| | <<<Back 1 day (to 2012/08/26) | 2012/08/27 |
| kens | tor8 ping | 08:23.57 |
| tor8 | morning kens | 08:24.23 |
| kens | Hi tor8 | 08:24.29 |
| | Which branch spec should I use when 'pull'ing a gti repo for MuPDF ? | 08:24.50 |
| tor8 | kens: master is a safe bet | 08:25.16 |
| kens | OK so just git pull master ? | 08:25.32 |
| | AH no, that tells me 'master' is not a git repository, I guess I need the URL in there | 08:26.02 |
| tor8 | git pull <remote> <ref> | 08:26.10 |
| | so: git pull origin master | 08:26.17 |
| | or just "git pull" if you're sitting on a checked out (but old) master branch | 08:26.35 |
| kens | git pull didn't work. | 08:26.43 |
| tor8 | git remote -v | 08:26.54 |
| kens | It told me I hadn't specified a branch | 08:27.01 |
| | git pull origin master failed to auto-merge. I guess maybe its too old.... | 08:27.26 |
| tor8 | kens: yeah, there's something in the config about which checked out branch has a 'pull' equivalent | 08:27.29 |
| | kens: well, if all else fails this is what I do: | 08:27.46 |
| | git fetch origin && git reset --hard origin/master | 08:27.54 |
| | I assume you don't have any local work you need to keep here | 08:28.28 |
| kens | I'll try that, thanks, I don't care about the contents to fhte current reposotiry, I just want to look at Robin's linearisation code | 08:28.34 |
| | Well it seems that onvocaiotn worked, thanks | 08:29.18 |
| tor8 | kens: git status will tell you if you're on a detached HEAD or on some odd branch, which may be the cause of trouble pulling | 08:29.46 |
| | but if it worked, good enough :) | 08:30.00 |
| kens | I think its just that my copy was so old it couldn't cope | 08:30.03 |
| | git logg shows a tag for vcersion 1.1 which I'm sure is up to date enough for me | 08:30.45 |
| tor8 | well then it looks like you're set :) | 08:31.01 |
| kens | Indeed, thanks for that | 08:31.09 |
| | tor8 would you mind doing a ping test for me please ? | 08:35.25 |
| | Can you ping 212.48.119.62 ? | 08:35.59 |
| sebras | kens: yes. time: ~25ms... | 08:47.31 |
| kens | thanks sebras | 08:48.44 |
| | Looks like my ISP has a table routing problem or something | 08:48.56 |
| | I cna ping it from Casper, but not from here.... I've opened a ticket and attached the tracertoue log, bet that will confuse them :-) | 08:49.25 |
| sebras | kens: probably. :) | 08:50.05 |
| tor8 | Robin_Watts: raed just mailed me asking about mupdf + metro ... seems to be a popular question these days | 09:27.10 |
| | do you have a canned reply I can send? | 09:27.36 |
| Robin_Watts | tor8: I don't, sorry. | 09:27.56 |
| tor8 | Robin_Watts: what are the problems with compiling it as a metro library? I haven't looked into this enugh. | 09:37.58 |
| | enough* | 09:37.59 |
| Robin_Watts | I haven't looked into mupdf at all. | 09:38.18 |
| | For gs, all the problems with the compilation were to do with printer access. | 09:38.40 |
| | For mupdf that doesn't apply. | 09:38.55 |
| | So I'd hope the compilation/linking for mupdf should be simple. | 09:39.08 |
| | He might have problems that only certain paths can be opened, but that's a matter of restricting reading/writing to certain dirs. | 09:39.47 |
| | If he has a specific question, I can try to find time to try a TIFKAM build of mupdf later. | 09:40.18 |
| tor8 | Robin_Watts: nah. his question is of his usual character "do you know of any problems with XXX?" | 09:43.18 |
| Robin_Watts | Just fixed my current problem I think, so let me have a shower, then I'll try a build. | 09:44.37 |
| tor8 | I replied and CC:d support | 09:45.05 |
| Robin_Watts | Has anyone done a TIFKAM Gtk lib ? | 09:50.32 |
| sebras | tor8: hasn't Ch3rryCok3 been trying to port mupdf to metro? | 10:00.38 |
| Robin_Watts | sebras: possibly. | 10:16.12 |
| vtorri | is cross compilation to metro possible ? | 10:33.08 |
| | or is it only with MS tools ? | 10:33.17 |
| Robin_Watts | vtorri: The only method for building for the metro interface itself is VS2012, AIUI. | 10:38.54 |
| vtorri | ok | 10:39.20 |
| kens | I see raed doesn't understand the difference between 'Metro' and 'Windows 8' | 10:40.14 |
| Robin_Watts | I was tempted to reply with "I refer the honorable gentleman to the question given previously." | 10:40.58 |
| | s/question/answer/ | 10:41.24 |
| kens | Robin_Watts : how did you test your hint stream generation ? | 10:42.53 |
| Robin_Watts | kens: Poorly. | 10:43.11 |
| | I looked at it in a binary editor. | 10:43.20 |
| kens | ROFL | 10:43.20 |
| Robin_Watts | and I checked to see that Acrobat was happy with it being a web optimised PDF. | 10:43.35 |
| | and that was it | 10:43.45 |
| kens | OK well I guess I will do that too. | 10:43.47 |
| Robin_Watts | VS2012 is incapable of importing older VS projects. What a pain in the ass. | 10:44.18 |
| kens | The full version can't import older projects ? :-O | 10:44.41 |
| Robin_Watts | Express | 10:45.22 |
| kens | Ah yeah I think I read that somewhere, I believe the full version can (but I could be wrong) | 10:45.46 |
| Robin_Watts | Let me try importing it to VS2010, then into VS2012. | 10:46.01 |
| | Aha, that worked. | 10:47.02 |
| kens | I guess that's not totally unreasonable | 10:47.23 |
| Robin_Watts | Well, partially. | 10:47.25 |
| | libmupdf and libthirdparty imported. | 10:47.37 |
| | generated/mubusy/mudraw/mupdf didn't. | 10:47.54 |
| ||arifaX | where can I tell ghostscript to always use "-dPARANOIDSAFER" ist there a kind of gsrc file? | 11:26.12 |
| kens | No, there is no such file, you need to specify the option on teh command line, as far as I'm aware | 11:26.51 |
| ||arifaX | kens: I have a program that calls gs and I cannot specify addditonal parameters to it | 11:27.21 |
| Robin_Watts | You can do a custom build of gs that automatically sets that as a param. | 11:28.08 |
| | Or you can write a 'shim' app to do it. | 11:28.26 |
| ||arifaX | Robin_Watts: ok sim might be an option - thanks for that idea. will try that out. should also work with a link pointing to a script that invokes gs with the parameter | 11:29.23 |
| Robin_Watts | ||arifaX: Yes, if you're on *nix. | 11:29.49 |
| ||arifaX | I am | 11:29.54 |
| Robin_Watts | That's the easiest method then. | 11:30.02 |
| | tor8: Why do we need winsock2.h ? | 11:41.35 |
| tor8 | Robin_Watts: for "struct timeval" IIRC | 11:43.55 |
| Robin_Watts | I'm answering Raeds second mail. | 11:44.10 |
| tor8 | which is only needed for our gettimeofday implementation, which is only used for benchmarking in mudraw and the x11 app | 11:45.12 |
| kens | nyone know if there is any existing support in GS for writing a stream of bits to a file ? | 11:59.51 |
| Robin_Watts | kens: look in the lzw encoder? | 12:00.33 |
| kens | Yeah, that was what I was wondering. | 12:00.48 |
| Robin_Watts | That must handle bits of various token lens to a stream. | 12:00.52 |
| kens | I was hoping someone would know where to look adn save me the effort ;-) | 12:01.18 |
| | Looks like the LZW code does it all itself | 12:02.40 |
| ryanakca | I'm trying to compile mupdf, but get http://paste.debian.net/185841/ . Is it due to a missing build-dep? Any suggestions? | 13:27.31 |
| Robin_Watts | ryanakca: You're using an old version of openjpeg. | 13:28.17 |
| | We strongly recommend you use the thirdparty libraries we supply. | 13:28.33 |
| ryanakca | Robin_Watts: Will libjpeg version 6.2 do? | 13:29.57 |
| Robin_Watts | It's not libjpeg. | 13:30.07 |
| | It's openjpeg. | 13:30.12 |
| ryanakca | Oops, sory | 13:30.14 |
| Robin_Watts | Our version has bugfixes in that aren't in their released version yet. | 13:30.38 |
| | (pulled in from their SVN). | 13:30.53 |
| | You *can* build with any version of OpenJPEG 1.5, I believe, but we support stuff built with our version of the library. | 13:31.23 |
| ryanakca | Robin_Watts: I see. And I take it 0.9 doesn't have autoreload for pdfs? | 13:31.39 |
| | (The reason I'm trying to update mupdf is that I'm tired of having to press 'r' after each LaTeX run) | 13:32.20 |
| Robin_Watts | Probably not. But 0.9 is over a year old, and we've made MANY fixes and improvements since. | 13:32.21 |
| ryanakca | Robin_Watts: Alright. | 13:32.45 |
| Robin_Watts | Building with our libs is dead easy. You just unzip the zip file into "thirdparty" and 'make' as usual. | 13:32.50 |
| ryanakca | nods | 13:34.23 |
| Robin_Watts | kens: ufst question on support, and chrisl is away today... | 13:50.50 |
| kens | I know, and I don't know the answer | 13:51.01 |
| | I believe it is 'yes' but I would rather chrisl answered it | 13:51.18 |
| Robin_Watts | kens: Could send a "Chrisl is away today, but we *think* the answer is yes. Please wait til tomorrow for a definitive answer" reply? | 13:52.27 |
| kens | I could, but I'd really prefer to let chrisl answer it, I have never worked with UFST on Ghostscript, let alone PCL | 13:54.34 |
| Robin_Watts | fair enough. | 13:54.41 |
| sebras | ryanakca: for linux the mupdf view does however support reloading the pdf on SIGHUP. | 14:35.46 |
| Robin_Watts | sebras: In 0.9? | 14:48.02 |
| sebras | Robin_Watts: I believe so. | 14:54.27 |
| Robin_Watts | fair enough. | 14:54.43 |
| sebras | yes, my hunch (sp?) was right. it is indeed in 0.9 | 14:55.13 |
| | four commits before the 0.9 tag. :) | 14:55.55 |
| | ryanakca: I remember we implemented this for someone that wanted to be able to reload a pdf in mupdf that was generated from some latex source. was this you? maybe you can add "pkill -HUP mupdf" at the end of your command/Makefile to trigger the reload in mupdf? | 14:58.01 |
| ryanakca | sebras: No, it wasn't me, although I do remember stumbling across such a feature request when looking into mupdf. | 15:01.26 |
| | I'll add that pkill line :) | 15:01.38 |
| sebras | ryanakca: let me know if you have some problems with it. | 15:03.31 |
| | "any problems" even.... | 15:03.45 |
| henrys | looking for the monday morning frog ... http://www.goodreads.com/quotes/168105-eat-a-live-frog-first-thing-in-the-morning-and | 15:17.20 |
| sebras | so _that's_ why the french eat them... | 15:26.14 |
| henrys | the dentist appointment is the worst thing on my list I'll go for that. | 15:27.44 |
| ray_work | henrys: Robin_Watts: I hope that you don't mind my fixing up hpglpathmode to be 'bool'. It looks like it is not quite used yet, and it got rid of some VS warnings | 15:51.22 |
| Robin_Watts | ray_work: Not at all. | 15:51.48 |
| | Did you update the docs ? | 15:52.04 |
| ray_work | docs ? we don't need no stinking docs! | 15:52.26 |
| Robin_Watts | doc/Language.htm | 15:53.01 |
| ray_work | Robin_Watts: I'll do that in a moment. Sorry | 15:53.27 |
| Robin_Watts | no worries :) | 15:53.33 |
| ray_work | Robin_Watts: the docs weren't correct anyway (at least for .sethpglpathmode) | 15:54.44 |
| Robin_Watts | I just ran a test to collect statistics for the different types of color transformations we do on the cluster. | 15:55.13 |
| henrys | ray_work it is used by pcl what do you mean? | 15:55.31 |
| Robin_Watts | (obviously, the results are dependent on our test files and the devices we test) | 15:55.53 |
| | The most popular transform we use is... Gray to floating point Lab. | 15:56.18 |
| | I would not have predicted that. | 15:56.26 |
| rayjj | henrys: one place in pcl is is in an #if 0, but all of them used 'true' 'false', AFAICT | 15:56.28 |
| Robin_Watts | rayjj: Yes, it was being used as as bool. | 15:56.57 |
| | s/as as/as a/ | 15:57.02 |
| | Second most popular is RGB -> FP Lab and third is FP Lab -> FP Lab | 15:57.48 |
| henrys | ray_work:the call that sets the boolean is in pgrdaw.c | 15:57.50 |
| | your change is fine I'm just pointing out it is used. | 15:58.04 |
| Robin_Watts | Those 3 transform types account for more than all the rest put together. | 15:58.21 |
| henrys | but I'll remove the pctop.c call, I forgot to do that. | 15:58.31 |
| | I can see how that would be confusing | 15:58.49 |
| rayjj | Robin_Watts: OK, doc fixed. | 16:06.37 |
| | Robin_Watts: what was wrong previously was the notation for input and output parameters (a PS specific style of notation) | 16:07.39 |
| | Robin_Watts: and now it mentions 'bool' and 'true' or 'false' | 16:08.10 |
| | Robin_Watts: it doesn't mention, and I doubt anyone will care, but I assume that this is subject to the graphics state handling (gsave, grestore, grestoreall, initgraphics, ...) | 16:09.36 |
| Robin_Watts | rayjj: I don't think so. | 16:10.15 |
| henrys | pcl will maintain its state irrespective of gstate level | 16:10.58 |
| ray_work | henrys: ONLY if the value is set in the 'base' level graphics state (used by initgraphics), right ? | 16:12.18 |
| henrys | pcl explicitly sets it before drawing anything. | 16:12.39 |
| ray_work | but my comment was more to the specifics of the PS operator extensios, which (AFAICT) _will_ follow PS rules | 16:12.59 |
| | the doc (presumably by Robin_Watts) just says 'future path creation operations' | 16:14.02 |
| Robin_Watts | The ps stuff is really just for me to test it. | 16:14.40 |
| henrys | yes it seems reasonable it should follow along with the other ps operators but it doesn't seem to be a pressing issue maybe a p4 enhancement | 16:14.41 |
| Robin_Watts | and it may already work. | 16:14.54 |
| | I added it down the path of least resistance. | 16:15.06 |
| ray_work | henrys: totally agreed, which is why I didn't update the docs to be more specific. | 16:15.15 |
| | Robin_Watts: right -- I think it works as all other PS graphics state settings the way it is. The doc just doesn't mention that, but not worth changing | 16:16.04 |
| | I assumed that even adding the PS operators was just for convenience in testing (since PS and PDF are so much easier to edit than PCL) | 16:16.55 |
| Robin_Watts | exactly. | 16:17.04 |
| ray_work | except for henrys who can write PCL in his sleep ;-) | 16:17.16 |
| henrys | PDF is painful too. | 16:17.29 |
| Robin_Watts | I believe that may technically be a nightmare. | 16:17.38 |
| ray_work | actually, writing HPGL with an editor, using decimal coordinate values is pretty simple | 16:18.15 |
| | I used to code CalComp 907 graphics language by hand as well (since I developed the syntax, I knew the 'easy' way to write compliant 907) | 16:20.54 |
| Robin_Watts | evening chrisl | 17:06.52 |
| henrys | struggling to understand how these regression emails are generated - last one says diff between d840d and 11bb4 but those commits are chronologically separated by 4 commits | 17:08.24 |
| Robin_Watts | henrys: Some of the regression emails are from jobs that run overnight, right? | 17:16.03 |
| | which one are you looking at? | 17:16.38 |
| | ah, the performance one. | 17:17.00 |
| | That's run on a single machine once a day, right? | 17:17.24 |
| | hence if we do 8 commits in a day, the revisions tested will be 8 apart. | 17:17.51 |
| henrys | that is an odd way to set it up. Once a day go out and get the revs since last run and them in sequence. | 17:31.11 |
| | and do them in sequence | 17:31.43 |
| | there's no need to test against "master" just rev - 1, right? | 17:33.55 |
| Robin_Watts | henrys: ? | 17:35.01 |
| | Every night marcos does a performance test using the latest version on a single node and records the times. | 17:35.38 |
| | He compares those times to the times he recorded last night. | 17:35.54 |
| | and keeps them for tomorrows night to be compared against. | 17:36.10 |
| henrys | ah okay got it. | 17:36.35 |
| Robin_Watts | I suspect that if he tested every revision, then it would take 1 machine full time. | 17:37.10 |
| | (at least, the above is my understanding for how it's done) | 17:37.28 |
| henrys | yes I was thinking all revs were tested nvm | 17:38.34 |
| | Robin_Watts, tor8:we now have the signature from simon, great job guys. | 17:59.38 |
| tor8 | henrys: great news! | 17:59.48 |
| henrys | but it looks like he wants a regular payday or something and I'd like to slow things down with him. | 18:00.34 |
| Robin_Watts | henrys: you mean he keeps pushing for bounties? | 18:01.09 |
| henrys | he expect to keep patches at sumatra until they can be rolled into something with a bounty. | 18:03.54 |
| | wasn't that in the email I sent you? | 18:04.11 |
| | it is in the email from Simon Aug 24 | 18:05.21 |
| Robin_Watts | That's not how I read it. | 18:06.19 |
| | I read that email as "Miles says he only wants to work via bounties. So how do we do that?" | 18:07.02 |
| | I think we can send an answer back saying "We don't have to work through bounties - we just have to be clear that Artifex gets the required rights. For large pieces of work, we can certainly work for bounties to reward you for your efforts, but we are quite happy to accept smaller patches as any open source project would, as long as you are happy to assign us the required rights." | 18:09.34 |
| | or something like that. | 18:09.49 |
| | I didn't read it as "I'll expect to be paid for everything I give you in future". | 18:10.44 |
| henrys | I was going to post mortem this with miles, I believe the fundamental problem must be our agreement, if our agreement read like google's everything would be fine. So I'm going to find out what is wrong with the agreement then clarify what the agreement means with Simon. | 18:11.07 |
| Robin_Watts | Googles agreement read (to me) like "Here is my code. I promise I have the rights to give it to you." | 18:12.53 |
| | Ours seemed much more lawyery than that. | 18:13.54 |
| henrys | he is fairly clear "miles kindly insisted on preferring bounties to a CLA..." | 18:23.31 |
| Robin_Watts | Right. So either Miles has a reason for wanting bounties rather than a copyright assignment, or there was a misunderstanding there. | 18:25.16 |
| | I think it's worthy of a point at the forthcoming meeting (unless you want to get this postmortemed before then) | 18:26.30 |
| henrys | I'm working on it now if it isn't beaten to death by meeting time it will be on the agenda | 18:27.43 |
| alexcher | I have new results for the bug 693301, customer 393. The cause of the problem is now clear but our response to it isn't. | 19:10.47 |
| henrys | alexcher let's talk about it at the meeting tomorrow | 19:26.46 |
| ray_work | bug 693301 (test for chatzilla settings) | 21:00.41 |
| henrys | Robin_Watts:The glaring distinction between the google agreement and ours is google's covers future contributions so even if he signed it's worthless going forward. He'll have to bounty until I can get Miles to draft up a new agreement. | 22:12.02 |
| | I have been operating under the assumption it covered future submissions and I think everyone else has, ray_work? | 22:13.30 |
| | Forward 1 day (to 2012/08/28)>>> | |
IRC Logs