| <<<Back 1 day (to 2013/01/01) | 2013/01/02 |
sebras | marcosw: thanks! :) | 01:42.57 |
Robin_Watts | has spent the entire year either on drugs (Contact is really good stuff) or feeling like he's been mugged. And then at midnight last night, my glasses snapped. My optician is away until Monday. Let's hope it gets better :) | 09:14.43 |
kens | Robin_Watts : the year can't get much worse.... | 09:26.58 |
| Hmm, that's odd, my regression test didn't seem to send me an email | 10:04.22 |
| ah 'log file too big' | 10:04.36 |
| I don't believe my change should have caused any increase in the log file size. | 10:05.28 |
| Maybe I need a marcosw :-( | 10:06.17 |
kens | tries a regression test of a clean HEAD | 10:07.37 |
| OK so a clean HEAD fails the same way, its not my code. I guess I need marcosw now. | 10:20.04 |
| But I'm off out for a couple of hours, I'll email him later | 10:20.40 |
tor8 | Robin_Watts: ping. | 10:40.11 |
sebras | tor8: pong. | 10:45.05 |
Robin_Watts | tor8: groan | 10:49.22 |
tor8 | Robin_Watts: under the weather today? | 10:59.14 |
| Robin_Watts: thirdparty/freetype artifex branch, is that of your doing? | 10:59.29 |
Robin_Watts | yeah. horrible cold. feel like I've been mugged. | 10:59.33 |
| tor8: yes. | 10:59.37 |
tor8 | Robin_Watts: cd /home/git/thirdparty/freetype.git; git remote update | 10:59.51 |
Robin_Watts | just pulled in some changes from freetype to fix one of the bugs from customer 395. | 11:00.17 |
| There are various reviews up for you. | 11:00.35 |
tor8 | Robin_Watts: I'd do fz_read_best differently. I'd make it never throw, but in the fz_read_all check if truncated and throw there instead. | 11:04.35 |
Robin_Watts | tor8: OK. | 11:06.03 |
tor8 | Robin_Watts: the rest LGTM. now I'll stop annoying you and let you rest. I had a nasty cold all christmas, probably same one as you got now telling by your description :( | 11:07.57 |
Robin_Watts | tor8: Thanks. I'm planning to work as much as possible today. | 11:14.00 |
| Annoyingly I snapped my glasses last night, and my optician is away until monday, so I'm having to live with an old pair. | 11:14.29 |
phiscribe | that sucks. last time i broke mine i closed the hood of a car on them. | 11:15.18 |
Robin_Watts | phiscribe: oops. | 11:15.29 |
phiscribe | anyone know where i can find out about the adobe photoshop file format .csh (custom shape file). with the intention of just converting one to some other vector format like eps. can't seem to find much in googling. the header of the file doesnt look like adobe or pdfish stuff. says M.a.l.t. | 11:17.53 |
Robin_Watts | tor8: ping | 11:42.52 |
tor8 | Robin_Watts: just heading out for lunch, I'll be back in an hour | 11:43.20 |
Robin_Watts | fz_read_best is documented to do exactly the same as fz_read_all does in the case where truncated is NULL. | 11:43.28 |
| ah, ok. | 11:43.33 |
| I'd have to break that if I move the truncated handling to fz_read_all. | 11:43.49 |
| Also, I'd lose the ability to rethrow a meaningful error. | 11:45.46 |
tor8 | Robin_Watts: can you do rethrow outside of catch? | 13:27.30 |
Robin_Watts | tor8: Possibly, but it's bad style. IMHO. | 13:46.56 |
| kens: Have you run temp/tests_private__comparefiles__Bug693490.pdf.pdf.ppmraw.72.0.log locally? | 13:49.56 |
| cos it's that file that has the huge log. | 13:50.11 |
kens | Robin_Watts : No I haven't | 13:50.59 |
| How would it help me ? | 13:51.05 |
| The cluster would still barf on it | 13:51.13 |
Robin_Watts | Right. I suspect that someone has left some logging in that is causing log file explosions with that file. | 13:51.42 |
kens | I suspect you are right | 13:51.52 |
Robin_Watts | If you can reproduce the problem, you can see the logging in question. | 13:52.01 |
| Either you can then fix it, or find the commit that caused it and back it out. | 13:52.14 |
| or know who to point the finger of shame at. | 13:52.26 |
| It might be ec5b8eb ? | 13:55.11 |
kens | I guess I'll go and try to find out | 13:55.44 |
| Hmm, yes that does look likely | 13:56.54 |
Robin_Watts | tor8: I can't see an obvious way to slice this that makes nicer code. | 14:01.15 |
| I dislike fz_rethrowing from outside the fz_catch, because the 'most recent error' may have changed. | 14:02.36 |
| We call fz_trim_buffer, for example, and if that uses try/catch etc, then we might end up throwing something spurious. | 14:03.04 |
tor8 | Robin_Watts: right. | 14:37.59 |
| Robin_Watts: maybe put fz_read_best in fitz-internal.h then? | 14:38.30 |
Robin_Watts | tor8: ok. | 14:41.05 |
faenil | good afternoon :) | 14:47.27 |
| was looking for some help with gs | 14:47.33 |
Robin_Watts | faenil: Hi. | 14:47.33 |
faenil | I want to convert a color pdf to grayscale, but when I do so | 14:47.47 |
| I get some errors from adobe reader in a few pages | 14:47.57 |
Robin_Watts | gs cannot convert a color pdf to a greyscale pdf, AIUI. | 14:48.04 |
faenil | so I should not use | 14:48.29 |
| gs -sOutputFile=normal_bw.pdf -sDEVICE=pdfwrite -sColorConversionStrategy=Gray -dProcessColorModel=/DeviceGray -dCompatibilityLevel=1.7 -dNOPAUSE -dBATCH normal.pdf | 14:48.30 |
| ? | 14:48.31 |
Robin_Watts | I think you need kens to comment on that. | 14:49.26 |
kens | SOrry, was fetching coffee, justa moment | 14:49.37 |
faenil | alright ;) | 14:49.57 |
kens | faenil that basically won't work in the current code | 14:50.06 |
faenil | kens oh :/ | 14:50.15 |
| it actually creates a grayscale pdf | 14:50.22 |
kens | Hopefully the next release will allow that, if I get the code finsihed | 14:50.28 |
faenil | but I have 3 pages with errors (out of 169) | 14:50.32 |
| I see...so it works but it creates errors? | 14:50.45 |
kens | faenil if it does you are in luck, buyt in the general case it won't work | 14:50.46 |
faenil | I see | 14:50.50 |
| strange, it's a command I found on forums on old threads :) | 14:51.01 |
kens | The next releasee I'm working on integrating the Colour Management System | 14:51.08 |
faenil | cool, ok ;) | 14:51.23 |
| kens, what do you recommend for that purpose? pdf2ps and back? | 14:51.36 |
kens | If I get it finished in time, you shoudl eb able to do this | 14:51.40 |
| faenil I reccomend not trying it | 14:51.50 |
faenil | :O | 14:51.59 |
| well, I have to :( need to print something to study | 14:52.08 |
| and the guy wants a grayscale pdf... | 14:52.13 |
kens | I know of no reliable way (using open source software) to do this | 14:52.13 |
| You could convert to a gray image and print that | 14:52.32 |
faenil | mmm I see...weird, I would have never imagined there were such problems :) | 14:52.37 |
kens | Consider an Indeexed colour space wiiht a base of DeviceN which has a ICCXBased alternate | 14:53.10 |
faenil | :) | 14:53.23 |
kens | COnverting that to DeviceGray is problematic | 14:53.25 |
faenil | I see... | 14:53.29 |
kens | CIEBased colours are difficult too. You need a proper CMS for all these things | 14:53.49 |
faenil | alright | 14:54.00 |
| I just need a readable result :) | 14:54.09 |
kens | THe current code works 'most of the time' but is not (in my opinion) of commercial grade | 14:54.12 |
faenil | the code you're working on you mean? | 14:54.26 |
kens | No, the current release works 'most of the time' | 14:54.38 |
faenil | oh ok | 14:54.41 |
kens | THe code I'm working on should be100% reliable | 14:54.51 |
faenil | cool stuff | 14:54.58 |
kens | Currently there are cases where we just 'give up' | 14:55.09 |
faenil | kens, what if I accepted to convert it to image? (as I have to print it today :/ ) | 14:55.12 |
| kens, oh.. | 14:55.18 |
kens | To 'convert to image' use an image device instead of pdfwrite | 14:55.32 |
Robin_Watts | faenil: You can certainly render to greyscale images. | 14:55.41 |
kens | I thikn there's a tiffgray device | 14:55.43 |
Robin_Watts | gs -sDEVICE=tiffscaled8 -r900 -dDownScaleFactor=3 -o out%d.tif in.pdf | 14:57.22 |
kens | Ah, marcosw has fixed my cluster trouble | 14:57.37 |
faenil | Robin_Watts, will try that, thanks | 14:57.41 |
Robin_Watts | That'd be what I'd try. | 14:57.48 |
faenil | Robin_Watts, oh...but that is a lot of images | 14:58.17 |
| I thought you guys meant a pdf with images | 14:58.23 |
kens | one per page | 14:58.26 |
faenil | kens, yeah noticed | 14:58.30 |
kens | If you realy want a PDF< open the images and print them to PDF (or to PostScript and hten use pdfwrite to make a PDF) | 14:58.54 |
faenil | yeah but it's 169 images xD | 14:59.11 |
| isn't there a pipe to do it all together? :D | 14:59.22 |
kens | Not 'out of the box' | 14:59.42 |
faenil | okay | 14:59.50 |
kens | There are some PS programs for reading image formats | 14:59.52 |
faenil | I'll make some bash script otherwise | 15:00.28 |
| but first of all I'll try pdf2ps and back, and see if that creates monsters | 15:00.43 |
kens | I don't believe that will work any betrter than pdfwrite (unless it is using a differnt program than GS+ps2write) | 15:01.10 |
faenil | kens, trying in the meanwhile, who knows :/ | 15:02.26 |
kens | :) | 15:02.31 |
faenil | or I could just take those 3 pages out | 15:02.33 |
| and get done with it :P | 15:02.37 |
| or print the color version of those 3 pages :D | 15:02.48 |
| is it just me or it seems ps2pdf will take like hours for a 10mb pdf? :O | 15:04.59 |
| it's raising few kbs at a sec | 15:05.06 |
| 0.7KB/sec | 15:05.23 |
| :/ | 15:05.25 |
Robin_Watts | tor8: I'm having submodule woes here. | 15:07.38 |
faenil | kens, done...no problems! :/ | 15:08.20 |
| kens, though quality is worse now...anything I can do? | 15:08.30 |
kens | OK good stuff | 15:08.33 |
faenil | must befont problem | 15:11.18 |
| kens, anyway I guess it's an optimization/compression issue | 15:14.59 |
| because this one is 27mb instead of 9mb | 15:15.05 |
kens | Possibly | 15:15.07 |
faenil | and it has no errors, even though text sucks | 15:15.13 |
kens | All converted to images, so bigger and ugly | 15:15.33 |
| No surprise | 15:15.37 |
faenil | seems like text isn't antialiased | 15:15.57 |
| is a ps supposed to have so bad fonts? | 15:16.05 |
kens | THey may be converted to images | 15:16.14 |
| depsnds on what you used to do it | 15:16.19 |
faenil | pdf2ps, no arguments | 15:16.26 |
| it uses gs, but I don't know the pars it passes | 15:16.38 |
kens | probably uses pswrite, bnad idea | 15:16.45 |
faenil | mm okay | 15:17.17 |
| you know what, I'll stick the 3 colour images, and get done with it :) | 15:17.27 |
| 3 colour pages | 15:17.37 |
| won't buy adobe professional but such a task :D | 15:18.02 |
| also guys | 15:22.20 |
| I was thinking about doing a pdf viewer | 15:22.27 |
| with ink annotations and highlights | 15:22.32 |
| for tablets | 15:22.34 |
| is mupdf okay for the task? or is there any other library you'd suggest? | 15:22.54 |
| I've seen poppler already has Qt bindings, which is interesting | 15:23.02 |
| there is also podofo, but it does not render, as I've read | 15:23.53 |
| what would you suggest for my purpose? | 15:24.06 |
| it's an app for tablets, to take ink notes on PDFs and highlight text | 15:24.24 |
| accademic purposes mainly | 15:24.32 |
sebras | appreciates the logs so he can find -dDEVICE(HEIGHT|WIDTH)POINTS again. gs doesn't really have an apple-ish UI now, has it... ;) | 15:27.38 |
faenil | kens, Robin_Watts any thought on my plans? :D | 15:27.52 |
Robin_Watts | faenil: Mupdf sounds appropriate for what you want. | 15:28.18 |
kens | I'm inclined to agree with Robin_Watts a viewer is a *lot* of work | 15:28.37 |
Robin_Watts | Whether it's academic or not, as long as you abide by the GNU GPL you will be fine. | 15:28.38 |
faenil | alright, I couldn't find info about how to highlight text so I thought maybe it had to be implemeted | 15:28.54 |
| Robin_Watts, I meant academic for the user | 15:29.17 |
| I'd like to sell it ( later, maybe) | 15:29.26 |
Robin_Watts | faenil: In MuPDF? It's there (we use it to highlight search results etc). | 15:29.28 |
faenil | Robin_Watts, alright | 15:29.39 |
Robin_Watts | faenil: Right, so you'd need a commercial license. | 15:29.42 |
faenil | Robin_Watts, GPL already allows that, doesn't it | 15:29.52 |
| share source, sell app | 15:29.58 |
Robin_Watts | Sure, but IME you'll find it hard to sell many copies if people can just download the source and build their own. | 15:30.33 |
faenil | Robin_Watts, yeah I know, but it's just to earn something and give people a way to appreciate the work :) | 15:31.06 |
| I don't think I can afford a commercial licence, I'm a uni student :) | 15:31.36 |
| Is there a place where I can know more about how to use mupdf? I've only found header on the website under "documentation" | 15:32.32 |
| headers* | 15:32.37 |
| I know almost nothing about pdf file format... | 15:33.01 |
| nothing at all about mupdf | 15:33.15 |
Robin_Watts | faenil: There is limited documentation in the 'docs' directory of the source, but mostly, look at the example apps, and read the headers. | 15:33.36 |
faenil | Robin_Watts, the example app seems very complex, from the pdf noob pov :/ | 15:34.01 |
| I mean, no explanation of any call | 15:34.11 |
Robin_Watts | faenil: doc/example.c | 15:35.36 |
faenil | Robin_Watts, look great, for rendering :) | 15:37.23 |
| is there anything like that for annotations? | 15:37.31 |
Robin_Watts | No. Because the functionality you are talking about (for adding annotations etc) is not quite there yet. | 15:37.56 |
faenil | Robin_Watts, oh...so how can I use it for my purpose? :D | 15:39.04 |
Robin_Watts | faenil: At the moment, adding annotations etc, requires some knowledge of the underlying PDF structure. Come back in a month or so and ask again, and we may have exposed a nicer API for you. | 15:39.58 |
faenil | I see...perfect timing, I think I'll start making that app in february | 15:40.30 |
| so that should be cool | 15:40.37 |
| otherwise, I'll learn something about PDF structure and do it myself :) | 15:41.18 |
| in that case, is there any good guide about pdf structure? | 15:41.32 |
Robin_Watts | The PDF Reference Manual. | 15:42.26 |
faenil | something like 10k pages :D | 15:43.08 |
| let's see | 15:43.10 |
| oh only 745 : | 15:44.21 |
| Robin_Watts, thanks ;) | 15:44.43 |
Robin_Watts | faenil: np. | 15:45.03 |
faenil | Robin_Watts, have you already tried porting mupdf to BlackBerry10 by any chance? | 15:45.12 |
| that's what I'll need to do | 15:45.18 |
Robin_Watts | tor8: Updated patch pushed. | 15:45.23 |
| faenil: Nope, but I believe others have. | 15:45.41 |
| tor8: plus another couple of changes. | 15:45.51 |
faenil | Robin_Watts, others in mupdf group? notice that BlackBerry10 isn't out yet | 15:46.09 |
Robin_Watts | OK. I believe that other people have ported mupdf to that OS or something. I can't remember any details. | 15:47.17 |
| but not from people within Artifex. | 15:47.34 |
faenil | Robin_Watts, alright, couldn't find any detail on Google, that's why I asked ;) | 15:48.34 |
| is there platform-specific code in mupdf? or should it just be a matter of using a different compiler? | 15:49.01 |
| or system libs | 15:49.10 |
Robin_Watts | mupdf is pretty portable. No compulsory system specific magic in there. | 15:50.04 |
faenil | cool | 15:50.12 |
henrys | isn't blackberry qnx - bug 692947? | 15:51.12 |
Robin_Watts | henrys: That was the TLA I was trying to remember :) | 15:51.31 |
sebras | henrys: yes it is, but their windowing-layer might not be photon (at least I'm not sure). | 15:54.09 |
| on blackberry I mean. | 15:54.55 |
faenil | henrys, oh cool, even though that's not BB10 most likely :) | 15:56.03 |
| 714kb patch file? holy crap :O | 15:56.12 |
sebras | faenil: but half of it just a hexdump of some icons or something. | 15:57.45 |
faenil | sebras, well 300kb is still pretty big isn't it :D | 15:58.07 |
sebras | it is. | 15:58.32 |
faenil | it's 6.5k LOC | 15:58.49 |
| the viewer that is :) | 15:58.59 |
| Ra+k | 16:01.05 |
| Robin_Watts, is there a reason why that was never merged? | 16:01.16 |
sebras | I guess the reasons for this not being merged so far is lack of test devices, incentive to do it and the fact that the viewer probably doesn't share enough code with the existing viewers. | 16:01.47 |
Robin_Watts | Yes. We lack the ability (or particular desire) to test it. | 16:01.49 |
faenil | alright ;) | 16:02.20 |
| anyway, it should work (TM) :) | 16:02.38 |
sebras | Robin_Watts: once the new pdfviewer layer is finished apps like this ought to be smaller at least. | 16:03.05 |
faenil | last question (sorry to bother...) would it make any sense to just make a tablet GUI for Okular? this applies to those who have already read its sourcecode at least :) | 16:04.33 |
Robin_Watts | Okular is (according to Professor Google) based on KPDF. So I'm not sure how that would work. | 16:07.06 |
faenil | alright | 16:07.21 |
| yeah it's based on KPDF and poppler | 16:07.28 |
Robin_Watts | mvrhel_laptop: Hey. How are you feeling? | 16:07.36 |
faenil | Robin_Watts, thank you very much for the patience :) | 16:07.56 |
| see you all guys, got to get back to studying compilers :/ | 16:08.10 |
mvrhel_laptop | Hi Robin_Watts . Pretty crappy still | 16:08.12 |
Robin_Watts | tor8: PDFRM section 3.9.3 talks about a "Functions" entry. | 16:08.19 |
| As far as I can tell we only look for a "Function" entry. | 16:08.29 |
| mvrhel_laptop: Commiserations. If it's any consolation I'm feeling similarly crap. | 16:09.00 |
mvrhel_laptop | Robin_Watts: sorry to hear that. | 16:09.40 |
Robin_Watts | My mother in law is 2 or 3 days further through the course of the cold we've got, and she just looked awful when she left this morning. So I have that to look forward to :( | 16:09.49 |
marcosw_ | henrys: I ran across an Adobe web site, http://acroeng.adobe.com/, that contains a large number of PDF test files. I've downloaded them and was thinking of running a comparison of Acrobat, Ghostscript, and muPDF but before I started wanted to see if you knew about the site and had looked at any of the files before? | 16:10.11 |
mvrhel_laptop | Stephanie just came down with this yesterday and is following me now... | 16:10.25 |
| bbiab | 16:10.33 |
Robin_Watts | ew. | 16:10.39 |
sebras | marcosw_: I mentioned that site long ago, so tor8 has known about it at least. don't know about henrys. | 16:11.19 |
| marcosw_: mupdf doesn't segfault on any of those files, but I'm not sure all of them are rendered perfectly. | 16:11.49 |
henrys | marcosw_:geez I missed it. | 16:12.02 |
sebras | Robin_Watts: load_stitching_func() parses "Functions"..? | 16:12.18 |
marcosw_ | sebras: not seg faulting is a good first step :-) | 16:12.20 |
Robin_Watts | sebras: Yeah, I'm off by 1 layer of structure here. Ignore me :) | 16:13.13 |
sebras | marcosw_: I haven't got gs and bmpcmp setup so I have had difficulties comparing them myself. | 16:13.14 |
| Robin_Watts: I know. I will. ;) | 16:13.26 |
henrys | marcosw_:while you are about I had cluster questions above about cron and all. | 16:15.32 |
marcosw_ | henrys: okay | 16:15.43 |
henrys | somewhere back in the logs - let me just repeat it. Why not one script with a loop at startup time instead of a syslog'd cron every minute? | 16:16.55 |
sebras | Robin_Watts: henrys: http://goo.gl/JEu9N apparently bb10 is based on qnx, but uses Screen for graphics rather than photon. | 16:17.05 |
marcosw_ | sebras: bmpcmp isn't well suited for running large comparisons, I have different comparison tool (based on fuzzy.c) that i used. I should check it into git but then I'd have to document it. | 16:17.14 |
henrys | I am trying to troubleshoot henrysx6 but syslog is of no help it is all cron entries. | 16:18.00 |
sebras | marcosw_: oh. well if you manage to run it on the adobe-files (avoid the ones named *32k* and *64k* as that refers to the number of pages), I'm quite curious about the results. | 16:19.07 |
faenil | sebras, exactly | 16:19.10 |
marcosw_ | henrys: If I had to do it again I probably wouldn't use a cron job, but it seemed like a good idea at the time. In any case, we'd still have to have a cron job running to make sure the cluster code was still running :-) but presumably it wouldn't have to run every minute. | 16:21.25 |
henrys | marcosw_:I've run various diagnostics and can't produce a failure on henrysx6 and I compile code with -j frequently. Does it usually fail with the large language switch build? | 16:22.32 |
kens | marcosw my cluster test just failed again | 16:22.52 |
| Not sure why yet | 16:23.07 |
marcosw_ | henrys: as I recall it fails with any of the builds | 16:23.09 |
| kens: I'll take a quick look. | 16:23.14 |
kens | It could be me this time ;-) | 16:23.30 |
marcosw_ | same error. odd why didn't the svn update I did work? | 16:23.49 |
kens | Umm, no idea.... | 16:23.56 |
marcosw_ | kens: no wait, the file I was looking at is several hours old. | 16:24.18 |
kens | It snot completed failing yet | 16:24.27 |
| But all the nodes aborted | 16:24.42 |
henrys | marcosw_:okay let me try again. | 16:24.51 |
kens | That's why I said I didn't know why it had failed | 16:24.56 |
| marcosw it has finsished and is still failing with log files too big | 16:25.38 |
marcosw_ | kens: sigh, another file has the same problem. | 16:25.50 |
kens | Looks that way yes | 16:25.55 |
| Bug690497.pdf | 16:26.04 |
marcosw_ | kens: okay, please try again. | 16:27.52 |
kens | OK :-) | 16:27.58 |
marcosw_ | This time for sure :-) | 16:28.00 |
| That was a "Rocky and Bullwinkle" reference, for those of you across the pond. | 16:28.27 |
kens | We get (or used ot) Rocky and Bullwinkle too | 16:28.43 |
marcosw_ | http://www.youtube.com/watch?v=e7mmrF-4rUE | 16:28.55 |
| I thought you only got Dallas and Baywatch, didn't realize the quality shows made it to you. | 16:29.35 |
kens | Our channels will buy anything if its cheap enough | 16:30.15 |
| got to fill the schedules somehow.... | 16:30.24 |
marcosw_ | kens: I thought you only had 3 channels? | 16:33.18 |
kens | You are out of date Marcosw, we've had 4 for ages | 16:33.32 |
| Of course, now we've gone digital we have as many as the US and nobody watches TV any more, we spend the entire evening flipping channels trying to find something that isn't terrible, or a repeat.... | 16:34.49 |
Robin_Watts | kens: This is why God gave us Sky+ (Tivo for the USians). | 16:39.36 |
kens | wwwEven more channels to flip through.... | 16:39.59 |
Robin_Watts | kens: Nah, set stuff to record. Never channel flip randomly again. | 16:40.43 |
sebras | Robin_Watts: yes, and watch when you retire (if you have kids). | 16:41.29 |
| that's what my former colleagues do at least. :) | 16:41.58 |
kens | 2 TVs :-) | 16:42.13 |
Robin_Watts | I am becoming more and more convinced that kids shouldn't be allowed to watch childrens TV, ever. | 16:42.33 |
kens | That's just a symptom of age :-) | 16:42.59 |
Robin_Watts | Brightly coloured blobs singing in stupid voices. We're breeding morons, and we need to be breeding smart people to earn lots, and pay taxes, and keep me in my (rapidly apporaching) decrepitude. | 16:43.35 |
sebras | Robin_Watts: I can tell from experience that watching TV news while eating might be somewhat...unsettling. | 16:43.51 |
Robin_Watts | Don't get me started pn rolling TV news... | 16:44.28 |
sebras | Robin_Watts: we didn't have that in .se until recently. | 16:44.49 |
Robin_Watts | "This is the view from our Bagdad hotel roof. We've heard stuff might be happening down the road, but we can't go and investigate because we've got to do another live link up in an hours time..." | 16:46.17 |
kens | (translation) It might be dangverous :-) | 16:46.55 |
sebras | Robin_Watts: unfortunately that wasn't the case when I was a kid. and it was Iran at the time I believe. hard to combine the TV images with my mother's food though. conflicting input... | 16:48.02 |
kens | marcosw looks like tht is working now | 17:00.29 |
| Hmm we have a *lot* of "new makefile issues" | 17:05.07 |
| OK I'm off for the night, bye all | 17:25.00 |
ascarpino | hi all, I get this error when converting a file to pdf using ps2pdf | 17:42.29 |
| Error: /undefined in ESCEESC*r0FESC&f101y0XESC&l1c12EESC*p0x0YESC*c5772x8238y0TESC | 17:42.31 |
| where should I begin to find the issue? | 17:42.49 |
Robin_Watts | Is the source file postscript ? | 17:43.35 |
henrys | marcosw_:just built gs with -j infinity with no problems | 17:43.41 |
ascarpino | Robin_Watts: no | 17:43.56 |
| Robin_Watts: it's a stream of raw data | 17:44.33 |
Robin_Watts | ascarpino: So you're using a tool that's designed to take postscript and turn it into pdf, and you're feeding it something other than postscript ? | 17:44.44 |
ascarpino | ehm right. | 17:45.02 |
marcosw_ | henrys: the problem occurs infrequently, so you may have to run it multiple times. If you remind me of the ip address I can try and duplicate it. | 17:45.15 |
ascarpino | Robin_Watts: what do you suggest to create a pdf file? | 17:45.28 |
Robin_Watts | ascarpino: From what? | 17:46.02 |
ascarpino | the stream could be a doc, or a pdf, or a web page... | 17:46.26 |
henrys | I don't know if there is a nice way for you to get to henrysx6 right now. I'm in the middle of a router change anyway. let me repeatedly compile | 17:47.14 |
| marcosw_:I do think this cron business should be revisited it blows away syslog⦠now doing 5 make -j; make clean in the ghostpdl directory, building all the languages hope that trips it. | 17:51.26 |
| marcosw_: I got this wndr4300 router a later release than chris' suggestion and saw there was code for it in openwrt but apparently dd-wrt hasn't taken up the changes yet so I'm stuck with net gear firmware and no dns, but it is fast particularly using the 5GHZ | 17:54.41 |
marcosw_ | henrys: it is possible that the problem only occurs during a cluster job. I can't think of how this could be, since the cluster just does shell call with 'make -j12', but I can setup a job on casper that causes a compile on henrysx6 to be launch repeatedly. | 18:08.44 |
henrys | I am just using -j and don't see anything. Please go ahead and set up the job and I'll monitor syslog | 18:09.59 |
| can you record the time it fails? | 18:10.13 |
marcosw_ | henrys: http://serverfault.com/questions/31334/how-can-i-prevent-cron-from-filling-up-my-syslog | 18:10.23 |
| apparently you aren't the first person to have this problem :-) | 18:10.42 |
henrys | marcosw_:looks like there are reasonable workarounds, but it doesn't seem the cluster should use cron at all. | 18:14.23 |
Robin_Watts | henrys: Why not? | 18:14.45 |
| Either we need to have a process that runs all the time as a top level watchdog, and we need to be really careful to catch every possible route by which it might fail and die, and hence leave a cluster node unconnected.... | 18:15.53 |
marcosw_ | henrys: there would still have be a cron job to check to make sure that the clusterpull script hasn't crashed, though that cold run every 5 or 10 minutes. | 18:15.54 |
henrys | because there is nothing about it that is scheduled that is what cron is used for. A script should be started in rc.local that is a simple infinite loop with a sleep. | 18:15.56 |
Robin_Watts | henrys: It *is* scheduled. | 18:16.37 |
| Every minute we check for timeouts etc. | 18:16.47 |
| and when your simple infinite loop with a sleep dies because of running out of memory, or process exhaustion, or any one of a myriad of other things, you'll lose the cluster node. | 18:17.55 |
| The cron based solution is low tech, but it's effective. | 18:18.14 |
marcosw_ | Robin_Watts: as I said earlier, we could still have a cronjob that check to make sure the infinite loop hasn't died, but that just changes the cron from once a minute to every N minutes | 18:25.16 |
henrys | Sorry I was called away | 18:34.45 |
| why isn't the server responsible for the time tracking/ | 18:35.43 |
| ? | 18:35.45 |
| anyway I think we can all agree syslog should not be flooded so all system messages are rotated out. | 18:36.41 |
| marcosw_:have you started the casper job? | 18:39.07 |
marcosw_ | henrys: yes, but I still need to add code to detect failure and quit. | 18:39.36 |
henrys | okay this is puzzling my makes seem to have worked okay. | 18:41.26 |
paulgardiner | faBvLon5 | 18:55.06 |
henrys | entropy is probably good enough but your usage of it needs work. | 18:57.31 |
paulgardiner | henrys: Yes: best not type it into a black screen without first checking there's a password prompt. :-) | 18:59.02 |
henrys | hi mvrhel_laptop feeling any better? | 19:55.06 |
marcosw_ | henrys: I've run about 10 test compiles on henrysx6 and so far no problem. You haven't changed anything since it last failed have you? | 20:29.05 |
henrys | marcosw_:I did enable smart diagnostics on the hard drive (http://en.wikipedia.org/wiki/S.M.A.R.T.) | 20:36.09 |
marcosw_ | henrys: I can't imaging that would make a difference; it really feels like a memory issue. | 20:39.20 |
henrys | well we tested memory and that didn't show anything then we said let's see about the drive so I checked that with smart. | 20:46.21 |
| marcosw_:I also changed the location of the box and use power line networking instead of 10 base T now, seeming unrelated. | 20:52.45 |
marcosw_ | Is it plugged into a difference circuit? I believe you have a color laser jet 4550, if it's anything like my 4500 and it causes all the lights on the same circuit to dim that could be it. I don't recall suggestion the drive; it really doesn't feel like a drive issue. | 20:54.54 |
henrys | marcosw:still here? | 21:03.40 |
marcosw1 | henrys: I'm back online. | 21:37.33 |
henrys | I do still have the printer on the same circuit but it is a different one and I'm not using it as frequently. I could print something while you do the test. | 21:38.24 |
marcosw1 | it's worth trying, though if the compile doesn't fail it won't prove anything :-) | 21:40.12 |
henrys | I would just enable the node and let's see what happens. | 21:42.41 |
marcosw1 | henrys: sounds good to me. I'll do that now. | 21:42.56 |
JakeSays | i'm tweaking the output of mudraw's xml text device. i need to capture paths as well as text. i've been able to add path tags to the xml easily enough, but the coordinates are not the same as the text. i'm assuming i need to apply the transformation matrix to the path, but i'm not sure how to go about it. any suggestions? | 22:42.39 |
| hmm. maybe i just call fz_transform_point on each point in the path | 22:49.57 |
| lol or maybe fz_transform_path | 22:52.57 |
emrosenf | Hi, does anyone here know if it's possible to compile ghostpdl as a static library for iOS? | 23:26.44 |
| The build system is quite complex and I've been having trouble configuring it. I'd really appreciate any insights. | 23:26.45 |
| Forward 1 day (to 2013/01/03)>>> | |