| | <<<Back 1 day (to 2015/04/01) | 20150402 |
| Robin_Watts | tor8: For the logs. I have 2 commits on robin/master that add rudimentary dropout avoidance to the mupdf renderer. | 00:00.52 |
| | There will shortly be a bmpcmp run you can look at too. | 00:01.08 |
| fredross-perry | new gsview builds available. Fixed bugs reported today by chrishl and henrys. | 00:39.37 |
| marcosw1 | Robin_Watts: I don't understand the point of the upload directories. | 05:40.48 |
| kens | Well, that file from Marcos is invalid, it has nested BT operators.... | 07:15.16 |
| chrisl | Which file? | 07:15.54 |
| kens | It looks to me like that's what's causing the problem. | 07:15.54 |
| | The bug assigned to you :-) | 07:16.02 |
| | #695897 | 07:16.05 |
| | You'd probably better keep on ignoring it | 07:16.23 |
| chrisl | Ah, I had a quick look last night, but I was too tired. I thought we handled nested BT/ET? | 07:16.39 |
| kens | Well, we do, in as much as we don't throw an error, but I'm going to have to read the code in detail to see what exactly we're doing# | 07:17.05 |
| | But the missing sections are in portions following a BT/BT/ET sequence | 07:17.27 |
| | Yeah if I knock out the second BT at least some of the missing text comes back | 07:18.14 |
| | I'd guess htis has been edited by some dumb tool to add the bold text, and it just squirts it straight into the content stream wihtout regard to what's in there. | 07:19.11 |
| chrisl | Interesting, I thought a BT inside a BT/ET block was a null op, and an ET outside a BT/ET block was a null op.... I guess not | 07:19.53 |
| kens | Don't know yet, I'mjust removing the extraneou BT ops to make sure that's the problem | 07:20.18 |
| | H, IT LOOKS LIKE THERE ARE EXTRANEOUS et OPS AS WELL. | 07:21.21 |
| | oops sorry caps | 07:21.27 |
| chrisl | So pretty broken then..... | 07:22.37 |
| kens | Yeah, pretty much. Of course 'Acrobat reads it'..... | 07:23.00 |
| | Ah, apparently creaated by iTxtSharp 5.0.6 but I suspect actually something *using* iTextSharp since the application is 'Mobile Service invoice generator' | 07:24.06 |
| | And teh current version is 5.5.5 anyway | 07:24.44 |
| chrisl | We've seen PDFs from iTextSharp before and I don't remember them having this problem.... | 07:24.58 |
| kens | Maybe the application uses it properly :-) | 07:25.19 |
| chrisl | I guess stranger things have happened.... | 07:25.43 |
| kens | This genuinely looks like something has injected chunks of BT/ET text blocks into an existing content stream | 07:25.45 |
| | The first one is "BT 10 832 Td ET" which is spectacularly pointless | 07:26.10 |
| chrisl | Yeh, I did notice that | 07:26.44 |
| kens | I can't remember what ops are valid inside a BT/ET block, and this file appears to be setting all kinds of graphics state inside. | 07:27.42 |
| chrisl | Ah, when we hit a nested BT we execute a ET - that seems bad | 07:27.46 |
| kens | I'm not entirely sure what else we can do | 07:28.02 |
| chrisl | Just ignore it, I'd think | 07:28.17 |
| kens | I'd have thought the outcome would be the same | 07:28.28 |
| chrisl | Oops, looks like I left some debug code in there..... | 07:29.03 |
| kens | :-D | 07:29.08 |
| | OK so if we *ignore* the nested BT then we won't initialise the text matrix, which we are specifically supposed to do in response to a BT. | 07:30.21 |
| | Which is probably why we do an ET then process the BT | 07:30.43 |
| | Of course the spec simply says 'you can't do that' | 07:31.07 |
| | Oh OK its legal to have graphics state changes inside a BT/ET that's OK then | 07:32.18 |
| chrisl | I'm baffled as to why we don't print a warning/error on this - it looks like we should | 07:33.54 |
| kens | It owuld be a good idea if we did | 07:34.07 |
| chrisl | We're not going through the code to handle a nested BT..... | 07:35.14 |
| kens | O.O | 07:35.19 |
| chrisl | But we *are* seeing spurious BT and ET ops, so..... | 07:36.49 |
| kens | Hmm, if I remove the first nested BT then I get a warnign about spurious ET operators, and although some text apperas, it is incorrectly placed. | 07:42.03 |
| | MuPDF manages to read it fine :-( | 07:42.45 |
| chrisl | So, I'm going to leave this to you, unless you want me to take it | 07:43.06 |
| kens | No it hsould have been assigned to me in the first place | 07:43.19 |
| chrisl | Well, I am capable of hacking on the pdf code, when required.... but no point both of us suffering ;-) | 07:44.02 |
| kens | Its supposed to be mine, sadly :-( | 07:44.18 |
| | If you tell me where that debug code is I can remove that if I come up with a fix | 07:44.37 |
| chrisl | I just removed it | 07:45.39 |
| kens | Ah, OK | 07:45.44 |
| chrisl | I had another typo commit to push - we shouldn't have the string "c*nt" in the sources, really....... | 07:47.15 |
| kens | Err, no, that's not ideal | 07:47.32 |
| pipitas | kens: Have you seen my question from last night, 21:50:13 h? (About 12.5 hrs ago) | 08:27.48 |
| kens | pipitas, no sorry | 08:29.27 |
| | On IRC ? | 08:29.34 |
| | pipitas, chrisl is the font expert. However you can load a TrueType COllection into Ghostscript. | 08:32.57 |
| | Why would you want to execute PostScript to load it ? | 08:33.46 |
| pipitas | kens: I want to use a simple command line which loops through a series of TTC fonts to write a one line sentence (in PDF) for the font sample⦠| 08:42.48 |
| kens | Well, that's not really how the font loading code is intended to work. It might work if you try it, I have no idea | 08:43.28 |
| | I'm presuming you are asking because it doesn't work ? Or have you not tried ? | 08:43.49 |
| pipitas | (kens: I've however also failed to get it to work after creating a customized fontmap that I load with '-sFONTMAP=mymap'.) | 08:43.51 |
| kens | If you want to load a TTC I think you may have to use a CIDFont | 08:44.15 |
| | Not a font | 08:44.19 |
| pipitas | The command line I gave didn't work (of course) â I just gave it as an example about how I image the lines along it COULD be working... | 08:44.58 |
| | Ah⦠CIDFont may be the magic word⦠| 08:45.19 |
| | (Back to trying itâ¦) | 08:45.29 |
| kens | I'd start by using cidfmap to load it as a CIDFont, that *should* work. When you get that correct then you'll have the correct runes to try from PostScript. WHich *migth* work, or might not. Neither would surprise me | 08:46.17 |
| | Well, somehow executing 'q' is clearing the TextSaveMatrix from the current dictionary <boggles> | 08:49.32 |
| | Ah, if we get a '/q' then we do a gsave and '//nodict begin', which of course means that 'currentdict' has no TextSaveMatrix present..... | 08:55.56 |
| chrisl | pipitas: you can't loop through the font indices in a TTC in Postscript (at least, not without getting horribly embroiled in Ghostscript internals) - PS doesn't understand TTCs | 09:09.51 |
| | It would probably be good to augment the Fontmap code to allow TTC entries to be loaded as font substitutes | 09:17.12 |
| | ... at some point | 09:17.21 |
| pipitas | â¦at some point :-) | 09:19.46 |
| | chrisl: Should I submit a feature request in Bugzilla so it doesnt get forgotten? | 09:20.13 |
| chrisl | pipitas: sure, and please make sure it get assigned to me..... | 09:20.38 |
| pipitas | chrisl: so when "looping" through a TTC doesnt work â there is also no possibility to directly access a single TTC resource/font from PostScript? | 09:23.00 |
| | chrisl: But going the indirect path, creating a cidfmap should work? | 09:23.02 |
| chrisl | pipitas: as I said, the issue is that Postscript doesn't understand TTCs (or TTFs, for that matter), so from Postscript there's no access to the "internals". Going through the cidfmap will work, as it pulls the TTF out of the TTC, and makes it into a CID Type 2 font | 09:25.22 |
| kens | Oh ffs | 09:28.19 |
| | When we do a gsave we make 'nodict' the current dictionary, and that's set to be 'executeonly' so I can't store things in it. WHat a pile of cr*p this interpreter is | 09:28.59 |
| chrisl | kens: .forceput | 09:29.22 |
| kens | I know, but its just annoying | 09:29.30 |
| | ANd I wonder why this doesn't fil generally | 09:29.37 |
| | fail* | 09:29.42 |
| | If I do a gsave then do a BT, how is the current dictionary not 'nodict', and BT does a def, so why does that not fail ? There must be more black magic going on somewhere | 09:30.20 |
| | Oh, actually nodict is readonly, but it has a self key which is executeonly | 09:32.26 |
| pipitas | chrisl: Is it possible to amend 'lib/mkcidfm.ps' so it works for Mac OS X as well? Or should this rather go into a separate 'mkosxcidfm.ps' file derived from the existing for-Windows code? | 09:41.25 |
| chrisl | pipitas: as far as I can see, the only thing that would need changed would be the directory separation definition..... | 09:43.06 |
| | We could easily allow the directory separator to be optionally set on the command line, just like the path to the fonts | 09:44.34 |
| | pipitas: having said that, the utility of mkcidfm.ps is much more limited now that we have the droidsansfallback substitution in place | 09:47.30 |
| pipitas | chrisl: "directory separator" â you mean for Win it's ";" and for Unix/OSX it's ":" ? | 09:48.32 |
| chrisl | No, on windows it's "\" and on OS/X it's "/" | 09:48.59 |
| pipitas | chrisl: Ah, I see it means "/". | 09:48.59 |
| | chrisl: but it's using "/" already in mkcidfm.ps | 09:49.38 |
| chrisl | Oh, so it is - well, there you go, there's no reason it won't work then | 09:50.07 |
| pipitas | chrisl: I just tested it, but it only creates an empty file (with some commented header lines). Probably because of the hardcoded font names inside the file? OSX has other font names⦠(That's why I asked if it should rather be done as a separate 'mkosxcidfm.ps') | 09:54.41 |
| chrisl | pipitas: Frankly, I doubt we'll be able to spend the time checking through the fonts on a "normal" OS/X install to find those compatible with various glyph orderings. | 09:56.00 |
| pipitas | chrisl: Oh⦠I wasn't know it would require so much manual work. But then, I've not much clues about the "various glyph orderings" either. | 10:02.12 |
| chrisl | pipitas: there's no reliable way to automate checking the glyph ordering, unfortunately | 10:03.06 |
| pipitas | chrisl: I'd be willing to help with that "manual work", once I have enough clues about how to do it, or if I get some hand-holding to start it | 10:03.08 |
| chrisl | pipitas: TBH, I'm not even sure where I would start doing it myself, never mind guiding someone else through it | 10:04.44 |
| pipitas | chrisl: So if I have "Avenir.ttc", and I know it contains 12 TTF fonts â how do I use the 9th of these inside a PS file? I tried to enter a line "/Avenir-09 <</FileType /TrueType /Path (/System/Library/Fonts/Avenir.ttc) /SubfontID 8 /CSI [(Unicode) 0]>>;" into an avenir.map file which I put into "Resource/CMap/" and using "/Avenir-09-CMap findfont 180 scalefont setfont" in PS code, running "gs -sFONTMAP=â¦". But it returned: "Erro | 10:13.08 |
| | Operand stack: | 10:13.08 |
| | Avenir-09-CMap" | 10:13.08 |
| chrisl | It's loaded as a CIDFont, you have to use it as such | 10:14.08 |
| | pipitas: So, you'll have to compose the CIDFont with a CMap, then scale and set it. You might get away with doing something like "/Avenir-09-Identity-H findfont 180 scalefont setfont" | 10:25.12 |
| | Although, I would be tempted to name the CIDFont "/Avenir09" or "/Avenir_09" as the dash in the name might confuse the name splitting code | 10:26.20 |
| t4nk575 | hello, i just wanted to ask if the release of jbig2dec (http://www.ghostscript.com/jbig2dec.html) has symlinks intentionally or? | 10:42.10 |
| | i'm working in MSYS2 and get "cannot create symlink" errors | 10:42.45 |
| chrisl | t4nk575: they were there intentionally, sort of, but I think the next release will not have them | 10:44.56 |
| t4nk575 | oh, ok | 10:45.34 |
| chrisl | Oh, unfortunately, some are still there :-( | 10:45.45 |
| t4nk575 | ya :/ | 10:45.54 |
| chrisl | Could swear I changed that.... | 10:46.16 |
| | t4nk575: I'll fix it for the next release | 10:47.02 |
| t4nk575 | chrisl: alright :-) | 10:49.02 |
| chrisl | t4nk575: do you have the various autotools installed? | 10:49.25 |
| t4nk575 | yes, i was advised by a few people in #mingw-w64 (which i'm using to compile this) to run "autoreconf -fi" | 10:49.55 |
| | i'm not that experienced with autotools, but it worked and compiled fine | 10:50.03 |
| chrisl | Ah, okay, cool. For some reason I've never understood, automake defaults to using symlinks instead of copying the actual files - which I think is just bonkers.... | 10:50.50 |
| kens | Oh boy, not only does this thing nest BT and ET, it also does a Do inside a BT/ET block | 10:51.49 |
| t4nk575 | chrisl: chrisl: oh well :-) | 10:53.58 |
| kingihai | Hi there.. | 10:55.15 |
| Robin_Watts | kingihai: hi. | 10:55.56 |
| kingihai | Hi there..Robin! | 10:56.07 |
| Robin_Watts | tor8, paulgardiner: kingihai opened bug 695896 yesterday. | 10:56.31 |
| | It's about wanting to use PDFs to submit form data. | 10:56.44 |
| kingihai | Yes.. | 10:57.02 |
| Robin_Watts | It looks like it relies on various javascript things like SOAP, and NET.Http etc, as well as being able to 'getDetails' from the form fields for submission. | 10:57.21 |
| | I've said that it's not on our immediate roadmap for MuPDF, but he might be interested in pursuing it as a commercial thing. | 10:58.09 |
| | (i.e. paying for our time). | 10:58.15 |
| | Do we have even a ballpark idea of how much effort that is? | 10:58.30 |
| kingihai | OK.. right... | 10:59.04 |
| | firs of all.. | 10:59.08 |
| | It is possible? | 10:59.17 |
| | Adobe Reader can not in Android device | 10:59.42 |
| | QPDF is possible.. I knew.. | 11:00.21 |
| | I have to go out for my bussiness, but I want to know your answers.. | 11:02.03 |
| Robin_Watts | kingihai: Leave the window open, and I'l sure paul and tor will reply when they get a mo. | 11:02.23 |
| kingihai | Thanks.. good Idea... | 11:02.44 |
| | and will come again after 3-4 hours.. | 11:03.19 |
| | In case something wrong, I want to give my email... kingihai@hotmail.com | 11:04.52 |
| paulgardiner | kingihai: you'd need to add the submit method to our JS dom. You can see what we currently support by looking at declare_dom in source/pdf/js/pdf-js.c. The submit method would need to walk the form object tree to pick up the field names and valuse to send. | 11:13.16 |
| | There are several methods that can be used to send data html or FDF. You'd need to wrap up the names and values appropriately for which ever method is required. | 11:16.10 |
| | There's a good deal of work involved. | 11:16.34 |
| Robin_Watts | You mean "submitForm" ? | 11:21.38 |
| | this.submitForm({ cURL: "http://whatever", cSubmitAs: "HTML" }); ? | 11:22.15 |
| paulgardiner | Yesm that's the one | 11:23.14 |
| Robin_Watts | The example PDF looks to be getching data using SOAP too. | 11:23.36 |
| tor8 | I think we'd want to delegate actual socket/network code to the client | 11:24.36 |
| | since that sort of stuff really should be non-blocking/async/multithreaded | 11:24.48 |
| Robin_Watts | tor8: That's been our approach so far to such things, and it seems to have worked well. | 11:25.08 |
| | Can that be hidden as a javascript level interface? | 11:25.26 |
| tor8 | mujs can call C functions | 11:25.42 |
| | so we just need to implement the JS hooks to the client hooks | 11:26.08 |
| | hooks everywhere! | 11:26.11 |
| Robin_Watts | OK, but it would be specific to the mujs integration with mupdf, rather than a generic interface that any javascript integration would benefit from. | 11:26.26 |
| tor8 | I think it'd go through our generic javascript interfaces | 11:26.39 |
| | if the submitForm call is a function we haven't implemented yet | 11:26.56 |
| paulgardiner | JS hooks talking to client hooks is scary. We have cases of it already, but it requires storing messages from the thread on which mupdf runs, and having the client pick them up | 11:27.10 |
| | We've done it before so that JS can invoke alert dialogs | 11:27.56 |
| Robin_Watts | AIUI submitForm would be a java function that would talk to MuPDF to get the data out (that would go through our generic interface, AIUI). It would also have to talk to network stuff. | 11:28.19 |
| | That talking to network stuff, I would have hoped need not involve MuPDF at all. | 11:28.32 |
| | or am I showing my ignorance here? | 11:29.17 |
| paulgardiner | Actually it may not be as bad as the alert dialog stuff: that required the JS to wait for the user's input | 11:29.28 |
| | All depends on whether submitForm needs to return a success/failure result | 11:31.02 |
| Robin_Watts | runs. 3 months into this year, and I have now fallen flat on my face 3 times. | 11:32.50 |
| tor8 | Robin_Watts: only 9 more times to go! ;) | 11:33.14 |
| | Robin_Watts: one more mingw fix on tor/master | 13:04.57 |
| | zooming with the mouse wheel, I tried it on my windows box and the shortcut seems to be shift+mwheel | 13:05.26 |
| | shouldn't it be ctl+mwheel to match web browsers? | 13:05.39 |
| Robin_Watts | tor8: Did you try the latest version? | 13:09.58 |
| | It should be ctrl+mwheel to zoom. | 13:10.16 |
| | shift+mwheel pans left/right | 13:10.24 |
| | seems to work for me. | 13:11.10 |
| | https://www.indiegogo.com/projects/ergodox-ez-an-incredible-mechanical-keyboard | 13:11.27 |
| | Tilts the wrong way as far as I'm concerned. | 13:11.34 |
| | Did Henrys get his new keyboard in the end? | 13:11.45 |
| henrys | Robin_Watts: nah I'm stickin' with this one. If I'm typing people ought to be awake anyway ;-) | 13:16.51 |
| tor8 | Robin_Watts: I did. ctl+mwheel did nothing (it just flipped pages like regular mwheel) and shift+mwheel zoomed | 13:19.06 |
| | might be a difference between mingw and msvc? | 13:19.12 |
| | though I somehow doubt that | 13:19.19 |
| Robin_Watts | tor8: Don't see how. | 13:19.23 |
| | what SHA are you on? | 13:19.26 |
| tor8 | origin/master (e8a8d7e) | 13:19.47 |
| | hang on, let me rebuild just in case | 13:20.00 |
| | yeah, sorry. false alarm. | 13:20.49 |
| Robin_Watts | phew :) | 13:21.01 |
| tor8 | Robin_Watts: matrix layout looks neat, but ugh for tilt | 13:21.24 |
| | ...and I'm rather partial to Topre switches. cherry switches are overrated. | 13:21.46 |
| Robin_Watts | The tilt is the killer thing for me. | 13:24.26 |
| henrys | tor8: when you get a chance could you have a look at sodapdf and see if there are any IP issues? I didn't see anything other than jasper and freetype use but the outfit looks suspicious. | 13:52.04 |
| | kens, chrisl : pipitas bug could be bountiable I wonder if he'd work on it if paid? | 14:09.08 |
| kens | Which bug ? The documentation ? | 14:13.42 |
| | Or 'augment teh fontmap... ' ? | 14:14.27 |
| | Because I doubt he would want to do the latter, and I doubt we would want him to | 14:14.51 |
| | The former looks more like a 'Read the PLRM' to me, but I'm currently beating my head off a PDF file and don't have time to look at it | 14:15.28 |
| chrisl | Using TTCs in the cidfmap is documented, and he appears to have understood it well enough to create the mapping file. It's not our place to document a) the difference between fonts and CIDFonts, nor b) how to use a CIDFont....... | 14:17.46 |
| kens | That's what I meant by 'read the PLRM' | 14:18.01 |
| pipitas | kens: I'll try to read again the relevant parts of the PLRM and see if I understand better this time. | 14:23.41 |
| kens | look for 'composefont' | 14:23.51 |
| | See PLRM p94 | 14:24.19 |
| pipitas | kens: you are right â "augment the Fontmap" is way beyond my skillset | 14:24.49 |
| kens | I suspected it may not be soemthing you would want to do, it would mean messing inside Ghostscript, and few people would want to do that.... | 14:25.17 |
| pipitas | a documentation snippet I could contribute easily, once I figured it out myself⦠| 14:25.32 |
| Robin_Watts | tor8: That commit looks good to me. | 14:55.49 |
| kens | network problem, going to reboot | 15:14.07 |
| Robin_Watts | tor8: 3 commits on robin/master | 15:15.41 |
| | 1 is trivial. | 15:15.48 |
| | The next 2 add rudimentary anti-dropout to the mupdf renderer. | 15:16.06 |
| | For axis aligned rectangles, we force them to take up a full subscanline. | 15:18.03 |
| | The tests show many AA differences. | 15:18.27 |
| | but none that I've spotted as being visible yet (except for the odd few which are clear progressions) | 15:18.51 |
| chrisl | Well, the cidfmap docs seem pretty clear to me, and the example records make it pretty clear what's going on. Perhaps we should remove the FAPIFontmap and FAPIcidfmap stuff, now | 15:20.30 |
| henrys | kens, chrisl: I was thinking documentation work. In fact I assumed there was a solution to the problem as I said in the logs last night. | 15:21.42 |
| kens | henrys, PostScript doesn't support TreuType fonts or collectiosn directly at all, so no there isn't really | 15:22.11 |
| | In order to mess with these, you need to use Ghostscript's internals, and to be honest, the answer there is 'don't even try' | 15:22.33 |
| | As chrisl pointed out, there is documentation for using cidfmap, and pipitas has (I htink) beenable to use that. Once a TT/TTC is loaded as a CIDFont then you use it as any other CIDFont, its not really our place to be providing a PostScript tutorial on how to use COIDFonts. | 15:23.32 |
| henrys | I guess it is pretty well layed out in the CIDFontSubstitution section. I wish we had something akin to the FAQ question: How do I specify custom substitutions on Windows? That looks very straightforward and simple for ttf. | 15:33.12 |
| chrisl | henrys: I'm not sure how we can make it much simpler than it already is..... I can put it in the FAQ, but it will be the same information, and the same examples | 15:34.51 |
| | And nobody reads the FAQ anyway...... | 15:35.09 |
| kingihai | Came back.. | 15:40.49 |
| henrys | chrisl: fair enough. | 15:42.29 |
| Robin_Watts | kingihai: Tor and paulgardiner both answered the questions above. It's a significant amount of work. | 15:43.43 |
| | I guess the way to proceed here is for you to talk to sales@artifex.com about what you need. | 15:44.08 |
| kingihai | Sorry about it.. my connecting was closed.. | 15:44.27 |
| Robin_Watts | They'll consult with us to come up with a figure for how much it costs. | 15:44.33 |
| | kingihai: It'll be in the logs. There is a link at the top you can click to read back. | 15:44.52 |
| | http://ghostscript.com/irclogs/current.htm | 15:45.15 |
| | http://ghostscript.com/irclogs/current.html | 15:45.30 |
| | (sorry) | 15:45.35 |
| kingihai | not working... my computer has some data till 22: 24 ..this is last message ==> [22:24] <+Robin_Watts> The tilt is the killer thing for me. | 15:47.22 |
| kens | Logs are here: | 15:47.48 |
| | http://ghostscript.com/irclogs/current.html | 15:47.50 |
| Robin_Watts | kingihai: shift reload the page. All the logs are there. | 15:48.23 |
| kens | I checked, the logs are all present, ghostbot didn't die or anything | 15:49.00 |
| kingihai | OK something opened.. Thanks a lot .. I got it | 15:49.11 |
| | HaHa...OK | 15:49.28 |
| Robin_Watts | tor8: OK, tests passed this time on the antidropout stuff. Some interesting progressions on page 4 of the bmpcmp stuff (I ran bmpcmp with -t 33 to suppress most of the AA changes). | 15:50.04 |
| kens | Time to go, bye folks | 16:02.57 |
| kens | just saw htis: | 16:04.52 |
| | http://www.theregister.co.uk/2015/04/02/android_apps_to_run_on_linux_windows_and_macs/ | 16:04.52 |
| | maybe we should try Smart Office on it :-) | 16:04.52 |
| kingihai | I read that, but I have a question, your JS dom is different with Adobe's one? | 16:10.02 |
| Robin_Watts | Our JS DOM is an incomplete version of Adobe's one. | 16:10.22 |
| | Part of the work that needs to be done is for the missing parts to be written. | 16:10.39 |
| | We've only done as much as we needed. | 16:10.48 |
| | tor8: 4 commits on robin/master | 16:12.53 |
| | mvrhel_laptop: You about? | 16:13.03 |
| | mvrhel_laptop_: Hey | 16:17.52 |
| | I've just done a commit to robin/master to give a nicer API for you to call for the pdfclean stuff. | 16:18.10 |
| | http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=4341f019b5bb5499821544209602078a28075b4b | 16:18.31 |
| | It needs to be reviewed before it goes in, but does it seem reasonable to you? | 16:18.50 |
| | fredross-perry: Do you offer the same pdfcleaning options as mvrhel_laptop does in gsview ? | 16:19.51 |
| fredross-perry | I donât think so. Let me see what heâs got. | 16:20.18 |
| Robin_Watts | kens: Do you remember a bug about it taking a very long time for gs to extract page bboxes from a million page file? | 16:29.49 |
| pipitas | chrisl: I just saw you dismissed my request for documentation improvement on the grounds that it it seems "fairly clear" to you :( | 16:32.35 |
| | chrisl: Ok, I'll try again â maybe I'm in a really dump phase of mind right now, and my brain will work better over the weekend. | 16:32.37 |
| | chrisl: I hope you'll accept a patch for the documentation (or the FAQ) once I successfully figured out what right now fails for me⦠| 16:33.27 |
| chrisl | pipitas: I'm not sure how I can make it much clearer, particularly given the several examples..... | 16:33.55 |
| | If I follow the steps outlined in the documentation, I get a working cidfmap file | 16:34.22 |
| fredross-perry | robin_watts: are you referring to âadvanced conversionsâ? | 16:36.46 |
| pipitas | chrisl: To me, the examples given seem to deal with CJK, and this confuses me. (Me, I just want to use one of the TTC component fonts to print ASCII characters onlyâ¦) | 16:38.21 |
| | chrisl: I'll try again over the weekend with a fresh mind â right now I'm maybe too confused about all the different paragraphs I've read and can't tell apart... | 16:39.14 |
| | chrisl: Thanks for considering and testing to make sure it worked for you. | 16:39.43 |
| | chrisl: I'll get back if I keep failing⦠:-P | 16:40.07 |
| Robin_Watts | malc_: I have a vague memory of you pointing out a problem to me the other day... but I can't remember what it was... | 16:40.38 |
| chrisl | pipitas: I can easily add an example with "Identity" rather than "Japan", for example, but I don't really see how that will clarify anything | 16:40.52 |
| Robin_Watts | fredross-perry: dunno. I just know that mvrhel calls pdfclean. | 16:41.01 |
| | Presumably to linearise or to extract pages from documents. | 16:41.12 |
| kingihai | I have to go out, I sent email to sales@artifex.com , I wish to meet Robin again.. bye | 16:42.32 |
| fredross-perry | Robin_Watts: yes, to liearize. I do that too. Is there an issue? | 16:45.22 |
| | *linearize* | 16:45.29 |
| Robin_Watts | fredross-perry: It depends how you do it. | 16:45.45 |
| | linearisation can be done just by calling fz_doc_write. | 16:46.13 |
| | But if you want to extract specific page-ranges from a document, you need to call pdfclean. | 16:46.31 |
| | mvrhel_laptop is doing that in a slightly nasty way, and I've just prepared a commit that should let him do it more neatly. | 16:46.56 |
| | If you're not offering that ability yet, then no problem. | 16:47.21 |
| | If you're just calling fz_doc_write (or pdf_doc_write etc) then don't worry. | 16:47.50 |
| malc_ | Robin_Watts: uhm... neither do i to be honest... | 16:47.58 |
| Robin_Watts | malc_: Excellent. Then I shall assume I fixed it, or dreamed it. | 16:48.28 |
| malc_ | Robin_Watts: "i" still have a problem but tor wanted to "fix" it not you | 16:49.12 |
| Robin_Watts | what problem was that ? | 16:50.14 |
| fredross-perry | Robin_Watts: calling pdfclean_main (from mupdf/source/tools) with -l, two filenames, and a page list. | 16:50.17 |
| Robin_Watts | fredross-perry: Ah, right. | 16:50.43 |
| | That's what I hope to change. | 16:50.48 |
| | See: http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=4341f019b5bb5499821544209602078a28075b4b | 16:51.03 |
| malc_ | Robin_Watts: pdf_lookup_page_obj being on the slow side of things | 16:51.22 |
| Robin_Watts | You'll call pdfclean_clean instead. | 16:51.26 |
| | malc_: Ah! Yes. | 16:51.32 |
| | (the old way will still work, but the new way avoids the need to build a fake command line and parse it) | 16:52.13 |
| fredross-perry | Robin_Watts: as you wish. | 16:52.38 |
| malc_ | uh... framboy | 16:53.14 |
| | farm rather | 16:53.19 |
| henrys | Robin_Watts: if you run in a group it cushions the falls ;-) | 16:54.32 |
| Robin_Watts | more people to laugh at me. | 16:54.47 |
| henrys | Robin_Watts: it could be your form, I learned a lot from the Pose Method. Might be scary because "falling" is a big part of it, but I think it's improved my balance. | 16:56.57 |
| Robin_Watts | henrys: 2 of the falls were down to slippery conditions. | 16:57.22 |
| henrys | a bunch of stuff on the net about it. | 16:57.23 |
| Robin_Watts | The last one was down to me tripping over a kerb. | 16:57.30 |
| | henrys: interesting. | 16:59.18 |
| | mutool info is bonkers. | 17:06.54 |
| dans_ | anyone had issues with gs intermittently failing to find gs_init.ps ? Seeing this issue with the rghost gem rendering pdfs. gs version 8.70 | 17:34.57 |
| rayjj | dans_: I've never seen it be actually "intermittent" in that different runs with exactly the same environment sometime fail | 17:42.34 |
| | dans_: but I am not familiar with "the rghost gem" | 17:43.03 |
| dans_ | it is odd. some pdfs render just fine while other's repeatedly fail on the same server | 17:43.49 |
| rayjj | dans_: I don't recall offhand when we went to having COMPILE_INITS=1 be the default. With COMPILE_INITS=1 the gs_init.ps gets built into the executable in the %rom% file system | 17:44.28 |
| dans_ | under the hood the rghost gem runs a system call to the gs executable. nothing too exciting there | 17:44.46 |
| rayjj | dans_: so you can't get the output from "gs -h" ??? | 17:45.12 |
| Robin_Watts | dans_: Someone should do the obligatory: "dans_: 8.70 is over 5 years old. Have you considered moving to a less hopelessly out of date version?" | 17:45.50 |
| rayjj | dans_: that would tell you if the %rom% file system was on the LIBPATH list | 17:45.53 |
| | Robin_Watts: yeah, thanks for mentioning that :-) | 17:46.11 |
| dans_ | lol, thanks Robin_Watts. I know but its the version of the package provided on RHEL6 so its what we've got for now | 17:47.11 |
| rayjj | even our printer customers that take a long time to upgrade due to extensive testing needed have moved past 8.70 | 17:47.17 |
| | RH 6 is pretty old too | 17:47.44 |
| Robin_Watts | dans_: Yes, but using versions that distributions have broken for you will cause you other problems. | 17:47.51 |
| dans_ | RH 7 is pretty new though, so... | 17:48.02 |
| Robin_Watts | I would seriously recommend building from our most recently released source. | 17:48.16 |
| | No system libs, just use what we supply. | 17:48.34 |
| | Then if it doesn't work, we'll be much more minded to help you out. | 17:48.49 |
| | We try to avoid hunting for bugs that can't be reproduced with the source as we supply it, as otherwise it turns into a guessing game for "how has the distro broken it this time"? | 17:50.13 |
| dans_ | I hear ya. Just thought I'd ask here since I'm at a bit of a loss | 17:52.56 |
| | 'gs -h' outputs a search path that includes the location of gs_init.ps | 17:53.19 |
| Robin_Watts | dans_: No worries. Grab 9.16 from the website, install it, try it, then come back here with the results. | 17:53.33 |
| rayjj | dans_: does it have %romesource/Init/ as one of the paths | 17:53.49 |
| | dans_: does it have %rom%Resource/Init/ as one of the paths | 17:54.07 |
| dans_ | no | 17:54.34 |
| | just full file system paths | 17:54.47 |
| mvrhel_laptop | Hi Robin_Watts | 18:04.17 |
| Robin_Watts | hi mvrhel_laptop. | 18:08.20 |
| | sorry, almost missed you there. | 18:08.26 |
| mvrhel_laptop_ | no problem, I am getting a late start today | 18:08.43 |
| Robin_Watts | mvrhel_laptop: Did you see my burbling earlier about the pdfclean change? | 18:11.37 |
| mvrhel_laptop | yes. let me know when you push it to golden | 18:11.46 |
| Robin_Watts | Will do. | 18:11.58 |
| | I suspect tor may want to beat me up (rightly) over the naming of it. | 18:12.25 |
| mvrhel_laptop | ok. I have the x64 printing issue fixed | 18:41.24 |
| | it was a silly issue with the wrong runtime lib getting copied. The 32bit and 64bit have the same name and nsis did not recognize that they were different files | 18:42.12 |
| | also, the application is a bit cleaner now in that it should not leave any temp files laying around | 18:42.33 |
| | henrys: ok so I think we are ready for a beta release | 18:47.37 |
| | maybe by monday, to give everyone sometime to beat on it more | 18:47.55 |
| | or not | 18:47.57 |
| | I am feeling pretty good about it now | 18:48.06 |
| Robin_Watts | mvrhel_laptop: In the usual place on casper? | 18:51.48 |
| mvrhel_laptop | Robin_Watts: yes | 18:51.54 |
| Robin_Watts | downloading now. | 18:54.06 |
| mvrhel_laptop | cool | 18:54.11 |
| Robin_Watts | I'm getting towards the list of things I want to get done for the mupdf release... just need to get tor to review the stuff in the queue and fix any comments he has. | 18:55.43 |
| mvrhel_laptop | does it include pdf merging? | 18:56.09 |
| | ;) | 18:56.10 |
| Robin_Watts | s/towards/towards the end of/ | 18:56.13 |
| | mvrhel_laptop: It does not, sadly. | 18:56.20 |
| mvrhel_laptop | we need that... | 18:56.28 |
| Robin_Watts | That's more than a quick change. | 18:56.29 |
| | I'd love to do it. | 18:56.34 |
| mvrhel_laptop | I need to add form filling in gsview | 18:56.42 |
| | as well as the ability to add annotations | 18:56.48 |
| Robin_Watts | Now that's more than a quick change :) | 18:56.54 |
| mvrhel_laptop | is the Android code a good place to review how best to do that? | 18:57.05 |
| Robin_Watts | mvrhel_laptop: The android code is the only place to review how to do that :) | 18:57.17 |
| mvrhel_laptop | ok. and there is no documentation right? | 18:57.29 |
| Robin_Watts | self documenting code! | 18:57.43 |
| mvrhel_laptop | of course. my favorite | 18:57.52 |
| Robin_Watts | Where did the license text for gsview come from? | 18:59.30 |
| mvrhel_laptop | from Miles | 19:00.04 |
| Robin_Watts | The "File Type Associations" window looks a bit blank. | 19:00.47 |
| mvrhel_laptop | what did you have in mind? | 19:01.11 |
| Robin_Watts | It could perhaps have "Select the File Types you would like to be associated with gsview by default:" or something. | 19:01.14 |
| | Rather than just "File Type Associations:\nSelect..." | 19:01.34 |
| mvrhel_laptop | fair enough | 19:01.37 |
| Robin_Watts | Can the associations be changed easily once installed? | 19:02.00 |
| mvrhel_laptop | we are talking about windows.... | 19:02.10 |
| | yes | 19:02.12 |
| | it is | 19:02.14 |
| | you can right click on the file type | 19:02.19 |
| | and acrobat, gsview, etc will come up | 19:02.31 |
| | and you can pick a default | 19:02.35 |
| Robin_Watts | but there is not a preferences pane in gsview to set them? | 19:02.47 |
| mvrhel_laptop | no that is a system level thing really | 19:02.58 |
| | I have not seen many apps that do that | 19:03.10 |
| Robin_Watts | Ok. Lots of apps say "To change the preferences at a later date, do xxx" | 19:03.30 |
| | s/preferences/associations/ | 19:03.40 |
| mvrhel_laptop | I don't think that is preferred though | 19:03.45 |
| Robin_Watts | OK. | 19:03.54 |
| mvrhel_laptop | I recall reading that on a MS page some place | 19:03.57 |
| | especially some of these file types are set in funky places in the registry depending upon the system | 19:04.23 |
| | I cant guarantee that the app can successfully do the association on the various system types | 19:04.56 |
| Robin_Watts | mvrhel_laptop: First impression: the "gsview" in the top corner looks lonely. | 19:05.00 |
| | 'GSview' ? | 19:05.05 |
| | Perhaps we should have a bundled PDF file that says "Welcome to gsview" or something and that could load by default with simple instructions in? | 19:05.40 |
| mvrhel_laptop | interesting on windows 8 it is in the center of the window | 19:05.47 |
| | on 7 it is the left. | 19:05.52 |
| | that is a good idea | 19:06.02 |
| | not sure if we are calling it gsview or GSview | 19:06.31 |
| Robin_Watts | Ctrl-mouse wheel should zoom. | 19:06.38 |
| | (he says, all smugly, cos the old viewer has been doing that for at least 2 days now) | 19:06.54 |
| mvrhel_laptop | oh I can do that | 19:07.08 |
| Robin_Watts | shift mousewheel should scroll left right within a zoomed in page. | 19:07.28 |
| | We can't follow internal links at the moment? | 19:08.15 |
| mvrhel_laptop | oh I see AR does that | 19:08.18 |
| | never new it | 19:08.21 |
| | yes | 19:08.25 |
| Robin_Watts | so does chrome, and most other things. | 19:08.26 |
| mvrhel_laptop | internal links should work | 19:08.37 |
| | as well as web external ones | 19:08.43 |
| Robin_Watts | Load pdf_reference17.pdf, and none of the links in the Contents page work. | 19:09.27 |
| mvrhel_laptop | hmm odd | 19:09.32 |
| Robin_Watts | You have 2 writable fields at the top, for the page and the zoom. | 19:09.44 |
| | both should be a tad wider I feel. | 19:09.52 |
| | 128/1310 when it means 1280/1310 :) | 19:10.15 |
| | being called away, sorry. | 19:10.30 |
| mvrhel_laptop | no problem. need to head to lunch anyway. | 19:10.54 |
| | I will check links on windows 7 with the ref manual | 19:11.02 |
| | links work fine for me | 19:12.26 |
| | odd | 19:12.27 |
| | as well as external | 19:12.40 |
| kens | Robin_Watts : y I recall that million page file | 19:13.06 |
| mvrhel_laptop | Robin_Watts: when you click on the link button at the top right of the menu bar, the links should all be highlighted | 19:14.37 |
| | then you can click on them | 19:14.43 |
| | to go places | 19:14.45 |
| | maybe you did not activate them | 19:14.51 |
| kens | mvrhel_laptop : I tested taht wth PDF_reference17.pdf, it worked for me | 19:15.54 |
| mvrhel_laptop | ok cool | 19:15.58 |
| kens | Also external URLs as I recall | 19:16.12 |
| mvrhel_laptop | ok. that worked for me too | 19:16.23 |
| | kens: so the issue with x64 and selection of the printer should be fixed | 19:17.09 |
| kens | is amused by Robin_Watts noticing the ctl-wheel zooming. I did try that too, but I didn't mention it. | 19:17.28 |
| mvrhel_laptop | hehe. I never use that | 19:17.37 |
| kens | mvrhel_laptop : I see that from the logs | 19:17.39 |
| mvrhel_laptop | so I was not aware | 19:17.42 |
| | I will add it in | 19:18.00 |
| kens | I felt it was a app-specific thiing. I can't recall if Acrobat does that, and I'm on my laptop atm | 19:18.05 |
| mvrhel_laptop | they do do it | 19:18.13 |
| | I just checked | 19:18.16 |
| kens | oh ok | 19:18.21 |
| mvrhel_laptop | I fear there are going to be a million of these little ui things like that | 19:18.51 |
| kens | No doubt :-( | 19:19.00 |
| mvrhel_laptop | as different people try to use it | 19:19.04 |
| kens | y can probably find a lot by looking at the MuPDF bug reports :-) | 19:19.24 |
| mvrhel_laptop | yes | 19:19.35 |
| kens | I'll try and make some time to test your (and Fred's) latesa nasty PDF interpreter bug to sort outI havt binaries but | 19:20.13 |
| mvrhel_laptop | ok thanks kens | 19:21.03 |
| | heading out for lunch for a bit | 19:21.23 |
| kens | runs away again | 19:21.26 |
| mvrhel_laptop | Robin_Watts: if you can put your comments in an email that would be great | 19:21.36 |
| Robin_Watts | tor8: ping? | 19:24.47 |
| | mvrhel_laptop: I still get a problem on swapping printer. | 19:47.28 |
| kens | Robin_Watts : ping | 20:06.15 |
| Robin_Watts | pong | 20:06.26 |
| kens | Did you want to talk about that million page file ? | 20:06.35 |
| Robin_Watts | kens: Nope. I found the bug thanks. | 20:06.47 |
| kens | Ah OK then I'll go off again | 20:06.57 |
| | See you tomorrow | 20:07.00 |
| Robin_Watts | I also implemented a "mutool pages" option that dumps the info that hcl wanted. | 20:07.05 |
| | kens: have a good one. | 20:07.09 |
| kens | :-) HCL could have speeded GS by a factor of 2 simply by not doign 2 passes over the file | 20:07.30 |
| | anyway.... | 20:07.38 |
| mvrhel_laptop | Robin_Watts: odd about the printer problem. a clean install of windows 7 x64 worked fine here | 21:31.43 |
| | I may have you tell me what all is in your bin directory tomorrow | 21:31.58 |
| | ah I see the text selection issue. had not noticed that before | 21:33.22 |
| | within a line there is an issue | 21:33.37 |
| | ok, quite a few things to do then | 21:38.46 |
| | Forward 1 day (to 2015/04/03)>>> | |
IRC Logs