| <<<Back 1 day (to 2015/02/24) | 20150225 |
Robin_Watts | kristian_on_linu: what is mogrify? | 00:33.17 |
| You've picked a bad time of day to ask on here. better to come back in 8 hours time. European or US day time. | 00:33.49 |
| If you're having a problem with a program that calls ghostscript, then the first thing we'll ask you to do is to reproduce the problem calling ghostscript directly - not calling through some other piece of code. | 00:34.24 |
| We maintain ghostscript and try and fix any bugs that people find in it. We really try to avoid having to workout how third party code might be calling gs. | 00:35.14 |
| That's not part of our job :) | 00:35.20 |
kristian_on_linu | mogrify is part of ImageMagick... however, this problem persists with gs | 00:35.34 |
Robin_Watts | I would suggest that you try the latest version of gs. (9.15 at the time of writing) | 00:35.50 |
| Then if the problem happens with that, give us the exact command, and a copy of the file you are using. | 00:36.13 |
kristian_on_linu | yes, as stated it seems to be a bug in 9.10 specifically... it's gone in 9.15 | 00:36.30 |
Robin_Watts | gotta reboot. | 00:39.20 |
rayjj | kristian_on_linu: glad to be of help (by already fixing the problem you had). If it's already fixed in 9.15, our job is done | 00:39.26 |
kristian_on_linu | no idea if it was the GS project or Linux Mint who borked it | 00:40.26 |
| never mind, it's fixed now | 00:40.38 |
| by the way, how old is the GS syntax? It seems a bit different from most other software... | 00:41.21 |
rayjj | kristian_on_linu: the gs syntax was started in 1987 | 00:41.49 |
| kristian_on_linu: we've added a few things that are more modern, such as -o ___ instead of: "-sOutputFile=___ -dNOPAUSE -dBATCH" | 00:42.35 |
kristian_on_linu | I see... | 00:43.17 |
rayjj | kristian_on_linu: and the ability to take 'k' 'm' and 'g' suffixes for numbers for arguments such as -dMaxBitmap=1g | 00:43.32 |
kristian_on_linu | a fine tool, but one I rarely use without googling the commands or using an alias | 00:44.17 |
| have you been with the project since the start, rayjj? | 00:44.42 |
rayjj | kristian_on_linu: but generally we can only "add" since dropping previous parameters mean that apps that call us (such as imagemagick) will break if we drop the old parametersm and we resist having multiple parameters that do the same thing (mostly) | 00:44.49 |
kristian_on_linu | makes sense | 00:45.06 |
rayjj | kristian_on_linu: no, L. Peter Deutsch was the one and only developer then. I started using ghostscript at a previous company as a customer in 1991 | 00:45.30 |
| kristian_on_linu: then after licensing it and admiring its utility, it was irresistible to not dive into the code :-) | 00:46.27 |
| My previous company has the distinction of being the first to embed Ghostscript as the PostScript RIP in a printer | 00:47.34 |
kristian_on_linu | I don't even understand what that means :) | 00:48.00 |
| is GS embedded a lot? | 00:48.08 |
rayjj | kristian_on_linu: a RIP is the "Raster Information Processor" that takes high level (PostScript) in and turns it into the dots needed to drive a printer | 00:48.49 |
| kristian_on_linu: GS is not embedded as much as we like, although we are slowly gathering printer companies to our fold | 00:49.30 |
kristian_on_linu | it's strange... I have a media player thing, and it's running open source software... but it's very unfriendly towards hacking | 00:50.27 |
| so from a free software point of view, it does not make much of a difference | 00:51.14 |
rayjj | kristian_on_linu: and GS is 'embedded' on other devices such as airplane cockpit display systems, seismic exploration vehicles and heavy construction machinery | 00:51.33 |
kristian_on_linu | I understand how you could have other types of motivation, of course | 00:51.34 |
| I don't know what OS you're on, but a lot of Linux software calls on GS too... which is usually fine, but a total disaster when GS is not working :) | 00:52.57 |
rayjj | kristian_on_linu: yeah, that's the problem with a lot of "cheaters" w.r.t. open source, although some open source licenses such as MIT and BSD don't require 'hackability' as GPL tries to | 00:52.58 |
| kristian_on_linu: I run linux and windoze | 00:53.16 |
| I avoid mac os, but nowadays, it's close enough to unix that it is useful | 00:53.45 |
| kristian_on_linu: I also run android, but let others have the fun of developing for that (and iOS) | 00:54.24 |
kristian_on_linu | this is GPL software, and it's there... but it's a lot of work getting to it. I guess you could find a closed media player that would be easier to work with. | 00:54.29 |
| what Linux distro, out of curiosity? | 00:54.55 |
rayjj | kristian_on_linu: and of course the "strange" os'es such as VRTX (mostly gone) and Thread-X (unfortunately still used in Japan) | 00:55.05 |
kristian_on_linu | I hear there's a lot of otherwise abandoned software used in Asia... they're also pretty big on XP, as I understand it. | 00:56.02 |
rayjj | kristian_on_linu: ubuntu is the one I've found the handiest, but I've installed RedHat and Suse on VM's for looking at problems, and sometimes had to look into HP-UX or AIX or SunOS specific issues | 00:56.25 |
| thankfully, most of that falls to Marcos or Chris now | 00:56.38 |
kristian_on_linu | I've mainly used Ubuntu/Debian and friends, but also played around with Slax and all kinds of weird stuff | 00:57.38 |
rayjj | kristian_on_linu: the strange linux variants I work with are the 'embedded linux' versions for specific hardware such as Raspberry Pi and other manufacturers I can't name (NDA) | 00:58.52 |
kristian_on_linu | for GS? | 00:59.08 |
rayjj | kristian_on_linu: yes, and GhostPCL (the PCL parser based on the GS graphics lib / device layers) | 00:59.54 |
| The last printer I worked on at my previous company used C-Exec (which is what Adobe used in all of their PostScript printers until switching over to VRTX) | 01:01.28 |
| embedded linux is _so_ much nicer | 01:01.54 |
kristian_on_linu | I have little understanding of printers... I'm just grateful when they work | 01:02.09 |
rayjj | all of the printer embedded customers we support now either run embedded Windows or linux | 01:02.47 |
| Some of them are really amazing products -- a 350 page per minute book on demand printer or a 36" wide full color printer that prints from a roll of paper at 3 inches per second. That's a 36"x44" poster in 15 seconds ! | 01:05.50 |
kristian_on_linu | it seems that the printers are much better than their software, in general | 01:06.53 |
rayjj | except for those that use Ghostscript, of course ;-) | 01:07.24 |
kristian_on_linu | I heard they work really really hard on making you buy the official cartridges and such... | 01:08.27 |
rayjj | kristian_on_linu: that's why the printers are so cheap. We did the same thing at my previous company, putting chips in the ink bottles and using software drop counting to write the chip to prevent refilling the bottle. | 01:11.09 |
| it wasn't entirely unjustified since most of those customers had service contracts and if they put something in that sludged up the inkjet head (which was $800 per color) it would eat us alive | 01:12.20 |
kristian_on_linu | yes, I understand these things are not done out of spite... but the customer gets a bad deal | 01:13.02 |
rayjj | kristian_on_linu: it's actually a good deal for the low-end user in that they can get a color printer cheap and as long just 'pay as you go'. The few companies that have tried pricing the consumables low and the print engine with reasonable profit margins have all failed | 01:16.59 |
| the cost of manufacturing, shipping, inventory and marketing a $300 color laser printer is about $400. If they didn't make money on the toner later, the printer would cost $1200-$1500 | 01:18.26 |
kristian_on_linu | not if the low end user gets a POS printer which needs very expensive ink every second month | 01:18.52 |
rayjj | (at least the "MSRP". The "street price would still be at least $800) | 01:19.02 |
| kristian_on_linu: that's why I (a rather typical low end user) got a color laser for $300 instead of a $60 inkjet printer that cost me $60 every other month | 01:20.02 |
kristian_on_linu | yup :) | 01:20.32 |
rayjj | at least this way I pay $150 for a set of 4 toner cartridges every other year (and it prints faster) | 01:21.39 |
kristian_on_linu | I was headed for that kind of solution, but the salesman talked me into getting a decent inkjet... a Brother DCP-J4110DW, would you happen to know it? | 01:21.44 |
| I was easy to persuade as I now and then need color prints for work | 01:22.07 |
rayjj | I have an inkjet I use at home ONLY when I want really high quality output (an Epson Workforce). Otherwise it's laser | 01:23.15 |
kristian_on_linu | yes, I'm thinking of getting a cheap laser printer as well | 01:24.16 |
kens | chrisl that one about RPM, I happen to know (because I saw it on SO already) that he has indeed built GS from source :-) | 08:09.11 |
chrisl | kens: well then, he's an idiot | 08:09.38 |
kens | Indeed he is, which is why I was less than sympathetic | 08:09.59 |
| I know nothing about RPM and even I could figure out why there was a discrepancy | 08:10.27 |
chrisl | I don't know about rpm, but it's not massively hard to make a .deb package and install it, but it's certainly nothing I'm going to try to explain on our bugzilla..... | 08:12.23 |
kens | Its also not in any way *our* bug.. | 08:12.38 |
chrisl | It's not in any way a bug. | 08:13.28 |
kens | Well yes that's true as well | 08:13.40 |
| Drat that idiot with the CIE colour stuff has responded, I was going to close that this morning | 08:14.01 |
| Oh and still nospecimen file. | 08:14.34 |
chrisl | Well, I would close it, with a comment about reopening *with* a test file | 08:15.31 |
kens | I'm just adding a comment, less than totally polite. | 08:15.47 |
| paulgardiner : Can you shed any light on this ? | 09:34.03 |
| http://stackoverflow.com/questions/28704332/password-protected-pdf-not-opening-with-mupdf-android | 09:34.03 |
paulgardiner | I'll take a look | 09:34.14 |
kens | Thanks | 09:34.19 |
paulgardiner | Nothing springs to mind. Probably good for Robin and Tor to have a look too. | 09:46.54 |
kens | OK well hopefully they'll see it in the logs, thanks for looking | 09:47.09 |
| Have to go out for a coule of hours, back in a bit. | 09:49.53 |
Robin_Watts | puts an answer up. | 09:59.45 |
paulgardiner | My guess would be wrong password. | 10:03.58 |
jogux | I guess the first question is whether he can open the pdf in the official google play mupdf... | 10:13.33 |
Robin_Watts | far easier to ask such questions here than in StackOverflow. It's not an ideal forum for back and forth. | 10:19.25 |
jogux | robin_watts: I've seen a few back/forths that work fairly well. The general principle is to comment on the question until the asker has editted the question to actually contain all the relevant info. | 10:29.51 |
Guest4058 | Hello, I have a problem to compile the MuPDF library for android. | 10:32.53 |
| Whenever I try to access ndk-build in the latest release comes the following error: | 10:32.59 |
| jni / mupdf.c: 140: error: undefined reference to 'fz_free_page' jni /../../../ source / fitz / document-all.c: 11: error: undefined reference to 'html_document_handler' jni /../../../ source / fitz / document-all.c: 11: error: undefined reference to 'epub_document_handler' collect2: error: ld returned 1 exit status make: *** [obj / local / armeabi-V7A / libmupdf.so] Error 1 | 10:33.07 |
| Someone here knows how to fix this error? | 10:33.14 |
jogux | Hi Guest4058 | 10:36.28 |
Guest4058 | hi jogux | 10:36.56 |
jogux | the 'master' branch of mupdf isn't that stable just now, if you try building some older code you might have more success. let me just find you a git sha that might be good. | 10:37.09 |
Guest4058 | oh okay, thank you. | 10:38.11 |
jogux | try 'git checkout 4fe9caafead6' - that's ~2 weeks ago before various changes happened that the ports haven't all caught up on yet. | 10:38.47 |
| master should be working well again in around a week I'd guess. | 10:39.50 |
Guest4058 | ok it's working. thank you for your help :-) | 10:45.58 |
jogux | Great! no problem :) | 10:48.03 |
tor8 | Robin_Watts: 1 commit on tor/master. your 4 commits on robin/master LGTM. | 11:58.28 |
jogux | Robin_Watts: I presume android builds are bust on master (see above Guest 4058), did you fix that? | 12:09.22 |
Robin_Watts | jogux: maybe. I will look at that now. | 13:22.45 |
| after I finish plastering myself :( | 13:22.58 |
| faceplanted while running. | 13:23.32 |
| waiting for superglue to set on my thumb now. | 13:23.46 |
jogux | Robin_Watts: afaict your changes fix the problem I was seeing on iOS. | 13:25.42 |
| Robin_Watts: eek :-( | 13:25.51 |
Robin_Watts | thanks. | 13:25.51 |
| my own fault. I commented to helen last night that my knee had almost got better from where I fell over at new year. | 13:27.51 |
jogux | this running thing sounds dangerous | 13:28.15 |
pedro_mac | Robin: ouch. shooting seems a much safer passtime ;) | 13:28.40 |
Robin_Watts | yes, on both counts :) | 13:28.54 |
kens | ?me will stick with swimming, water doesn't hurt so much | 13:29.13 |
chrisl | Depends if you try to breath it...... | 13:29.33 |
kens | I'm not keen on trying drowning, I prefer swimming :-) | 13:30.03 |
| I've been told that drowning is a (comparatively) pleasant way to go. WHat I want to know is; Who did thy ask ? | 13:31.18 |
Robin_Watts | swimming would be great were it not for the half hour drive there, the struggle to park, the changing rooms, the morons in the pool, the half hour drive back etc. | 13:32.17 |
chrisl | I can't stop the water from gushing up my nose when I try to swim | 13:32.49 |
kens | Well, the pool is a 15 minute walk for mw, and they have this lane set aside apparently just for me. At least everyone else gets out of it when I start swimmiong | 13:32.56 |
| Hmm, cannot reproduce these cluster seg faults. I guess I'd better start up a VM :-( | 13:33.44 |
Robin_Watts | tor8: All look good to me. | 13:35.52 |
| s/look/looks/ | 13:36.06 |
| how wierd. Adding an s changes the meaning from plural to singular :) | 13:36.36 |
| will look at android build in a mo. | 13:36.54 |
kens | Huh, can't reproduce the fault on 32-bit Linux build either | 13:55.56 |
robmv | Greetings. is the pdf device option DSCEncoding documentation current? I have tried it and I don't see it working | 14:07.17 |
| greping the source I don't find that that option really exist | 14:07.38 |
| at least not on /* Ghostscript-specific parameters */ part of devices/vector/gdevpdfp.c | 14:08.19 |
kens | robm what do you expet DSCEncoding to do with PDF ? | 14:09.59 |
robmv | kens: I want %%Title: (Con Acentos \341\351\355\363\372) to be decoded and added correctly to the PDF metadata | 14:11.25 |
| instead of being copied as the doc says if the option is not present | 14:11.47 |
kens | Hmm, well I don;t think that's done by the pdfwrite device, I htink its done in PostScript | 14:11.57 |
| Its in gs_pdfwr.ps in ghostpdl/gs/Resource/Init | 14:12.34 |
jogux | does anyone know if the mupdf Makefile deliberately does CFLAGS += $(XCFLAGS) before the include Makerules? that ordering prevents me disabling unused function warnings in the iOS build. (or, perhaps asked a different way, should be actually fix unused functions? they're in places like freetype, jbig, etc). Having warnings come out when you just open the xcode project and build seems bad, anyway. | 14:12.55 |
kens | Of course, its not something that ever gets tested, so it may well have bit-rotted | 14:13.11 |
| Rob: mv it looks to me like that flag is only operated on if the output is PDF/X or PDF/A compliant | 14:15.00 |
| robmv ^^ | 14:15.42 |
robmv | looks like it. I found another fixed bug when the default changed to be PDFDocEncoding | 14:15.52 |
| http://git.ghostscript.com/?p=ghostpdl.git;a=commit;h=a3d00daf5f9abb1209cb750a95e23bc6951c1c63 | 14:15.52 |
| and the name on the comment bug is PDFDSCEncoding, will try forcing PDF/X or PDF/A | 14:16.27 |
kens | Offhand I can't see any good reason why it should be insistent onPDF/A or PDF/X, that doesn't look correct, at a first look | 14:17.08 |
| THe comment for that commit is not quite correct, of course, it should be just DSCEncoding, not PDFDSCEncoding. I must have been having a bad day | 14:18.05 |
Robin_Watts | jogux: sorry, what's the problem? | 14:18.12 |
| how would you want to disable unused functions? | 14:18.24 |
jogux | Robin_Watts: I would like to disable the warning for unused functions | 14:18.36 |
| but -Wall ends up in the command line after my -Wno-unused-functions, switching the warning back on. | 14:18.52 |
Robin_Watts | jogux: Where is -Wall added? | 14:19.22 |
jogux | Makerules | 14:19.27 |
Robin_Watts | I would have no objection to CFLAGS += XCFLAGS happening after the makerules include. tor8? | 14:20.08 |
robmv | kens: When using -dPDFX -sDSCEncoding=PDFDocEncodingxxxx it get an error, so it is being used, without -dPDFX it is ignored | 14:20.22 |
| but -sDSCEncoding=PDFDocEncoding doesn't do the decoding | 14:20.41 |
kens | robmv yes it seems quite clear in the PostScript htat its only active if PDF/A or PDF/X is set | 14:20.48 |
| DSCEncoding isn't a standard PostScript encoding | 14:21.25 |
| SOrry I mean PDFDocEncoding | 14:22.26 |
| Although Ghostscritp probably does have it available. Even so, the glyphs you are using probably aren't PDFDocEncoding values ? | 14:23.06 |
robmv | I was assuming (Con Acentos \341\351\355\363\372) was PDFDocEncoding :( | 14:23.52 |
kens | I don't know, it depends what you are expecting those octal character codes to be. | 14:24.23 |
robmv | \341\351\355\363\372 is áéÃóú generated via a cups printer to pdf tool (CUPS-PDF) :( | 14:24.59 |
kens | Sadly I don't think taht came out correctly in Miranda, it contains a copyright symbol for starters | 14:25.35 |
| 341 is aacute in PDFDocEncoding | 14:26.17 |
| 351 is eacute | 14:26.47 |
robmv | look right then | 14:26.58 |
kens | Well in that case going by the bug report you pointed to, it should just work, since the code will assume that the string is in PDFDocEncoding | 14:28.05 |
tor8 | Robin_Watts: no objection, both the CFLAGS and LIBS += XCFLAGS and XLIBS could well be after both Makerules and Makethird | 14:28.50 |
jogux | great, thanks | 14:35.30 |
robmv | kens: will add a bug report, sadly it doesn't work anymore. Thanks | 14:36.05 |
jogux | tor8: Robin_Watts: so I'm now on the hunt for 64bit build warnings. fz_malloc* (and this may well be a much more general question) doesn't use size_t, is that deliberate? | 14:36.25 |
kens | Don't forget to add a specimenf ile | 14:36.29 |
| Otherwise I won;t be able to duplicate the problem. | 14:36.42 |
Robin_Watts | jogux: yeah... that's something that might need fixing. | 14:37.13 |
tor8 | jogux: it's semi-deliberate from a long long time ago | 14:37.29 |
| worth fixing though | 14:37.33 |
Robin_Watts | Will need prodding in memento too, I fear. | 14:38.02 |
tor8 | jogux: we do a lot of overflow tests before mallocing | 14:38.04 |
jogux | I see. I wonder if that's something I should really embark on. | 14:38.30 |
tor8 | egads I'd forgotten just how crazy the PDF colorspace pattern setting madness is | 14:38.35 |
Robin_Watts | jogux: I wouldn't. | 14:39.11 |
| or rather, I wouldn't expect you to. | 14:39.29 |
jogux | I'm just pondering if I should find a way to silence the warnings | 14:39.51 |
kens | chrisl when you have a few minutes spare.... I can't seem to sort out this memory problem with the clist. I've pushed my latest commits to my repo, could you pull that and have a look at the changes in gxclist.c and tell me what I've done wrong please ? You may be able to get it to seg fault (I can't) by using the command line "-dMaxBitmap=400000000 -sDEVICE=pbmraw -r300 -Z: -sDEFAULTPAPERSIZE=letter -dNOPAUSE -dBATCH -K100000 | 14:40.56 |
| 0 -dClusterJob -dJOBSERVER Bug691352.pdf" | 14:40.56 |
chrisl | is glad he just brewed some coffee........ | 14:42.38 |
kens | Well I'mhoping I've done some really obviously dumb | 14:42.54 |
chrisl | Allocating from immovable and freeing to non_gc seems like a bad idea...... | 14:45.46 |
kens | That's probably it :-) | 14:45.55 |
| Let me try changin that to non-gc | 14:46.38 |
chrisl | I think non_gc is preferable in this case | 14:47.21 |
Robin_Watts | tor8: 1 commit on robin/master | 14:47.39 |
kens | I think so,I don't really want to add a GC descriptor for this to the clist device | 14:47.40 |
Robin_Watts | jogux: Are you doing a commit to move the lines in the Makefile? | 14:47.54 |
kens | I've made them both non-gc and started a clister | 14:47.56 |
| or even a cluster | 14:48.01 |
jogux | Robin_Watts: yeah | 14:48.08 |
tor8 | Robin_Watts: LGTM | 14:48.56 |
jogux | some of the warnings I see (which aren't 64bit specific) are related to a decprecated stream destroy function in openjpeg, that seems to be fixed upstream (they undeprecated it). I guess we want to merge in upstream sometime? | 14:52.32 |
Robin_Watts | jogux: possibly. We have someone gradually feeding them all our fixes. | 14:53.14 |
| when they take them all, we'll move to a fresh copy. | 14:53.30 |
jogux | ok. I shall ignore it. | 14:53.57 |
Robin_Watts | tor8, jogux: 1 more commit on robin/master | 14:54.12 |
chrisl | Robin_Watts: have we got stuff pending other than the memory stuff? | 14:54.12 |
Robin_Watts | chrisl: I don't know. Possibly not. | 14:54.28 |
jogux | any opinions on this warning: | 14:54.33 |
Robin_Watts | I've not been tracking it. | 14:54.34 |
jogux | thirdparty/jbig2dec/jbig2_text.c:899:73: warning: for loop has empty body [-Wempty-body] | 14:54.35 |
| for (SBSYMCODELEN = 0; (1 << SBSYMCODELEN) < SBNUMSYMS; SBSYMCODELEN++); | 14:54.36 |
| ^ | 14:54.37 |
| thirdparty/jbig2dec/jbig2_text.c:899:73: note: put the semicolon on a separate line to silence this warning | 14:54.37 |
Robin_Watts | My opinion is that it's a bollocks warning. | 14:55.10 |
jogux | okay, I'll disable then. | 14:55.21 |
chrisl | Hmm, decent quality 8 port KVMs are rather expensive....... | 14:56.08 |
kens | I've not used a KVM for ages | 14:56.35 |
jogux | Robin_Watts: android fix LGTM | 14:56.44 |
Robin_Watts | chrisl: You have that many machines that all share the same output? | 14:56.50 |
| I mean, VGA vs DVI, and inputs, PS/2 vs USB etc? | 14:57.13 |
chrisl | Almost...... I've just reached the point where a fifth would be (temporarily) useful | 14:57.36 |
Robin_Watts | jogux: Ta. I think tor already reviewed that one. It's the fz_begin_tile_id one that needs review now. | 14:57.57 |
| chrisl: Daisy chain 2 of them :) | 14:58.19 |
chrisl | Robin_Watts: I suspect that would wreak havoc with hotkey switching! | 14:59.01 |
| I have main computer, Windows laptop and 2 x Mac Minis all use DVI and USB. And now I have a new computer to set up | 14:59.08 |
Robin_Watts | What's the software KVM thing called that uses VNC ? | 14:59.37 |
jogux | robin_watts: I'm confused. fz_draw_begin_tile returns 0 on success. you're returning zero for failure? | 14:59.58 |
| well, actually, I guess it would be more correct to say always returns 0 whether it failed or not :-) | 15:00.21 |
Robin_Watts | jogux: It returns 0 for uncached. | 15:00.23 |
| or 1 for cached. | 15:00.28 |
chrisl | Robin_Watts: can't remember - as this is just a short term thing, I'll probably take one of the Mac Minis offline until the new machine is ready for use | 15:00.34 |
robmv | kens: done http://bugs.ghostscript.com/show_bug.cgi?id=695850 thanks again for helping me a way to confirm there is a bug | 15:00.39 |
jogux | robin_watts: ahhh. yes. I see now. | 15:00.40 |
kens | robmv OK I'll get to it when I have a chance | 15:01.01 |
henrys | Robin_Watts: sabrina is going to take Helen and Linda around Denver Friday but I'm afraid they are going to stranded Saturday. We are in a very remote area at the hotel, need a car to get anywhere interesting. Of course we usually finish early day 2. | 15:01.15 |
Robin_Watts | henrys: We are getting a car on day 2 anyway. | 15:01.38 |
kens | robmv I can't see any reason why DSCEncoding is limited to PDF/X and /A, I may well change that | 15:01.44 |
jogux | robin_watts: I guess that's consistent then. LGTM. | 15:01.48 |
kens | Unless I can find out why its like that..... | 15:01.57 |
Robin_Watts | If need be the girls can get the car, and we can then go back to the hire place to add us on as drivers later. | 15:02.04 |
henrys | Robin_Watts: oh she's okay driving here? | 15:02.08 |
Robin_Watts | Assuming her Miami experience hasn't put her off, yes. | 15:02.28 |
| She's pretty gung ho. | 15:02.34 |
| After hyde park corner in a van, anything else is easy. | 15:03.02 |
kens | chrisl that still causes tghe same seg faults on the cluster. I put the (very small change) on my repo. | 15:03.28 |
chrisl | kens: okay, give me a couple of minutes - darn! coffee's finished....... | 15:04.16 |
kens | NP, there's no rush | 15:04.25 |
Robin_Watts | pops to post office. | 15:04.27 |
chrisl | kens: seg fault in read_set_bits() | 15:08.02 |
kens | Hmm, well not entirely unexpected I guess | 15:08.25 |
chrisl | I think it's probably easy to fix - give me a few minutes | 15:08.45 |
kens | Oh OK, thanks | 15:08.53 |
| I wish it would go wrong for me, seems to require 64-bit Linux | 15:09.16 |
chrisl | Did you try with -Z@ ? | 15:10.20 |
kens | No, I didn't think ofg that | 15:10.33 |
chrisl | Hmm, maybe not so easy - we seem to be getting there with the chunk not allocated (or disappeared) | 15:15.02 |
kens | I found on an earlier trial that it had disappeared, but I'd forgotten to use non-gc memory then, so I assumed it had grestore'd away | 15:15.35 |
chrisl | Well, the pointer is NULL so it's not been gc'ed away | 15:16.21 |
kens | that's peculiar | 15:16.37 |
| I'm fairly sure it had 0xf1f1 or something when I saw it | 15:17.10 |
chrisl | This seems to be a pattern clist - not sure if that's significant | 15:17.20 |
kens | I thinkall the files that failed have pattern clists | 15:17.34 |
| I wonder if the pattern clist 'close' is freeing the memory and we aren't getting an 'open' for the regular clist, that might do it I guess. I'll try and see if that's the case. | 15:18.43 |
jogux | right. I've got the warnings down to just the 3 openjpeg ones. woop. only downside is I can't push my changes as I need fred's to go in first. | 15:25.17 |
Robin_Watts | jogux: let me try to tidy freds up. | 15:26.03 |
henrys | jogux: if Fred's changes are good aplly them. | 15:26.07 |
Robin_Watts | henrys: They aren't. | 15:26.21 |
jogux | Robin_Watts: there's no rush, leave it to fred :-) | 15:26.30 |
henrys | Robin_Watts: sorry we were writing at the same time. | 15:26.45 |
Robin_Watts | jogux: We have a release coming up in march. | 15:26.58 |
| so getting everything sorted before that would be good. | 15:27.08 |
henrys | let's not wait. | 15:27.17 |
Robin_Watts | They shouldn't be hard to fix up. | 15:28.03 |
chrisl | kens: Indeed, I don't think clist_close() is the place to release the cache chunk(s)...... | 15:28.47 |
henrys | chrisl: I don't think PCL is a concern for language switch if we tear down the allocator all memory will be released. Norbert wanted to do a partial tear down as I recall. | 15:30.17 |
kens | I was thinking that, it looks like the pattern and regular clist are confused | 15:30.24 |
henrys | chrisl: but if you know of something let me know. I'll look at it. | 15:30.36 |
Robin_Watts | jogux: Is it just "conform to recent mupdf API changes." you need? | 15:32.19 |
chrisl | henrys: the problem with the current setup, IIRC, is that the whole PJL/PCL/PXL/HPGL is allocated from the same memory pool that the interpreters use, so to avoid leaking, we'll have to throw away the entire "universe" | 15:32.19 |
jogux | Robin_Watts: it is | 15:32.26 |
Robin_Watts | tor8: I can't see any trailing spaces or non-tabs in that, so I reckon that's good to go. | 15:32.35 |
chrisl | kens: I *am* struggling to understand how you don't see a crash | 15:33.40 |
kens | chrisl , no idea, it won't crash for me on 32 or 64-bit WIndows, or 32-bit Linux | 15:34.02 |
chrisl | That's concerning..... | 15:34.28 |
kens | I haven't tried -Z@ | 15:34.42 |
chrisl | Well, the fact that the pointer is explicitly set to NULL and then later dereferenced suggests that -Z@ wouldn't make a difference | 15:35.16 |
kens | Hmm, I get two frees fro a pattern clist, but the main device is OK | 15:35.19 |
henrys | chrisl: I can see a case for separating out PJL but I don't know the rest is significant ... it will be separate from PS/PDF which in turn has the same problem right. I don't think you keep PS state and dump PDF, right? | 15:36.41 |
Robin_Watts | jogux: All your commits look good. | 15:37.01 |
kens | We can't, the PDF interpreter is written in PostScript | 15:37.03 |
henrys | s/I don't think you/I don't think youi can/ | 15:37.14 |
Robin_Watts | I'd like to understand what the reason behind cf_uncompressed_decode being there is. | 15:37.21 |
kens | henrys we don't really have a PDF interpreter. We have a PostScript program which interprets PDF. So we don't have a 'state' for PDF. | 15:37.48 |
henrys | right so let's say PCL is memory constrained and we decide to kill PS - we automatically kill PDF, which kind of sucks but I think we can live with it. | 15:39.12 |
kens | We don't really have an alternative | 15:39.23 |
chrisl | henrys: in the current setup, the list of interpreters is stored in memory from the allocator used for PCL and co, so if we destroy that allocator, we lose the list of interpreters | 15:39.39 |
henrys | similarly killing PCL knocks out PXL, HPGL etc. | 15:39.43 |
kens | Well, mooscript, but that's nto completed... | 15:39.43 |
henrys | chrisl: I'm assuming we'll continue using that only for PCL/PXL not PS and PDF there will be another layer above. | 15:40.34 |
| chrisl: we should just get rid of the whole damn thing now but that will be bad for norbert I think. | 15:41.25 |
chrisl | henrys: even then, it would be preferable not to have to rebuild that list every time we invoke a PCL variant | 15:41.25 |
| henrys: does the current language switch code handle PJL switching into Postscript? | 15:42.28 |
henrys | chrisl: it used to, not tested recently. | 15:43.11 |
tor8 | Robin_Watts: the 'library changes' commit has lots of mixed tabs/spaces and trailing whitespace | 15:43.35 |
Robin_Watts | tor8: Right. That one we aren't taking yet. | 15:43.58 |
tor8 | Robin_Watts: as well as 'hide the library view' | 15:44.04 |
chrisl | henrys: I guess I would just like to see PCL and co cleaning up after themselves properly | 15:44.07 |
tor8 | but the first one (conform to changes) LGTM | 15:44.15 |
henrys | chrisl: I'll review norberts thing again. I think the only real issue is passthrough everything else I was able to clean up. | 15:45.55 |
kens | chrisl I'm not seeing any problems with clist_open/close here. The pattern clists are a different device, they aren't interfering with thge 'real' clist device. Although I do see multiple 'close' calls for the pattern this shouldn't be a problrem, because the pointer to the memory to be freed is NULL. | 15:48.59 |
henrys | bbiab | 15:49.04 |
kens | So I'm puzzled :-( | 15:49.09 |
chrisl | kens: but I think the memory is being freed *before* we playback the clist | 15:49.56 |
kens | Well, I'm not seeing that here. | 15:50.09 |
| The pattern clist is allocated and freed for each band (as you might expect), but the mail clist only twice. (the second time is after the end of the job) | 15:50.47 |
| I don't see the memory being freed or set to 0 until the complete sequence of bands has been executed | 15:51.11 |
chrisl | So, the sequence I'm seeing is clist_open() being called from pattern_clist_open_device(), then clist_close() being called from gx_pattern_load() | 15:53.04 |
kens | Let me check | 15:53.19 |
chrisl | Then clist_close() again from gx_forward_close_device() | 15:53.38 |
kens | I start with open from 'setup_as_command_list' which is then closed from 'tear down' then another 'setup as command list' | 15:53.52 |
| Then a pattern open | 15:54.12 |
| causes a clist_open | 15:54.18 |
| then a close from pattern_load | 15:54.32 |
chrisl | Of source, there are several patterns in this file :-( | 15:54.36 |
Robin_Watts | oh, gawd, I have the million page file to look at. | 15:54.45 |
kens | I know, that might be required | 15:54.47 |
| then I see hte close form the forward, from pattern_load | 15:55.01 |
| The chbunk memory is already 0, but thaT'S ok | 15:55.14 |
| Then an open from pattern_paint_prepare | 15:55.35 |
| and a close from pattern_paint_cleanup | 15:55.45 |
| and another from pattern_cache_free_entry | 15:55.58 |
| an open from pattern_load | 15:56.10 |
| and a close from pattern_load | 15:56.23 |
| an open from paint_prepare | 15:56.32 |
| close from paint_cleanup | 15:56.41 |
| Then I'm into playing back bands | 15:56.54 |
| Lots of open and close until all the bands are done | 15:57.30 |
| Then a close from gdev_prn_tear_down | 15:57.52 |
| and an open from setup_as_command_list | 15:58.05 |
| and another close from tear_down | 15:58.13 |
chrisl | It's the second time we go into clist_playback_band() that I see the crash | 15:58.18 |
kens | And GS exits cleanly | 15:58.21 |
| The playback bands executes 80+ times or so for me without problems | 15:58.49 |
| There's at least one open and close per band, could be 2 | 15:59.02 |
Robin_Watts | interesting. According to process explorer, when I run mutool clean on that file, it takes almost no CPU time. But my antivirus software is maxing out a core. | 15:59.10 |
kens | Presumably scanning the PDF file for viruses ? | 15:59.40 |
| be right back | 16:00.00 |
chrisl | kens: how many bands do you get when you run the job? | 16:02.44 |
kens | I'm not certain, let me see if this tells me | 16:02.58 |
| 42 it looks like | 16:03.32 |
chrisl | It's in the debug output: width=3413, band_width=3413, band_height=61, nbands=42 | 16:03.33 |
kens | I had to rerun it to see the output :-) | 16:03.51 |
| and yes, those are the same numbers | 16:04.03 |
chrisl | Okay, trying upping the MaxBitmap value by a factor of 10 | 16:04.21 |
kens | OK one moment | 16:04.29 |
| So 10,000 to 100,000 | 16:04.49 |
chrisl | Er, that's not the command you gave me..... | 16:05.03 |
kens | Really ? | 16:05.14 |
chrisl | -dMaxBitmap=400000000 | 16:05.19 |
kens | Hmm, I could have sworn I pulled that straight off my VC properties. I did just switch back to 32-bit from 64-bit, maybe they were differnt | 16:05.52 |
| OK let me try 400,000,000 | 16:06.09 |
| But it seems to me that's not having any actual effect | 16:06.21 |
chrisl | I'm using: -dMaxBitmap=400000000 -sDEVICE=pbmraw -sOutputFile=stuff%03d.pbm -r300 -Z: -sDEFAULTPAPERSIZE=letter -dNOPAUSE -dBATCH -K1000000 -dClusterJob -dJOBSERVER Bug691352.pdf | 16:06.25 |
jogux | Robin_Watts: great, thanks, I've puished those to golden. | 16:06.56 |
kens | Apart form the paths and the MaxBitmap, yes those are the same as mine | 16:07.05 |
| Ah, and now I do indeed see a seg fault | 16:07.23 |
chrisl | With only 1 band? | 16:07.33 |
kens | Yes | 16:07.36 |
jogux | I've spent a while testing the 64bit iOS mupdf, it seems to work fine, so I think we can do the next appstore upload as a 32+64. | 16:07.42 |
chrisl | Me too | 16:07.47 |
Robin_Watts | jogux: excellent. | 16:07.53 |
kens | Well, I say yes, but that's as far as it got, I think it said '0 of 1' before it crashed | 16:08.03 |
| OK then feel free to tleave this to me, at least I can see it now | 16:08.16 |
chrisl | I see "0 bands skipped out of 1" a couple more lines, then the segfault | 16:08.53 |
jogux | in reality, I needed to do nothing except fix warning, but fixing warnings is good anyhow :-) | 16:09.06 |
kens | Yeah that sounds about the same before my windo curled up and went away | 16:09.08 |
chrisl | I think the trick is to have just one band | 16:09.22 |
kens | I think the trick is to have just a pattern clist | 16:09.35 |
| I htink this is running in page mode | 16:09.43 |
| Yeah its playing back the pattern that does it. | 16:10.12 |
| The only tgime I see clist_opencalled is for the pattern, not for the device | 16:10.33 |
chrisl | Ah, well, that's sort of just one "band"....... | 16:10.53 |
kens | Yes, except that we never even create a 'normal' clist at all | 16:11.10 |
| Odd, but.... | 16:11.17 |
tor8 | Robin_Watts: in fz_begin_tile_id, if the device has no begin_tile or it fails, ret is returned undefined | 16:17.26 |
Robin_Watts | tor8: My commit should fix that. | 16:17.42 |
| 09a165e4f469afe4ad4852211be73006dd905156 | 16:18.15 |
tor8 | Robin_Watts: d'oh! | 16:18.22 |
| how did you find that? I just ran into it while cleaning up the interpreter... | 16:18.48 |
Robin_Watts | Joseph pointed it out to me :) | 16:19.00 |
jogux | the ios build threw it up as a warning | 16:19.02 |
tor8 | coincidences! | 16:19.06 |
jogux | which was obvious once I'd disabled the other warnings that we didn't care about :-) | 16:19.16 |
Robin_Watts | so this 999999 page file thing... | 16:19.28 |
| mutool clean, even without any -g's runs out of memory on it. | 16:19.44 |
| In that it hits the 2Gig max for 32bit windows apps. | 16:20.10 |
| I am trying to figure out how much we care. | 16:20.23 |
tor8 | me personally? not one bit. recompile for 64-bit and be happy. | 16:20.50 |
chrisl | Robin_Watts: does it do a lot of reallocing? | 16:21.01 |
Robin_Watts | I'm trying the 64bit thing now. | 16:21.21 |
| chrisl: don't think so, why? | 16:21.31 |
chrisl | Robin_Watts: I was just wondering if it was really using 2+Gb, or encountering address space fragmentation | 16:22.03 |
Robin_Watts | I suspect it's true. | 16:22.25 |
kens | Its a big file, 300Mb compressed | 16:22.40 |
| and nearly 1 million pages | 16:22.52 |
| Gah, the ICC memory is specially cleaned up in the pattern code when the tile is released | 16:26.50 |
| SO for the clist the device can be closed, but still hang around like Banquo's ghost | 16:27.40 |
| marcosw the commit you referenced in bug #695851 doesn't seem to exist for me | 16:28.28 |
chrisl | kens: the link generation only works for ghostpdl repo commits | 16:29.10 |
kens | Oh fair enough | 16:29.25 |
chrisl | Which is why I prefer to use URLs | 16:29.47 |
tor8 | kens: you booked something yesterday that I didn't have time to ask about | 16:33.51 |
kens | I did ? | 16:33.59 |
| lift pass maybe ? | 16:34.05 |
tor8 | yeah, that sounds like it | 16:34.16 |
kens | Did it online as its cheaper, pick up the pass form the lift office on the day. | 16:34.33 |
| $240 for three days, IIRC its $120 per day at the office | 16:34.52 |
tor8 | kens: right. you got a link to the place you booked from? | 16:35.35 |
kens | I have an email let me look | 16:35.49 |
| http://www.coppercolorado.com | 16:36.13 |
| Copper Mountain snow store | 16:36.21 |
tor8 | so what should I get? | 16:37.17 |
kens | I went for a 3 day pass (3 of 5 I think it says) | 16:37.33 |
| You have to pick it up on the first day (clealry I picked Tuesday) and then its any 3 of the next 5 days | 16:38.07 |
| From'plan & buy' select lift tickets | 16:39.03 |
tor8 | if I click Tuseday in the calendar, it says Starting Monday, March 02... | 16:39.04 |
kens | Really ? | 16:39.12 |
| My confirmation says 03/03/2015 | 16:39.35 |
| Fromthe lift ticket page I selected prepurchase online | 16:40.43 |
| Then 'Lifts' + 'Adult' + '3' | 16:41.04 |
| THen clicked Tuesday 03 March | 16:41.18 |
| press the blue select button and then the green 'view itenerary button' | 16:41.41 |
tor8 | yeah, if I do that and click Tuseday 03, it ends up as monday | 16:41.50 |
kens | yesterday I got confused because the dialog didn't go away until I clozed it | 16:42.07 |
tor8 | but if I click wednesday it ends up as tuesday | 16:42.16 |
kens | Bizarre! | 16:42.24 |
| Where do you see the Monday/Tuesday ? | 16:42.32 |
| I haven't gone to the checkout yet | 16:42.42 |
| and my browser doesn't show a date there | 16:42.51 |
| sorry it does, and mine says 03/03/2015 | 16:43.04 |
tor8 | Copper Mountain Snow Store | 16:43.14 |
| Adult (13-64) 3 Day Lift Ticket (valid 3 of 5 days) | 16:43.14 |
| 03/03/2015 in the shopping basket "Selected Items" if I add the one when I click wed | 16:43.14 |
kens | just above the blue 'remove' link | 16:43.16 |
| Wild, I clikced Tuesday and it was happy to show me 03/03/2015 | 16:43.42 |
tor8 | might be confused by local time zones | 16:44.05 |
kens | I suppose that's possible, what browser are you using ? I'm using Chrome | 16:44.19 |
tor8 | firefox | 16:44.31 |
kens | Well it could be the browser I suppose | 16:44.42 |
tor8 | but I have it showing 03/03/2015 3 day lift ticket (valid 3 of 5 days) | 16:44.46 |
kens | I think that's probably good enough | 16:44.56 |
| You have to print out the email and take it along anyway | 16:45.19 |
tor8 | so I guess if it had said 03/02/2015 I would still have been okay | 16:45.20 |
kens | No, because you have to pick the ticket up on the first day | 16:45.41 |
| So say the rules somewhere | 16:45.49 |
tor8 | ah. good catch then. | 16:45.59 |
kens | Definitely! I'm glad to see you were awake there :-) | 16:46.23 |
| I suspect they'd probably have honoured it anyway if you plead ignorance and stupid foreign tourist status | 16:46.55 |
chrisl | Or shoot you.... ;-) | 16:47.17 |
kens | Or it could be I'm interpreting the written word too stringently | 16:47.20 |
tor8 | chrisl: wrong state ;) | 16:47.23 |
chrisl | tor8: good point | 16:47.36 |
kens | suspect Colroadans carry guns too | 16:47.39 |
Robin_Watts | http://www.bloomberg.com/news/articles/2014-04-01/colorado-town-considers-letting-residents-shoot-at-drones | 16:49.11 |
kens | THat's from last year isn't it ? Or is it a new bunch..... | 16:49.29 |
Robin_Watts | from last year. but I bet those yahoos haven't packed up and moved states... | 16:50.03 |
kens | No I wouldn't think so | 16:50.12 |
| Didn't somewhere in Texas actually pass a law offering hunting permits for drones ? | 16:50.30 |
henrys | it's surprising how many guns are confiscated at the airport security here. People are liike "What's the problem?" | 16:50.35 |
kens | Yeah I saw that in South Africa a lot too | 16:50.48 |
tor8 | kens: okay, that's that done. anything else I need to do? I got myself a new jacket, pants, gloves and goggles for skiing yesterday so I think I'm all set dress-wise. | 16:50.57 |
Robin_Watts | In SA they have (or had) gun lessons at school. | 16:51.14 |
kens | tor8 you have some kind of neck warmer ? Not a big deal you can always get something on the spot, they aren't epensive | 16:51.24 |
Robin_Watts | http://www.7dayshop.com/products/7dayshop-polar-fleece-snood-neck-warmer-scarf-black-SNOODBLK | 16:51.59 |
chrisl | Robin_Watts: did you get those neck warmers you ordered for all of us? | 16:51.59 |
kens | Robin_Watts : I wen to Sun City and was quite relieved to see they made the residents check their guns on entry..... | 16:52.06 |
tor8 | kens: I do, I have a long wool scarf and a face mask if needed. | 16:52.07 |
Robin_Watts | chrisl: I did. I have yours here. | 16:52.09 |
chrisl | Excellent, much appreciated :-) | 16:52.18 |
kens | tor8 sounds fine to me then | 16:52.19 |
| I'll probably be bringing 2 anyway, one fleecy one not | 16:52.46 |
Robin_Watts | chrisl: At the moment, the only thing we don't have that we need are face masks. | 16:52.48 |
| http://www.amazon.co.uk/dp/B00HFT4FNW/ref=dra_a_rv_ff_fx_it_P2000_1000?tag=dradisplay0bb-21&ascsubtag=b0395178ceabc60d7d8f795e86dee6b8_S&dra_hfr=x&dra_ohs=24-48 | 16:52.59 |
| That's the cheapest one I could see that looks halfway decent. | 16:53.09 |
kens | That is very cheap | 16:53.22 |
Robin_Watts | but unless we all buy one, we're going to be finding a rental place anyway, so... | 16:53.24 |
kens | I can leave a pair with you if that helps | 16:53.40 |
| As I'll not need my goggles again till next winter | 16:54.01 |
Robin_Watts | kens: We need 5 pairs to avoid having to find a rental place. | 16:54.06 |
kens | Hmm, can't do 5, sorry :-) | 16:54.18 |
Robin_Watts | I suspect that finding a rental place in vail won't be hard. | 16:54.40 |
chrisl | I got a pair on ebay like these: http://www.amazon.co.uk/Rayzor-Professional-SnowBoard-Mirrored-Anti-Glare/dp/B002MPIQJ8 | 16:54.40 |
kens | Could do 3 I htink, we do have some spare pairs but I wouldn't want to use them unless it was an emergency | 16:54.42 |
Robin_Watts | chrisl: Can you wear sunglasses under them? | 16:55.05 |
| chrisl: So you have a pair already? | 16:55.30 |
chrisl | Robin_Watts: Yes I have a pair, and I wouldn't want to try wearing glasses under them | 16:55.50 |
Robin_Watts | Then maybe I will order a couple of the amazon ones... | 16:55.57 |
| Ah, wearing glasses under them is an absolute requirement. | 16:56.12 |
chrisl | Not for me ;-) | 16:56.31 |
Robin_Watts | Conceivably I could just wear normal glasses, not sunglasses, but they'd need to be decently tinted already. | 16:56.37 |
| chrisl: I don't trust these new fangled contact lens things. | 16:56.48 |
| and you don't want me on a snowmobile without glasses :) | 16:57.05 |
kens | Most goggles would be pretty tricky to wear glasses under | 16:57.56 |
chrisl | It's a bit hard for me to check, the glasses are quite big, so not really representative | 16:58.13 |
Robin_Watts | kens: hmm. how the hell am I gonna cope then? | 16:58.17 |
kens | We have one friend with prescription gogles, and one who just uses sunglasses. Fortunately Austria is not so cold | 16:58.20 |
Robin_Watts | Then I'm definitely buying them off amazon. | 16:58.39 |
kens | Robin_Watts : There are *some* pairs that work, but I think you would have to try them out | 16:58.43 |
Robin_Watts | If I have to chop a section out of the frame to make my glasses fit, I want to do that before I travel :) | 16:59.00 |
kens | Like chrisl its not a problem for me so I'm nt well informed | 16:59.08 |
chrisl | My glasses *just* fit under those goggles, but not very comfortable. Smaller specs might work | 16:59.38 |
kens | Or you could trim the frame as Robin suggests ? | 16:59.56 |
Robin_Watts | I have a prescription snorkel mask... :) | 17:00.04 |
kens | Or is it the lens to goggles psacing ? | 17:00.09 |
| Robin_Watts : wear that :-) | 17:00.24 |
chrisl | It was around the edge of the frame that there was contact, so I reckon Robin's normal glasses would work. But the only way to be sure would be to try them | 17:01.01 |
kens | Ok lets see if this fixes the memory problem, fingers crossed..... | 17:01.01 |
| chrisl yeah that's what I thought. I have seen people use glasses under goggles, but I get the impression its a 'suck it and see' job | 17:01.29 |
| Buying a cheap pair and customising before departure looks like a decent idea to me | 17:02.05 |
chrisl | Pass the dremel :-) | 17:02.32 |
| Time for me to shuffle off..... | 17:04.51 |
Robin_Watts | I suspect that I have sufficient face covering that I can survive with just sunglasses if required, but I don't want to fall foul if the snowmobile renters are awkward. | 17:05.03 |
jogux | robin_watts: mountainwarehouse are selling off their ski goggles cheap at the moment if you can get into a branch to try them with your glasses. | 17:05.27 |
kens | I'd have thought the wind might make it awkward to see without goggles | 17:05.30 |
jogux | (I got a pair for £15 yesterday) | 17:05.37 |
kens | is tmepted ot hit mountain warehouse this weekend | 17:05.55 |
| Or possibly snow and rock, since I';ll be in London anyway | 17:09.37 |
Robin_Watts | I've bought 2 different pairs from amazon. Hopefully they will arrive in time. If not, I'll just return 'em. | 17:15.38 |
kens | Should be itme I'd have thought | 17:16.31 |
Robin_Watts | delivery between March 2-5. | 17:16.41 |
| and we fly on the 5th I think. | 17:16.46 |
kens | Well I'mflying on the 2nd :-) | 17:17.04 |
Robin_Watts | failing that, it's possible there will be stores in denver that sell stuff :) | 17:17.25 |
kens | I htink that may possibly be true :-) | 17:17.36 |
| and Helen will find them for you :-D | 17:17.49 |
henrys | well milk shakes within walking distance of the hotel, the bad news is nothing else is within walking distance. | 17:19.08 |
| https://www.google.com/maps/place/Moonlight+Diner/@39.8006444,-104.7761786,14z/data=!4m9!1m6!2m5!1srestaurants!3m3!1srestaurants!2sHampton+Inn+Denver-International+Airport,+6290+Tower+Rd,+Denver,+CO+80249,+United+States!3s0x876c65cbc0729b65:0xa447aec3f4a3f9e0!3m1!1s0x0000000000000000:0x4ef9bcd5ba9170b5 | 17:19.37 |
Robin_Watts | We're at the Hampton Inn? | 17:21.36 |
| Dining in a Diner suits me just fine. And there is an Italian place not too far away too as a backup. | 17:22.15 |
| And papa johns can deliver pizza. We're sorted! | 17:22.38 |
henrys | the old denver airport was right in Denver, very convenient, DIA not so much. | 17:24.12 |
| largest airport in the US needs cheap land I suppose... I've heard other reasons but I think that's the big one. | 17:25.11 |
kens | OK thankfully that solved my memory problems, chrisl_away thanks for your help (for the logs) | 17:25.26 |
| I'm heading off now, gnight all | 17:25.35 |
Robin_Watts | henrys: is it the largest because when you have limited air pressure, you need a longer runway? :) | 17:25.48 |
kens | Its all those skiers..... | 17:26.00 |
tor8 | Robin_Watts: the goggles I got are made to fit over glasses. you're welcome to try and borrow mine if you need to. | 17:29.38 |
Robin_Watts | tor8: ah, thanks. we'll see how the amazon ones work out, but I may take you up on that, thanks. | 17:30.06 |
henrys | Robin_Watts: I'm sort of wondering how you'll be jet lagged a mile high at the dreaded meeting ;-) | 17:30.58 |
tor8 | Robin_Watts: yeah. they fit my glasses with a bit to spare. it's the area around the nose that's the main bother with goggles, if the lens is far enough out not to touch the glasses | 17:31.01 |
Robin_Watts | tor8: Ok, I have a 64bit build working on windows. Will commit the msvc solution changes required for that in a mo. | 17:32.10 |
| henry: 5000 feet? Hah. I've slept at over 15000, and been to 17000. | 17:40.25 |
henrys | Robin_Watts: but not with the meeting going on ;-) | 17:41.18 |
Robin_Watts | The airport at La Paz is 12000 ft, and we flew there direct from Miami. | 17:41.35 |
| Now *that* was a shock :) | 17:41.42 |
| I hear you have some really 'radical' herbal remedies in Colorado anyway. | 17:42.43 |
henrys | when you drive to vail you can take hwy 6 if it's open that will take you over loveland pass well about 11K and a very scenic but treacherous at time drive | 17:47.45 |
| Robin_Watts: yeah I thought I'd bring the meeting snacks and we'd finish up early ;-) | 17:48.27 |
Robin_Watts | If the weather is good (and my passengers are up for it), that sounds great. | 17:48.35 |
| henrys: dumb question, but any car we hire is going to have snow tyres, right? | 17:50.06 |
rayjj | henrys: are chains required ever ? | 17:50.29 |
henrys | Robin_Watts: I would think but I'd really double check for what you're doing. | 17:50.37 |
Robin_Watts | I suspect they'll try and get us to upgrade into some mammoth premium SUV thing. | 17:51.13 |
henrys | rayjj: just trucks usually, if you need chains you need to check into a hotel and not drive. | 17:51.13 |
Robin_Watts | I rented a car in nevada once, and ended up driving it in a blizzard in... idaho? | 17:52.06 |
henrys | I'm saying all this and their could be kids in bikinis on the slopes the way the weather goes around here... | 17:52.32 |
| Robin_Watts: how many are you hauling? | 17:54.33 |
rayjj | Robin_Watts: we found that a full size van was cheaper than even a minivan. That way we have plenty of room for gear inside | 17:56.54 |
Robin_Watts | me+4 | 17:57.00 |
| rayjj: I've booked already. | 17:57.39 |
mvrhel | looks like the forecast is for some major snow this coming weekend through monday | 18:04.51 |
| should be a good time | 18:04.56 |
Robin_Watts | Aha. Dodge Grand Caravan or Similar | 18:06.22 |
| tor8, jogux, paulgardiner: The win64 build commit is on robin/master for review now. | 18:38.09 |
| along with a SEGV fix. | 19:05.13 |
mvrhel_laptop | watched the 3rd episode of black mirror last night. pretty crazy. google glass on steroids | 19:12.53 |
hyper_ch | good evening | 19:14.51 |
Robin_Watts | hyper_ch: evening. | 19:16.08 |
hyper_ch | the ghostscript-workaround-to-check-if-scanner-did-well works fine :) | 19:16.36 |
henrys | mvrhel: going to have a look at that after the Americans looks pretty good | 19:19.41 |
hyper_ch | look at? | 19:20.41 |
Robin_Watts | hyper_ch: different conversation. | 19:22.06 |
hyper_ch | The Americans sounds like a tv show | 19:22.19 |
Robin_Watts | It is. | 19:22.29 |
hyper_ch | so I wonder what the other one is | 19:22.37 |
Robin_Watts | Black Mirror | 19:22.48 |
hyper_ch | never heard of :) | 19:22.54 |
malc_ | Robin_Watts: how does one obtain document information via fz_meta? | 19:29.24 |
Robin_Watts | malc_: I prefer to answer questions in public, rather than on a private tab, cos the channel is logged, and can be searched by google. | 19:29.38 |
| Have a look at platform/x11/win_main.c | 19:30.02 |
| in the dloginfoproc | 19:30.11 |
| at line 455. | 19:30.21 |
| There we are passing in a buffer (buf) and the buffer length (256) to fz_meta, with the FZ_META_FORMAT_INFO reason code. | 19:31.12 |
| If you get a negative reply, then the document handler doesn't support that reason code. | 19:31.42 |
| Otherwise, you get the format info returned in 'buf'. | 19:32.05 |
malc_ | Robin_Watts: thanks, how to get the upper bound for the buf? | 19:32.38 |
Robin_Watts | malc_: hmm, that's not quite right. | 19:32.55 |
| oh, yes, it is. | 19:33.27 |
| A bit further down that function, see the SET_UTF8 macro. | 19:33.43 |
| That calls fz_meta again, this time with FZ_META_INFO. | 19:34.15 |
| On entry, *(char **)buf is a pointer to the string that identifies what info you want. | 19:35.02 |
| There is no way to get the upper bound, currently. | 19:36.51 |
| But the call guarantees not to overrun the end of the string. | 19:37.11 |
| s/string/supplied buffer/ | 19:37.17 |
rayjj | wow! for gs to just load the 1m page PDF uses 1,258 Mb RAM and takes 283 sec on my i7-2640 2.8GHz. Whatever would possess someone to put 1M pages in a single PDF ? | 19:37.29 |
Robin_Watts | so you could call it and if the buffer is filled, try with a larger buffer. | 19:37.57 |
mvrhel_laptop | oh crap. it is looking like the change that I did to "lock" the colorants was not needed. It appears that if the "order" of the colorants was specified then that prevented additional colorants from being added | 19:38.58 |
Robin_Watts | mvrhel_laptop: Did you update your stuff for the mupdf API change? Or was fred doing that? | 19:39.52 |
mvrhel_laptop | Robin_Watts: I need to do that. I will jump on that next | 19:40.10 |
Robin_Watts | ok, np. | 19:40.19 |
rayjj | mvrhel: FYI, I started looking into bug 695485, so I took it from you. | 19:44.24 |
mvrhel_laptop | rayjj: thank you! | 19:57.27 |
| just got done looking over the devicen sep stuff with nclr profiles and the specification of SeparationColorNamesand SeparationOrder. I am going to spend a little time reworking this mess. For now, though I think Stefan should have a workable approach. | 19:59.03 |
| Now though I will take a break and update for the mupdf api change and make a new installer with gsview with a couple things turned off so we can release the beta | 19:59.49 |
| things that don't quite work right related to printing | 20:00.03 |
| then I still want to wrap up my SOT project that is only 1/2 done | 20:00.16 |
| we may need to have a hack-a-thon in the evenings at copper next week... | 20:00.45 |
| breaking for lunch now | 20:01.33 |
fredross-perry | hi all. back at it. Iâll look at the white pace issue now. | 21:22.45 |
Robin_Watts | hi fred. | 21:26.33 |
| tor is a whitespace nazi :) | 21:26.40 |
fredross-perry | I can see that | 21:29.15 |
henrys | jogux: did marcosw send you printer test results? | 21:32.17 |
fredross-perry | I thought I read that my changes were not good, so I expected NOT to see them in the main repo. But there they are. | 22:04.19 |
Robin_Watts | fredross-perry: the first of your changes were fine, and they are there. | 22:18.38 |
| the ios ones. | 22:18.41 |
| the library ones aren't there yet, AIUI. | 22:18.49 |
fredross-perry | I know. I had meant to put thos in a branch, so thatâs what Iâve done today. | 22:19.15 |
Robin_Watts | fredross-perry: Ah, right. | 22:19.36 |
fredross-perry | they were not meant to be reviewd | 22:19.41 |
henrys | fredross-perry: we weren't sure what was going on, because you didn't come to the IRC meeting yesterday, maybe I missed something about you being away. | 22:40.11 |
fredross-perry | oh, sorry. I had something else at the last minute and forgot to let you know. | 22:40.56 |
henrys | fredross-perry: no meeting next Tuesday, many are on vacation and we don't have a meeting the week before the in person staff meeting. | 22:41.01 |
fredross-perry | got it. Thanks. | 22:41.16 |
henrys | fredross-perry: no problem. | 22:41.47 |
| Forward 1 day (to 2015/02/26)>>> | |