| <<<Back 1 day (to 2013/08/26) | 2013/08/27 |
sebras | Robin_Watts: btw, there are some #if 0 in source/fitz/output-pcl.c and source/pdf/pdf-write.c too. I'm guessing that the latter might be in development and the former is not. but yeah, they really ought to be on a branch I guess. | 00:04.21 |
| Robin_Watts: oh, and zenikos patch source/pdf/pdf-xref.c | 00:14.29 |
| argh! | 00:14.32 |
| zenikos patch c875b3a should probably not refer to fz_atof() as pdf_load_version() has removed this comment. | 00:15.48 |
| yoou guy really should register the domain ghostscript.ps :) | 06:48.02 |
joo0nas | kens: thanks for pushing - now I am here :-) | 07:25.14 |
kens | aha, welcome :-) | 07:25.25 |
| You can find someone here most of the time, Monday through Friday, not so much at weekends | 07:26.17 |
joo0nas | thanks | 07:30.35 |
joo0nas | guesses "time" means american timezones | 07:31.28 |
chrisl | joo0nas: there's folks in the US, a few of us in the UK. So, between west coast US, and the UK, there's *usually* someone around - of course, we're generally working, so our attention might not be on IRC, so it might take a while for a reply | 07:34.52 |
kens | Yes, sorry, I was debugging andmissed that. | 07:35.20 |
| But there is less likely to be someone here at a weekend | 07:36.02 |
| The channel is logged though, and we read teh logs, so you can always leave a message and get a reply later. | 07:36.35 |
chrisl | Logs are here: http://www.ghostscript.com/irclogs/ | 07:36.50 |
joo0nas | Iwas just curious: "opening hours" are not important to me; my pace is largely independent from that :-) | 07:36.53 |
chrisl | Working as we do, our working hours can be fairly random..... | 07:38.41 |
kens | chrisl ping | 08:11.03 |
joo0nas | can I safely drop .../Resource/Font/* if same fonts are available elsewhere in the font path? Or do some routine hardcode the path so I need to replace with symlinks? | 08:16.45 |
kens | We don't hard code to any fonts, so if you have replacements then you don't need the fonts (NB the default implementation of GS compiles the fonts and other resources into the executable) | 08:17.21 |
| In fact we don't have coded paths to anywhere (except the ROM file system), so you can safely all resources, as long as you include a way for GS to find the ones it needs. | 08:18.28 |
joo0nas | "compiles [...] into the executable" - is that what I disable with --disable-compile-inits ? | 08:20.47 |
kens | Yep, that's it | 08:20.55 |
| sets the flag COMPILE_INITS=0 | 08:21.06 |
joo0nas | good - for Debian we want fonts maintained independent - obviously preferrably without needing to recompile ghostscript if fonts are changed :-) | 08:22.04 |
kens | You don't need to recompiloe Ghostscript if the fonts change, you can just put them on the font path, they get located anyway | 08:22.45 |
chrisl | kens: pong | 08:22.57 |
kens | chrisl sending you an email, it got too complicated to explain. | 08:23.11 |
nomorebeer | hi guys can anyone help me avour mirc scripting? | 08:23.57 |
kens | chrisl mail sent | 08:24.00 |
nomorebeer | about* | 08:24.01 |
kens | nomorebeer : no probably not, we aren't involved with mirc | 08:24.22 |
nomorebeer | k thanks. | 08:24.34 |
chrisl | joo0nas: the one thing you need to be careful of is the CMap resources - various times people have attempted to share them between GS and poppler, and it's rather complicated, prone to disaster, and more effort than it's worth. Especially as we have CMap files (for internal use) that poppler doesn't. | 08:24.58 |
joo0nas | ok | 08:25.19 |
| hmm - such sharing is *exactly* what I do currently | 08:26.53 |
| I believe I did double-check that all files where identical, before synlinking them | 08:27.27 |
chrisl | joo0nas: but poppler won't have the Identity-UTF16-H CMap, as that is specific to Ghostscript - and without it, the CIDFont substitution can't work (as much as it ever "works") | 08:28.50 |
| kens: re your mail: sounds okay to me. | 08:30.08 |
kens | Thanks, just working on the changes now | 08:30.26 |
joo0nas | oh - that file got added for 9.06 - i.e. _after_ I compared files | 08:30.32 |
chrisl | joo0nas: actually it was accidentally removed from an earlier release, and put *back* in 9.06..... but yes. | 08:31.16 |
joo0nas | making a note to compare at build time | 08:31.22 |
| thanks a lot for that! | 08:31.44 |
| oh, and thanks a lot for Ghostscript at all :-D | 08:32.11 |
chrisl | joo0nas: the contents of the other CMaps files *should* match, as they are standard files released by Adobe - so not maintained by us, nor poppler | 08:32.19 |
joo0nas | right | 08:32.29 |
chrisl | joo0nas: you raised a question about the font license the other day...... and you mentioned that the current GPL might not be suitable for fonts? | 08:33.20 |
joo0nas | ...which annoys me about the package name in Debian - it should ideally reflect its true origin | 08:34.02 |
| yes | 08:34.12 |
chrisl | Is that just down to the exception for embedding, or are there other incompatibilities? | 08:34.54 |
joo0nas | I am no expert - but yes, I believe the issue is embedding | 08:37.18 |
chrisl | So, if I re-introduce the comment in the font files to include a reference to the GPL and the embedding exception, that should be okay. | 08:38.02 |
joo0nas | https://www.gnu.org/licenses/gpl-faq.html#FontException | 08:38.42 |
| I recommend to not touch the files, but put a README file in same folder stating the licensing | 08:39.42 |
| that helps identify duplicate fonts - as you are really not recompiling the fonts | 08:40.38 |
chrisl | I can't do that, as Ghostscript would then think there was a font called "README" in its font resource - that would be bad! | 08:40.46 |
joo0nas | then extend the LICENSE file in root dir | 08:41.10 |
chrisl | Yep, I'll do that shortly | 08:41.25 |
joo0nas | speaking of which: I am aware that you publish those fonts as a zip file, but do they have a canonical location closer to its source - e.g. at urwpp.de? | 08:42.21 |
chrisl | joo0nas: unfortunately not - somehow, we've ended up as the "upstream source" for the URW fonts :-( | 08:42.59 |
joo0nas | or rephrased: from where did _you_ get those updated fonts? | 08:43.05 |
kens | We got them from URW but URW aren't interested in open source fonts, so we get to take care of it | 08:43.27 |
joo0nas | oh - then I recommend to track them in a separate git | 08:43.54 |
kens | paulgardiner: , Robin_Watts, tor7 any of you have any ideas what's happening with Michael's winRT version of MuPDF ? THere's someone on Stack Overflow (a bit of a dim user, but still) wanting to do so: | 08:46.14 |
| http://stackoverflow.com/questions/18341802/any-one-using-mupdf-in-their-winrt-app | 08:46.14 |
| Note also answer 2 where someone else has done it. | 08:46.27 |
tor7 | kens: my brain hurts reading the question :( | 08:47.03 |
kens | Well, I did say a dim user...... | 08:47.13 |
chrisl | joo0nas: we get the fonts directly from URW+, by e-mail | 08:47.36 |
joo0nas | oh | 08:47.49 |
kens | He posted an earlier question (and didn't like the answers he got) which makes it clear that he's using the ordinary MuPDF which tries to use the forbidden C APIs | 08:47.52 |
chrisl | joo0nas: do you know Fabian Greffrath? | 08:48.36 |
joo0nas | sure do :-D | 08:48.42 |
chrisl | He can fill you in on some the background with the URW fonts | 08:48.59 |
tor7 | kens: the winRT version of mupdf is in the git and should work, but I haven't had the time to try building it myself yet. | 08:49.32 |
kens | tor7 OK thanks, I will post that info on the SO question | 08:49.47 |
| thanks BTW | 08:50.02 |
tor7 | kens: there are project files in the platform/winrt directory | 08:50.14 |
kens | excellent, thanks tor7 | 08:50.24 |
joo0nas | chrisl: seems he just did, few hours ago: http://bugs.debian.org/496070#104 | 08:50.27 |
kens | THe cluster really goes fast since marcosw added all those new nodes. I'm surprised he hasn't added a smoots though | 08:53.34 |
kens | coffees | 08:54.07 |
chrisl | joo0nas: just reading that mail from Fabian, just to clarify: the real reason we dropped the fonts forked with the cyrillic glyphs was that the cyrillic glyphs simply too low quality for high res print work. | 09:08.19 |
joo0nas | would be great if such decisions was reflected in git commit messages - seems it only now is "Revert to using the "pristine" URW fonts." | 09:11.14 |
| ...or perhaps even in changelog - some people do care about fonts, you know | 09:11.54 |
kens | You're talking to people who do :-) | 09:12.17 |
joo0nas | I know | 09:12.31 |
| my point is - expect others to do too | 09:12.41 |
| sorry if that landed as harsh - I only meant that those (other that yourselves) would want to follow evolution of font changes - reason why things change | 09:15.03 |
kens | Well we normally try to be specific in commit messages, I can't recall why we weren;t there. Maybe we didn't want to seem critical of the changed fonts. | 09:15.40 |
chrisl | joo0nas: the problem is, there were a whole raft of factors that went into that decision (you can see some in the bug referenced in the commit), and to put them all in the commit message, it would end up rather longer than normal. That's one reason we reference bug numbers in the commit messages. | 09:17.15 |
joo0nas | ohh - sorry: there _was_ a bug referenced to that commit. Makes perfect sense now! | 09:20.11 |
Robin_Watts | gets annoyed with RLE stuff and goes for a run. | 10:33.02 |
| chrisl: So... if I print a file to PDF and it encodes the URW font into that document, the license says that's fine. | 12:32.23 |
chrisl | Robin_Watts: basically, yes. | 12:32.44 |
Robin_Watts | If I then recode that into (say) an XPS file, that's against the license. | 12:32.44 |
chrisl | Apparently so, yes | 12:33.07 |
Robin_Watts | or PXL (PXL supports high level fonts, right?) | 12:33.31 |
| I wonder if that's a deliberate thing, or an accidental ommission. | 12:33.47 |
chrisl | I used the wording from the original font files from way back - the wording is not mine | 12:34.22 |
paulgardiner | Robin_Watts, tor7: Four commits on paul/android-signatures need an eye cast over them. | 12:57.04 |
joo0nas | chrisl: if you are talking about wording of font license exception from (A)GPL, then only the copyright holder can choose a wording | 13:09.30 |
chrisl | joo0nas: yes, exactly. I'm slightly assuming that the wording was agreed with URW, given that it's what we were shipping up to 9.05 or 9.06.... | 13:10.45 |
joo0nas | that's a reason for URW++ to release on their own: you "assuming" a license for a new work passed on from them is... not ideal | 13:11.40 |
atulagrwl | paulgardiner: Did you get a chance to look at the issue? | 13:12.23 |
joo0nas | ideal would be for the copyright holder to *include* a licensing statement with each work they release | 13:12.47 |
chrisl | It would, but they don't..... | 13:13.01 |
joo0nas | ...either embedded or as a separate statement - but not stuffed on by others after the fact | 13:13.28 |
paulgardiner | atulagrwl: Just having a look now. | 13:13.41 |
atulagrwl | paulgardiner: Thanks a lot. I keep bothering you with the same problem again and again :) | 13:14.01 |
paulgardiner | atulagrwl: No worries. It's another problem that we are glad to have had pointed out. | 13:15.01 |
| For some reason, the page objectst parent_num isn't being set, and that prevents it being incrementally updated correctly. | 13:16.13 |
atulagrwl | paulgardiner: Good to know you are still talking to me :) | 13:17.27 |
| Is annotation the first case mupdf adding some content to pdf? | 13:21.59 |
tor7 | paulgardiner: the first 3 look simple enough, the last one has too much java for me to reliabily review | 13:23.15 |
paulgardiner | Thanks tor7. That's most of it at least. | 13:24.13 |
Robin_Watts | paulgardiner: Looking at your last commit now. | 13:26.31 |
| The use of 0, 1 and 2 might be more nicely done as enums. | 13:26.44 |
paulgardiner | Yeah, especially as I convert to a Java enum in PassClickResultSignature. | 13:28.40 |
| I'll fix that. | 13:28.46 |
Robin_Watts | all the rest of it looks plausible. | 13:36.25 |
paulgardiner | Brilliant. Thanks | 13:38.30 |
Robin_Watts | Ok, so what's the best way to detect endianness at compile time? | 13:50.36 |
| static inline fz_is_big_endian(void) { union {int i;char c[sizeof(i)];}u; u.i=1; return u.c[sizeof(int)-1]=1; } ? | 13:54.27 |
| tor7: How would you feel about me adding that to mupdf ? | 13:54.40 |
| Turns out that my SWAR plotting enhancements cause problems on big endian machines. | 13:55.02 |
| or even: static inline fz_is_big_endian(void) { union {int i;char c[sizeof(i)];}u; u.i=1; return u.c[sizeof(int)-1]; } | 13:55.58 |
| chrisl: Sanity check please. | 14:05.10 |
chrisl | You seem sane to me..... | 14:05.24 |
| ;-) | 14:05.29 |
Robin_Watts | I'm running MuPDF on the Raspberry pi, and it suddenly says "Killed" and stops. | 14:05.30 |
| ulimit says "unlimited" | 14:05.43 |
| Is this going to be it running out of memory? or something else? | 14:06.01 |
chrisl | Oh, it could be running out of memory, yes. The kernel reserves the right to kill processes that it deems are hogging too much memory, if it's running low | 14:06.38 |
tor7 | Robin_Watts: linux will terminate a process if it overcommits memory and the kernel later realizes it can't provide all the memory it promised | 14:07.37 |
| malloc on linux (in the absence of ulimit) will never fail. it'll terminate the process instead. | 14:08.06 |
Robin_Watts | tor7: I found an "isbigendian" static inline in one of the crypt functions, and I'm reusing that. | 14:08.46 |
| malc: You about? | 14:08.52 |
chrisl | tor7: Hmm, is that a recent change? | 14:09.02 |
tor7 | chrisl: it's been that way for as long as I can remember (which arguably isn't more than a few years) | 14:09.32 |
| Robin_Watts: there's an is_big_endian test in platform/x11/x11_image.c | 14:10.18 |
| Robin_Watts: shouldn't that union-based variant be "return u.c[0] == 0"? | 14:11.49 |
| well, I guess both ways should work fine | 14:12.23 |
Robin_Watts | tor7: either u.c[0] ==0 or u.c[sizeof(int)-1]==1 should be the same. | 14:12.29 |
| but I've used the one from the existing code. | 14:12.41 |
chrisl | kens: did you see my mail CC-ing tech about the new release branch? | 14:18.00 |
kens | chrisl no, sorry, will look | 14:18.24 |
chrisl | It was yesterday | 14:18.33 |
kens | Then it didn't reach me | 14:18.42 |
| I'll check gmail spam | 14:18.48 |
chrisl | kens: subject was "9.10 branch test....." | 14:19.29 |
kens | let me check my Eudora mailboxes too, just a moment | 14:19.45 |
| chrisl ah yes I did get it, but it seemed ot be directed to Marcos, so I ignored it | 14:21.57 |
chrisl | kens: yeh, I just haven't had a reply - I'll ping marcos when I see him on here | 14:22.28 |
kens | I see, that makes sense yes | 14:22.40 |
stanv | hello. I want to export UTF8 text to PS, and paste it into my existent ps document. | 14:32.00 |
| u2ps generates new PS page | 14:32.00 |
| How to generate PS from UTF8 ? | 14:32.00 |
kens | stanv use an application like a word processor | 14:34.59 |
stanv | kens: I can'tâ¦.. I want to modify CUPS's banner..... | 14:35.40 |
| Justadd few words in UTF8 | 14:36.02 |
kens | Its not that easy | 14:36.12 |
| PostScript expects every character code in a string of text to have a 1:1 mapping throuhg the Encoding vector to a named glyph. Obviously that falls down when the input is of variable width | 14:37.09 |
| You could do it using a CIDFont of course, but I'm guessing you don't want to fo through embedding a CIDFont either | 14:37.54 |
stanv | I want ) | 14:38.43 |
| embedding isn't a problem | 14:38.54 |
| I need to do this task once. | 14:39.08 |
kens | OK so embed a CIDFont and create a custom CMap which maps UTF8 | 14:39.13 |
| If you want to do it once, then don't bother, use glyphshow to show the named glyphs instead. Or set up a font with a custom encoding and map the UTF8 codes to that encoding. | 14:40.01 |
| THis is only a problem if you want to use ITF8, the simple answer is 'don't use UTF8' | 14:40.32 |
| For more on re-encoding fonts see Acumen Training's Acumen Journal, November and December 2001: | 14:42.11 |
| http://www.acumentraining.com/acumenjournal.html | 14:42.11 |
stanv | I want to have a compatible PS template that will be printed without any problems on any printer. | 14:42.31 |
kens | Why do you want to use UTF8 ? What glyphs are you intending to use ? | 14:43.13 |
Robin_Watts | And I'd like earths primary satellite mounted on a wooden rod. | 14:43.19 |
stanv | )))))))) | 14:43.31 |
| kens: UTF8 - is primary locale in Linux. | 14:44.05 |
kens | stanv, so what ? You said you only wanted to do this once, which glyphs are you planning to use ? | 14:44.25 |
stanv | and I want to printout words in Cyrillic | 14:44.40 |
kens | Then you will need to embed a font, the basic fonts don't have full support for Cyrillic | 14:44.58 |
stanv | kens: is there any "howto" or manual ? | 14:45.28 |
kens | SO don't use UTF8, encode the font you embed so that each gly[ph can be represented using a 1 byte code, and use normal PostScript | 14:45.29 |
Robin_Watts | stanv: You want to be able to change the words for every job? | 14:45.33 |
| Or you want a fixed page? | 14:45.42 |
kens | stanv the PostScript Language Reference Manual | 14:45.47 |
| 3rd edition, its available form Adobe | 14:45.59 |
| http://www.adobe.com/products/postscript/pdfs/PLRM.pdf | 14:46.22 |
stanv | Robin_Watts: noâ¦. it will be all the same wordsâ¦. | 14:46.22 |
kens | stanv then you don't need to use UTF8 | 14:46.36 |
Robin_Watts | stanv: Then put the text in whatever font you want into a drawing package. Export it as paths. | 14:46.52 |
kens | Yes, was about to suggets that also | 14:47.06 |
Robin_Watts | Then send that at the start of every job. No font messing. Much easier. | 14:47.10 |
kens | Easiest is to create an EPS from the drawing package, and include the EPS in the final file, the EPS is guaranteed to be in an acceptable format that way | 14:49.22 |
stanv | include ? | 14:50.58 |
kens | cut and paste, insert programmatically, whatever | 14:51.13 |
| put the contents of the EPS file inside the PostScript to be pritned | 14:51.32 |
stanv | hmmm | 14:51.47 |
| kens: it seems the easiest way for me⦠should try it | 14:52.40 |
kens | Its probably the least pain. | 14:53.05 |
sysdef | when i use -sOutputFile=page-%d.png while converting a pdf to png i get a file named "page.png" instead of "page-1.png" :'( | 14:53.56 |
kens | sysdef what version of GS are you using ? WHat platform ? | 14:54.25 |
| I think maybe it doesn't like the '-' try omitting it | 14:54.42 |
sysdef | kens: version 9.05, debian gnu/linux 7.1 | 14:55.18 |
stanv | sysdef: just enclose in single ' | 14:55.38 |
| -sOutputFile='page-%d.png' | 14:55.48 |
| kens: i see a%%PageBoundingBox: 0 0 1280 800 | 14:56.48 |
| userdict begin | 14:56.48 |
| and huge binary data | 14:56.56 |
sysdef | wow, that works \o/ | 14:57.00 |
| thank you | 14:57.04 |
kens | stanv, I'd have to see the file, possibly you are looking at the EPS preview | 14:57.24 |
stanv | i have a EPS fileâ¦. | 14:57.43 |
kens | EPS files may contain a 'preview' which is a bitmap image representing the final output | 14:58.02 |
stanv | but I don't know how to do copy-paste to my existence file | 14:58.18 |
kens | text editor ? | 14:58.25 |
sysdef | with -sOutputFile='100-page-%d.png' i get "100-page-1.png" now. perfect =) | 14:58.39 |
stanv | don't know what exactly to copy-paste â¦.. PageBoundingBox - just this block ? | 14:59.10 |
| until word `end' ? | 14:59.54 |
kens | You need the whole thing if its an EPS (if there is a preview you can omit that). You need to scale it and place it as required on the page, and insert it into the final document in such a way that it does whatever you expect. Remember PostScript is an opaque imaging model, so later objects cover objects lower in the Z order. | 15:00.32 |
stanv | But there is a header: https://pastee.org/yv9up | 15:01.59 |
| is this normal ? | 15:02.07 |
kens | In general PostScript files always have a header, yes, its not the same as a preview | 15:02.28 |
Robin_Watts | stanv: Anything starting with % is a comment. | 15:02.30 |
| You can include it, or exclude it as you wish. | 15:02.41 |
stanv | hmm, thanks! | 15:02.52 |
kens | I have to say (wihtout seeing the whole file) that that looks like its converted it into a raster | 15:02.54 |
Robin_Watts | kens: yeah. | 15:03.02 |
| stanv: Where did you make the EPS from? | 15:03.11 |
stanv | convert ) | 15:03.17 |
| ImageMagic's convert | 15:03.24 |
kens | Not a good plan | 15:03.30 |
stanv | I know.... | 15:03.35 |
kens | ImageMagick only handles raster formats | 15:03.37 |
stanv | ))) But just for test | 15:03.42 |
| ))) | 15:03.43 |
kens | You need to use something which can export EPS | 15:03.45 |
Robin_Watts | What did you create the original thing in? | 15:03.46 |
stanv | Robin_Watts: I don't know for now. I 'm just looking an easiest way to modify CUPS's banners... | 15:04.29 |
| Will try inkscape | 15:05.09 |
| Thank you for yours help! | 15:05.49 |
| See you | 15:05.53 |
chrisl | marcosw: did you get my mail yesterday about the new release branch? | 16:05.24 |
marcosw | chrisl: yes. the regression testing started today, so will be done next Tuesday. | 16:06.27 |
kens | crosses fingers | 16:06.54 |
chrisl | marcosw: super, thanks - good morning, btw :-) | 16:06.55 |
marcosw | good afternoon to you. | 16:08.28 |
| Are we not meeting because henrys is out of town or is it just a very quiet meeting? | 16:09.12 |
kens | henry said no meeting | 16:09.27 |
marcosw | right, to work then. | 16:15.02 |
ray_laptop | henrys did say no meeting. | 16:17.34 |
| chrisl: Is there something we need to know or do to try and squeeze out the release ? | 16:18.03 |
chrisl | ray_laptop: how do you mean? | 16:18.29 |
ray_laptop | or are we just waiting on Marcos ? | 16:18.40 |
kens | Testing would be a good thing | 16:18.40 |
| Try every combination of safer, set safe,locksafe, setpagedevice etc you can think of | 16:19.07 |
chrisl | I'll do a release candidate on Thursday, it would be good if as many folks as possible could grab that. | 16:19.13 |
ray_laptop | and, is there any way to speed up Marcos' run through to get it out in August ? | 16:19.27 |
| kens: wilco | 16:19.40 |
chrisl | Why does it have to be in August? | 16:20.02 |
| Given that I've started a new release branch from (yesterday's) master, I don't think there's a way to short-cut Marcos's testing | 16:21.57 |
ray_laptop | chrisl: doesn't really. We've missed Aug//Feb once before, iirc | 16:22.00 |
chrisl | ray_laptop: we aim for August to fit into Ubuntu's schedule, but that's blown already | 16:22.27 |
ray_laptop | chrisl: I didn't mean to shortcut as much as to run on more computers at once | 16:22.33 |
| chrisl: true. I haven't heard what tkamppeter is doing. Is he just staying with 907 ? or merging patches in as we find things | 16:23.19 |
chrisl | ray_laptop: he'll take 9.09 and apply the appropriate patch I'd think | 16:23.46 |
ray_laptop | assumes that tkamppeter will see his nick being mentioned and reply | 16:24.47 |
chrisl | marcosw: so, might there be a to speed things using more machines - even if we sacrificed some normal cluster time to it? | 16:25.01 |
Robin_Watts | chrisl: Presumably we could fork off as many more amazon instances as we wanted. | 16:26.55 |
chrisl | Robin_Watts: yeh, I meant sacrifice "local" cluster time | 16:27.39 |
marcosw | chrisl: I could cut the testing to a couple of days if I used the cluster nodes in my garage. They are currently running fuzzing tests, but I could pause those until the release is out. | 16:28.11 |
chrisl | marcosw: unless it's going to be a significant amount extra work, it might be worth doing | 16:30.14 |
marcosw | it's not too much work, so I'll go ahead and do this. Presuming the power company doesn't decide that the increase in electricity usage means I've started growing pot the testing should be completed by Thursday. | 16:31.17 |
chrisl | :-) | 16:31.56 |
marcosw | i'm amazed the poweredge servers work in my garage, we've had 90 degree days the last week and they haven't fallen over (they are really, really loud, so presumably the fans are running at full speed). It will be nice to have have a heated garage this winter :-) | 16:33.26 |
kens | stores beer in the garage during the winter | 16:33.50 |
chrisl | keeps a car in the garage - must be strange....... | 16:34.35 |
Robin_Watts | converted his garage into a lounge/music room/cinema. | 16:34.54 |
kens | Well, we also keep a car in the garage, but there's plenty room for the beer too | 16:35.23 |
Robin_Watts | kens: You clearly don't have enough beer :) | 16:35.39 |
kens | Its a big garage :-) | 16:35.52 |
marcosw | Robin_Watts: he used to have more beer. can't let beer sit around for long, it goes off. | 16:36.39 |
kens | chrisl I just made a commit which is potentially destabilising, (88bce11...) please don't include that in any release soon, no matter what happens ot the RC..... | 16:36.52 |
Robin_Watts | marcosw: If only I drank beer. I have a handy brewery within staggering distance. | 16:37.04 |
kens | has 2 | 16:37.16 |
| just over the railway bridge | 16:37.28 |
ray_laptop | marcosw: What's the best place to buy a poweredge for my office? or should I just google/ebay ? | 16:37.54 |
marcosw | http://theserverstore.com/ | 16:38.14 |
Robin_Watts | can imagine kens speccing up his house... Must have decent broadband, 3 bedrooms, office, and at least 2 breweries within 5 minutes walk. | 16:38.34 |
chrisl | kens: okay, I'd like to think I won't be taking another new release branch any time soon..... <sigh> | 16:38.34 |
Robin_Watts | ray_laptop: How big is your office? | 16:38.54 |
kens | Robin_Watts : ands a mainline railway station too | 16:38.57 |
marcosw | The latest machine I bought is a c6100 with eight L5639 cpus | 16:39.12 |
Robin_Watts | ray_laptop: From what marcosw says, these poweredges are LOUD. | 16:39.20 |
kens | Well, down to 88 bugs, I'm going to call it a night and enjoy the warm fuzzy feeling until tomorrow. | 16:40.57 |
marcosw | the C6100 isn't as loud as the C1100 (it's a 2U machine, so the fans are bigger). there are a couple of forums with long threads devoted to trying to make them quieter, though I gather at full load those attempts all fail (it's hard to cram 8 cpus into an volume that small and not require massive airflow). | 16:41.31 |
chrisl | kens: ah yes, I remember fixing bugs and other useful things like that...... those were the days...... | 16:41.56 |
Robin_Watts | marcosw: How much RAM ? | 16:42.09 |
kens | I'll be happy just to not introduce any new ones..... | 16:42.20 |
marcosw | ray_laptop: btw, theserverstore.com also sells on ebay, under the name pdneiman <http://stores.ebay.com/THESERVERSTORE/DELL-SERVERS-/_i.html?_fsub=2873213013&_sid=4059453&_trksid=p4634.c0.m322> | 16:42.22 |
| Robin_Watts: there are four blades, each with 2 cpus and 24 gigs of ram, so 96 gigs total. total cost was ~$1400. | 16:42.57 |
Robin_Watts | with no hard drives? | 16:43.25 |
marcosw | with 4 250G drives. | 16:43.38 |
| ray_laptop: some configurations are cheaper on ebay and some cheaper on theserverstore.com site. | 16:44.18 |
ray_laptop | marcosw: are you running RAID ? | 16:46.49 |
atulagrwl | paulgardiner: Do you want me to log a bug for the problem? I could not log it yesterday. | 16:47.17 |
Robin_Watts | ray_laptop: That's one HD per node, so I'd guess not. | 16:48.21 |
marcosw | ray_laptop: nope, each blade has to have it's own drive (presumably you could net boot and not have a drive). | 16:48.28 |
| this is what I bought: http://www.ebay.com/itm/DELL-POWEREDGE-C6100-XS23-TY3-SERVER-8x-L5639-SIX-CORE-CPUS-96GB-MEM-4x-250GB-/200918620138?pt=COMP_EN_Servers&hash=item2ec7aed7ea | 16:49.49 |
paulgardiner | atulagrwl: Don't worry. I think I've fixed it. If we decide we need a bug as a record, I'll log it. | 16:50.50 |
marcosw | I'm still amazed that a computer than has 96 equiv. cpus in it, each running at 2.1 GHz can be bought for $1400 (and that it can be carried by one person, though at 80 lbs I don't recommend using it as a laptop). | 16:52.40 |
atulagrwl | paulgardiner: Good to know you fixed it :). | 16:53.20 |
Robin_Watts | marcosw: You just described a 2020 cellphone. | 16:55.24 |
| except it'd cost $600 :) | 16:55.43 |
marcosw | and still weigh 80 lbs | 16:56.06 |
Robin_Watts | and weigh less than the powercord for the server. | 16:56.15 |
marcosw | no, the batteries would be heavy... | 16:56.32 |
| unless we get those Mr. Fusion reactors the back to the future documentary promised us about now. | 16:57.00 |
ray_laptop | marcosw: Do you think a 2xQuad core can run the regressions OK with 16Gb ? | 16:58.18 |
marcosw | ray_laptop: no problem, we have cluster nodes that only have 4 gigs (single quad core). | 16:59.48 |
ray_laptop | marcosw: yeah, I think peeves only has 6Gb | 17:00.16 |
marcosw | I noticed that when I was updating the cluster node spreadsheet and thought it a odd it wasn't a power of 2 | 17:00.56 |
ray_laptop | marcosw: the memory control is split 3 ways interleave on the FSB | 17:02.46 |
| don't ask me why -- just how much silicon they wanted to use for it I guess ? | 17:03.11 |
| and how many pins | 17:03.36 |
| kens: In case you are on later... I have a question about the message vs. the code for Bug 690667 | 17:12.17 |
| chrisl: did you review this patch ? | 17:12.31 |
| iirc, we have a policy that patches get reviewed by another engineer facile in the area | 17:13.16 |
chrisl | ray_laptop: I didn't review it, no. But it looks okay to me - at a glance, anyway | 17:21.35 |
ray_laptop | chrisl_away: I'm testing it, but it doesn't seem right given the comments. I'll put the comments in the bug | 17:30.51 |
atulagrwl | paulgardiner: Did you submit the fix to repository? | 17:41.26 |
tkamppeter | ray_laptop, chrisl_away, I have packaged 9.09 for Ubuntu, is something wrong with it? Do I have to package a 9.10? What exactly happened. | 17:59.15 |
Robin_Watts | atulagrwl: Not yet. | 18:03.44 |
| paulgardiner is probably gone for the day. | 18:04.02 |
ray_laptop | reopened bug 690667 | 18:09.44 |
| Sorry, kens. :-( | 18:09.53 |
Robin_Watts | ray_laptop: From kens comments earlier, I think he suspected it might not be entirely perfect. | 18:10.24 |
ray_laptop | tkamppeter: you can see the git log for 2001f469 on Aug 24 | 18:10.48 |
| Robin_Watts: yeah, probably. That's why we are supposed to have review, and setpagedevice is particularly "hairy" | 18:11.23 |
Robin_Watts | yeah. You'll recall that I'm a fan of the whole review thing. I think it catches all sorts of errors for minimal cost. | 18:12.08 |
ray_laptop | tkamppeter: basically, -dUseCIEColor was broken | 18:12.23 |
atulagrwl | Robin_Watts: Thanks for info. | 18:12.34 |
Robin_Watts | tkamppeter: We hope to have an RC on thursday. It's undergoing testing now. | 18:13.54 |
ray_laptop | It's probably particularly frustrating for chris and kens if they have to wait on a review from Pacific time, but not much more so than me waiting for review if it's after they both quit for the day | 18:14.09 |
Robin_Watts | tkamppeter: I think chrisl and kens thought we'd already missed the ubuntu deadline. | 18:14.17 |
| ray_laptop: Well, given we all use git, it's not *too* much effort to park a commit and wait for it to be reviewed. | 18:15.05 |
| You can always keep working, and go back and do fixes etc. git is good like that. | 18:15.20 |
ray_laptop | Robin_Watts: right, I've learned how to have WIP parked on branches. | 18:22.05 |
| and if I can learn how to do it ... ;-) | 18:22.22 |
tkamppeter | ray_laptop, chrisl_away, kens, Robin_Watts, the Ubuntu deadline for a new feature release is Thursday (Feature Freeze), patches and pure bug fix releases I can accept until one week before release (around 6 weeks from now), but the quicker I get the fixes of known bugs before release, the more it gets tested before release. | 18:34.35 |
Robin_Watts | tkamppeter: Well, I think chrisl is off tomorrow, and he'll make the rc on thursday. | 18:35.06 |
| Hopefully marcosw's testing might have completed by then. | 18:35.17 |
| so it's possible that you'll have something to work with on thursday. | 18:35.43 |
tkamppeter | Robin_Watts, as 9.10 is a pure bug fix release (Am I correct?) I will also take it after feature freeze, and I also will take the RC to get more time for testing. | 18:36.31 |
Robin_Watts | tkamppeter: It's broadly a bug fix release. | 18:36.50 |
| One or two other things may have crept in, but it's pretty much just fixes for the nasty bugs we found. | 18:37.23 |
tkamppeter | Robin_Watts, do you know what changes are in 9.10? And what are the nasty bugs? | 18:38.00 |
Robin_Watts | http://git.ghostscript.com/?p=ghostpdl.git;a=shortlog;h=refs/heads/gs910 | 18:38.56 |
| Shortly after the 9.08 release happened Sags appeared complaining of problems with the NTFS filename enumeration code. | 18:39.41 |
| kens fixed that, but IIRC we weren't going to do a new release for that. | 18:40.00 |
| but then we had complaints that ghostview didn't work. | 18:40.10 |
| and it turned out to be that ghostview was nastily relying on some internal parts of gs that we'd changed a bit for adobe compatability. | 18:40.34 |
ray_laptop | tkamppeter: the 'stoppers' were gv and GSView related | 18:40.50 |
Robin_Watts | kens did some magic to make it work and still be compatible with adobe. | 18:41.03 |
| Then we released 9.09 with that in. | 18:41.10 |
ray_laptop | Robin_Watts: Note, that these aren't really "internal", but are documented in Language.htm | 18:41.18 |
Robin_Watts | Then someone found that gsview had the same problems. | 18:41.27 |
| so kens had to do the same magic somewhere else, AIUI. | 18:41.36 |
| and we had a customer report of a buffer overflow that ray has fixed. | 18:42.02 |
tkamppeter | Robin_Watts, OK. Thank you very much. I will take 9.10, independent of our Feature Freeze. | 18:44.47 |
ray_laptop | Robin_Watts: it was pdf2dsc that used -dDELAYSAFER and then pdf2dsc.ps that used .setsafe that was the gv related issue, but we heard that pdf2dsc was used fairly widely | 18:45.15 |
| my fix was totally optional, but a clear buffer overrun | 18:45.15 |
Robin_Watts | likewise there are fixes from me in there to do with interpolated landscape images. | 18:46.24 |
| which were optional for the release, but chrisl has gone back to master and started a whole new release process, so they got pulled in too. | 18:46.52 |
tkamppeter | Yes, gv dopes not work at all with GS 9.09. | 18:47.18 |
| s/dopes/does/ | 18:48.05 |
Robin_Watts | For all you lucky people that get SciFi: http://vark.me.uk/lazarus.88 | 18:48.26 |
| How high were the money men at SciFi when they said "yeah, we'll fund that!" ? | 18:49.17 |
marcosw | Robin_Watts: me.uk? Is that new? | 18:49.50 |
Robin_Watts | marcosw: I think it's been around a while. | 18:51.41 |
ray_laptop | kens: chrisl: I re-closed ken's bug 690667 after testing on Adobe CPSI. It's not at all obvious, but Adobe deactivates the device on EVERY setpagedevice call, even one with an empty dict: << >> That defeats any N-up using BeginPage EndPage procs and I don't see it docuemted anywhere :-( | 19:16.58 |
| have to run an errand. bbiaw | 19:21.40 |
| one last regression run before pushing the saved-pages changes. Then maybe somebody will look at it. At least Marcos can add --saved-pages-test to one of the weekly regressions | 21:38.12 |
| marcosw: (for the logs) once I commit it, I'd like to get an 'all devices' run to make sure that it doesn't choke some strange device. | 21:39.19 |
tkamppeter | Has anyone tried to build GS against the system's libopenjpeg? Is this possible? | 23:17.45 |
| Has anyone tried to build GS against the system's libopenjpeg? Is this possible? | 23:47.40 |
| Forward 1 day (to 2013/08/28)>>> | |