 |
Robin_Watts | kens: Random silly girl on the boat with us. | 14:31.11 |
kens | ah :) | 14:31.22 |
Robin_Watts | Crumbs: http://www.topgear.com/uk/car-news/video-boulder-nearly-smashes-car-2013-09-02 | 16:34.26 |
kens | Was on the BBC web site the other day | 16:34.40 |
Robin_Watts | henrys: We should discuss lcms and it's threading problems at the staff meeting. | 17:19.15 |
| (essentially, I think we need to disable lcms ever being used as a shared lib) | 17:19.40 |
henrys | okay adding it. | 17:20.15 |
kens | And I see it, that was quick | 17:21.29 |
henrys | Robin_Watts: it's under gs/color | 17:21.32 |
Robin_Watts | cool. | 17:22.22 |
henrys | kens:yeah that part is nice the major gripe I have with this system is easily seeing what's be recently added. workflow brags about their simplicity but somethings are too simple. | 17:22.53 |
Robin_Watts | henrys: Can you color code stuff? | 17:23.22 |
kens | Well its small enough that I just read it all every time | 17:23.29 |
Robin_Watts | i.e. could you 'select all' then "set color to grey" after each meeting? | 17:24.05 |
| that way new stuff would appear in black? Or maybe we should add new stuff in red, or something? | 17:24.22 |
| or have a 'color of the release' :) | 17:24.30 |
henrys | Robin_Watts: I can tag it and # will autocomplete the choices in the search. | 17:25.39 |
Robin_Watts | yeah, I was wondering about something orthogonal to the # stuff. | 17:26.08 |
henrys | kens:maybe if it is to big to digest in one swallow we should stop adding stuff , so maybe it's best left as is. | 17:26.59 |
Robin_Watts | Everything added for this staff meeting could have a light green background? Then everything added for the next one could have a light blue background etc... over time we'd get to see what was old that way... | 17:27.16 |
henrys | Robin_Watts: the sensible feature here is a mouse over pop up "date added" and "completed" if your showing completes. | 17:28.12 |
Robin_Watts | henrys: Can workflowy do that? | 17:28.30 |
henrys | no but I've requested it. I do get an email of all changes each day so I have the data. | 17:29.23 |
| Forward 1 day (to 2013/09/03)>>> | |