IRC Logs

Log of #ghostscript at irc.freenode.net.

Search:
 <<<Back 1 day (to 2013/05/07)2013/05/08 
mvrhel_laptop whew. finally got this thumbnail background task rendering working 06:05.53 
  and significantly cleaned up06:06.15 
  now I need to get the zooming working properly again, device rotation, and figure out the canvas UI binding for the link and text search results 06:11.08 
  and device rotation is working again06:31.20 
  night all07:04.40 
sebras Robin_Watts: there is? where?08:35.03 
  I must have missed that completely.08:35.17 
Robin_Watts paulgardiner: I rooted the nook last night. The market won't offer me mupdf, presumably because it needs multitouch ?12:18.31 
  I'll try a manual installation later.12:18.48 
  paulgardiner: I'll commit your documentation, then?12:22.23 
paulgardiner Oh yes please. I have a nook on order, so hopefully that will provoke me into sorting out non-multitouch!12:26.26 
  Robin_Watts: do you need the micro sd card so as to root it?12:27.09 
Robin_Watts paulgardiner: You do.12:27.19 
  You need to write the rooter to the microsd card using an external USB writer.12:27.55 
  then you put that into the nook, reboot, and voila.12:28.04 
  After that you don't need the SD card.12:28.24 
  I have an SD card and a writer here.12:28.34 
  so if you don't have one, I can root it for you.12:28.43 
paulgardiner ah could be handy. I have a 2GB one here. Not sure about a writer. Should have one somehere.12:29.26 
  I guess the Vario II could be used as a writer. Is 2GB enough?12:33.50 
Robin_Watts paulgardiner: 2gig is fine.12:34.01 
  You need direct access to the card.12:34.18 
  i.e. you need to be able to do the equivalent of dd in linux.12:34.33 
paulgardiner Right. May well take you up on your offer.12:35.07 
henrys kens:sorry the pattern in that character looked distinctly different than the rendered pattern, I thought that is what you wanted.13:28.14 
kens The pattern is resolution dependent13:28.30 
  But the single quotes are filled in solid blue13:28.41 
henrys kens:np I'll do it again shortly13:29.17 
kens sThanks henry13:29.22 
henrys kens:you certainly haven't tripped over the easy stuff ;-)14:19.41 
  complicated interactions in this test.14:20.19 
kens henrys yes, I'm sure that's true :-(14:25.11 
  Easy ones were all done some time ago14:25.32 
  I don't really know what's going on with the intellifont one, it *ought* to work....14:26.03 
henrys kens:is it a bitmap font?14:28.47 
kens the inteelifont one ? No14:28.54 
henrys it's an actual intelligent outline.14:29.08 
kens wwwwwwwapparently so yes14:29.20 
henrys s/intelligent/intellifont14:29.25 
  colloquy corrects my spelling14:29.41 
kens :-)14:29.57 
henrys interesting to see if it is broken with pswrite - then we'd know it was a coordinate system issue vs. a font issue.14:32.50 
kens THe font is odd, its 'upside down'14:33.37 
henrys well you do seem to be offset by the point size so I think you are on the right track.14:34.54 
kens that seems to be the problem, the font is going the 'wrong way' somehow14:35.15 
henrys I remember I had problems with the font matrix and xl -> pdfwrite but I can't remember the details.14:36.13 
kens well it renders fine, so it should be ok14:36.37 
Robin_Watts WooHoo! Tiling patterns!15:11.01 
tor8 Robin_Watts: speaking of tiling patterns, I had a thought last night15:13.53 
  wondering if we could have the ifdef TILE in the interpreters based on device hints15:14.23 
Robin_Watts devices can already control it.15:14.41 
tor8 #ifdef TILE in pdf_interpret.c is the one I'm talking about15:15.31 
Robin_Watts yeah, sorry, ignore me.15:15.57 
tor8 mostly thinking in case the svg tiling turned out to be too difficult15:16.20 
Robin_Watts The svg tiling just fell out quite neatly.15:16.35 
  (touch wood)15:16.39 
tor8 I started banging on the ghostsvg code, see if I could get it working in mupdf instead15:17.04 
Robin_Watts Who originally wrote the ghostsvg code?15:17.24 
tor8 I think I did...15:17.31 
Robin_Watts Ah :)15:17.37 
tor8 it looks like my code anyway :)15:17.42 
Robin_Watts How fully featured is it ?15:17.53 
tor8 not very, but it does tiger15:18.04 
Robin_Watts (I mean, ignore the DOM and the animation stuff etc)15:18.13 
tor8 text and images were the big pieces missing when I last worked on it15:18.38 
  text because doing text layout in ghostscript is ... well ... painful15:18.52 
Robin_Watts ok.15:18.54 
  doing text layout in mupdf is something we don't currently do either - but we possibly should.15:19.20 
  If we want to handle reflow for ourselves we'll need it.15:19.31 
tor8 I think we really ought to add it. both for reflow, svg and the forms stuff15:19.44 
Robin_Watts and if we want to handle other input formats (like .doc? or .txt?) we'll need it.15:20.01 
  Speaking of random ideas....15:20.10 
tor8 the "reflow" document type probably should be some cooked down html/css box model thing15:20.17 
Robin_Watts In the draw device, when we are asked to render text, we either convert it to glyph bitmaps and plot them, or we convert to paths, and plot them.15:20.58 
  I was wondering how possible it would be to make the "and plot them" stages be calls back into the device again.15:21.19 
tor8 something like ghostscript has with fallbacks for devices that can't cope?15:21.44 
Robin_Watts Exactly.15:21.51 
tor8 we could trivially do it with a "decompose operations" intermediate device15:22.08 
  or if we make a generic decomposition device, just add that as the default callbacks in dev_null.c15:22.25 
Robin_Watts That way we could use the draw devices text entry points for things like SVG to avoid having to send the fonts through, but still get perfect rendition.15:22.28 
  (ideally we'd want to send fonts through when possible, but...)15:22.42 
  tor8: right.15:22.51 
tor8 well, it'd be swell to have for type 3 fonts15:22.53 
Robin_Watts I haven't looked into it at all - that was just an idea that occurred to me last night.15:23.26 
tor8 so text -> paths, or text -> images decomposition15:23.33 
Robin_Watts yeah.15:23.42 
tor8 I'd guess we could add in tiles -> repeat the tile there as well15:23.48 
Robin_Watts yeah, maybe.15:24.05 
tor8 shading -> image15:24.25 
Robin_Watts marcosw_: ping ?15:37.17 
marcosw_ Robin_Watts: morning15:37.31 
Robin_Watts morning.15:37.35 
  Did you see the thing from henry about getting paulgardiner an @artifex.com address?15:38.06 
  tor8: ping15:48.52 
  mupdf seems to do a poor job of the tiling in fts_06_618.pdf - are you aware of this?15:49.27 
marcosw_ Robin_Watts: not yet, I've been dealing with my personal email so haven't checked my artifex account.15:50.39 
Robin_Watts marcosw_: ok, I'll shut up.15:51.04 
marcosw_ Robin_Watts: no problem, just dealing with an email from one of my children's teachers :-)15:51.40 
Robin_Watts marcosw_: Praise, I'm sure :)15:52.23 
marcosw_ Robin_Watts: they never email you with good news...15:52.49 
Robin_Watts tor8: gs does a similar job at low res. gets better at higher res. mupdf does not improve at higher res.15:54.30 
tor8 Robin_Watts: does it have anything to do with the tiling alignment / grid fitting settings that gs respects but mupdf doesn't (due to the device interface being resolution independent)15:58.37 
Robin_Watts tor8: could do, I guess.15:58.48 
  just running through the FTS now.16:00.04 
  Other than shadings and fonts, svg out is looking good.16:00.15 
henrys kens:very interesting I kept losing it because there needs to be a previous quote in blue to get the effect - presumably a cache issue.16:09.21 
kens ah, fascinating....16:09.38 
Robin_Watts kens: http://io9.com/old-spock-battles-new-spock-in-the-greatest-car-commerc-493836696?autoplay=116:12.00 
kens I bet that cost a bit :-)16:12.46 
  Oh god, and the hobbit song16:13.35 
  I wonder how much they had ot pay to get that16:13.43 
henrys so I'm surprised these other characters aren't being cached - the quote is smaller, but it isn't a particularly large font, has anyone looked at our cache hit rate recently?16:14.45 
kens Not at all no, but pdfwrite should not be using it anyway16:15.05 
  OK have to go now, henrys, did you reassign the bug to me ?16:19.58 
henrys I'm attaching a sample now and will make sure the assignment is okay bye.16:20.37 
kens OK thanks henry, night all16:21.24 
marcosw_ paulgardiner: I created a artifex.com email account for you. You should have received notification at your current laser point email address.16:26.52 
paulgardiner Great. Thanks Marcos16:28.53 
henrys paulgardiner:we'd like you to use that address if you talk to a customer, with artifex staff it doesn't matter.16:34.04 
paulgardiner henrys: yeah sure. I was aware using the Laser Point address wasn't ideal.16:35.41 
  Does this new address redirect to the laser-point one, or can I make it do so?16:38.24 
Robin_Watts sends test email.16:39.56 
paulgardiner Looks like I can set up a forward16:43.27 
Robin_Watts paulgardiner: Where does it currently end up?16:43.43 
paulgardiner I can access it from GMail16:44.19 
  That should do it16:44.53 
henrys marcosw:where is the scan build control flow? I see the text warnings17:07.48 
Robin_Watts oh, crap.17:12.55 
  SVG has a wierd interpretation of strokes within a clip path.17:13.18 
  it fills the stroked path :(17:13.38 
  so how the hell can I convert a 'clip to stroked path' into SVG?17:23.09 
marcosw henrys: I don't understand the scan build control flow question. 17:26.31 
henrys I thought scan build was supposed to have output like this: http://clang-analyzer.llvm.org/images/analyzer_html.png I only see a text summary on your machine where the warnings and other reports are kept17:27.48 
marcosw henrys: I'm not sure, let me look into it.17:29.19 
  Is this it: http://marcos-artifex.homeip.net:8080/artifex/e64ac4c83a4ae07dc15c4cacd7b60cba6294899a/gs/index.html17:30.36 
  btw, no one panic, but I've started re-entering the fuzzing bugs. 17:32.03 
Robin_Watts marcosw: yeah, I saw the CachedXForm ones go in. and out. and in. and out.17:32.45 
  The bug hokey pokey :)17:32.55 
marcosw Robin_Watts: bugs in my auto enter bugs script :-) we need a development copy of the bugzilla databae17:33.30 
Robin_Watts marcosw: Ah!17:34.00 
henrys marcosw:yes that's it, thank you.17:34.24 
mvrhel_laptop finally I figured out how to get these textbox and link rectangles bound to the xaml ui17:34.26 
  I had asked a friend at microsoft because I could not figure it out and he just got back with me with someone who would know almost exactly at the same time that I finally got it working 17:35.17 
  I mean like 5 minutes apart17:35.29 
  this is going to be much cleaner than what I had previously17:35.54 
Robin_Watts Ok. Stack overflow question time, I guess.17:40.15 
marcosw_ Robin_Watts: since you recently replied to Mateusz I'll ask you about the fuzzing bugs. Should I only enter bugs for those tests that cause a seg fault or for all that cause a valgrind report? 17:43.18 
Robin_Watts marcosw_: Both I reckon (but state which is which)17:43.51 
  I mean, you're opening bugs for valgrind things anyway, right?17:44.07 
marcosw_ Robin_Watts: yeah, it's just that there a lot of valgrind reports in the fuzzing files. I'm trying not to overwhelm the team with bugs.17:45.02 
Robin_Watts Speaking for myself, I am adept at ignoring bugzilla bugs that I don't want to think about. I suspect my colleagues are similarly skilled :)17:45.43 
marcosw_ is "ignoring bugs" listed on your CV under skills?17:47.52 
  :-(17:48.52 
  or maybe :-|17:48.59 
  ;-)17:49.11 
  sorry, amused by the emoticons in colloquy17:49.32 
ray_laptop marcosw: I didn't realize that you had closed the bugs as invalid before commenting. The description on that one didn't match the logs that were attached. Is that why you closed so many as INVALID ?18:01.40 
  I'm more than happy to wait for any fuzzing bugs to show up :-)18:02.30 
henrys chrisl:will the new build have the feature of having the each language as a static library or dll? If so I'll reassign one of my enhancements to you ;-)18:03.06 
Robin_Watts henrys: So... you assigned bug 693377 to me.18:07.45 
  which ones were you thinking were my fault?18:07.56 
  The gsroprun1.h ones ?18:08.18 
henrys I guess you need to explain the rop ones and push it along to the next victim18:08.29 
Robin_Watts henrys: The rop ones are essentially unsolvable.18:08.47 
  I cast a pointer to an int and then "&3"18:09.00 
  so as long as ints are at least 2 bits in size, the warnings are wrong.18:09.19 
henrys I know but I guess you have to say that if marcosw is going to close the bug - your blaming the messenger ;-)18:09.44 
Robin_Watts The ones in base/gxclrect.c are similarly bogus, but there may be a cunning hack.18:09.53 
  Rather than doing (X>>n) I can do ((X>>(n/2))>>(n/2))18:10.46 
chrisl henrys: my eventual plan is to have each language interpreter and the graphics lib as a static libs - I haven't thought about DLLs18:13.30 
henrys chrisl:my bug count is really coming down today.18:13.58 
chrisl henrys: I should I just reassign a few of mine to someone else - just to shuffle things round a bit?18:14.51 
henrys as long as you say: no backsies it should be okay.18:15.52 
Robin_Watts ooh.18:18.17 
  ptrdiff_t should be 64bit on 64bit platforms, right?18:18.30 
chrisl Robin_Watts: probably should be, but in practice it's probably rarely an issue - given how our memory management works18:20.14 
Robin_Watts chrisl: but if I cast to ptrdiff_t then I'm safe and I shouldn't get a warning.18:20.44 
chrisl Robin_Watts: Yes, I'd think so - how widely available is it?18:22.29 
Robin_Watts MSVC has it.18:22.52 
  It's in stddef.h18:23.07 
chrisl I have a feeling it might be a C99 addition, but not sure18:23.36 
Robin_Watts It's in K&R 218:25.00 
chrisl Robin_Watts: couldn't you just use a uint64_t?18:25.02 
  Ah, well in that case.....18:25.16 
Robin_Watts chrisl: But then I'd be casting it to larger than I needed on 32bit builds.18:25.27 
  and I distrust compilers :)18:25.32 
chrisl My concern was that the tiff configure script checks for the ptrdiff_t type, which suggests it's not universal18:26.40 
  But as it's in K&R 2 then I'd think you're fine18:27.35 
henrys size_t works too right?18:28.19 
chrisl size_t can also be 64 bit on 32 bit platforms18:28.58 
Robin_Watts ptrdiff_t is more correct, I believe.18:29.33 
henrys yeah I guess so18:30.23 
sebras also ptrdiff_t ought to be signed while size_t is unsigned, which might cause warnings.18:34.15 
Robin_Watts bah. ptrdiff_t is not defined where I need it.18:37.33 
  ptrdiff_t is defined in the same place as size_t - so what's the gs approved way of getting size_t ?18:41.05 
  Can I assume that stddef.h is available ?18:43.03 
henrys how about putting them in stdint_.h and using ARCH_SIZEOF_PTR to define them, so we don't have any platform surprises.18:47.48 
  but this is chrisl's domain18:48.31 
Robin_Watts I've fixed it with some local #definery where it's used.18:49.03 
henrys ptrdiff_t seems quirky on my mac in the debugger I can say "ptype size_t" and it says long unsigned but ptrdiff_t is not defined.18:51.38 
chrisl That may explain libtiff checking for it....18:52.16 
henrys and from my read of the header files it could turn out to be an int (32 bit) on a 64 bit system.18:52.19 
  strange18:52.33 
chrisl Well, my G4 MacMini is now resurrected from a dead disk, so I think I'll call it a night19:49.13 
tkamppeter mvrhel_laptop, hi21:29.36 
mvrhel_laptop hi tkamppeter21:31.20 
  I am hoping to work on my slides this week21:31.35 
  and I will get them uploaded after that21:31.45 
tkamppeter mvrhel_laptop, you answered exactly what I wanted to ask you, thanks.21:32.15 
mvrhel_laptop tkamppeter: and I am set for talking next week Wed. yes?21:32.43 
  on your schedule21:32.53 
tkamppeter mvrhel_laptop, yes, in the afternoon session.22:28.27 
mvrhel_laptop ok sounds good. thank you22:28.36 
  tkamppeter: ^^22:28.42 
 Forward 1 day (to 2013/05/09)>>> 
ghostscript.com
Search: