| <<<Back 1 day (to 2012/11/20) | 2012/11/21 |
Robin_Watts | mvrhel_laptop: oh, awesome. | 00:08.29 |
mvrhel_laptop | Robin_Watts: So I went to the store here after calling to make sure they had the 7 32G | 00:15.20 |
| and they said they did | 00:15.24 |
| went I go there, no they didnt | 00:15.31 |
| then I checked my phone back at google and there was the nexus 10 32G now available | 00:15.54 |
| when I went there ..... not went I go there | 00:16.25 |
Robin_Watts | mvrhel_laptop: Gotta love stores like that. | 00:36.57 |
ray_work | just for reference ... on an item I know is in short supply, when I call I usually get their name, have them put it aside (and tell them I will be there in ___ hours) to make sure I dpn't waste the trip. | 04:30.30 |
mvrhel_laptop | ray_work: yes. I guess in this case I was lucky that they did not have it since I was able to get the nexus 10 | 05:17.41 |
| good night | 07:11.53 |
tor8 | Robin_Watts: morning. | 09:36.03 |
Robin_Watts | morning | 09:57.03 |
| tor8: There is a fix for a stupid typo on my repo. | 09:58.46 |
| You would prefer _arm than _ARM ? | 09:58.54 |
tor8 | Robin_Watts: I'm not sure, _ARM sticks out a lot better but how would it go with other architectures (should we against all odds add them) | 09:59.46 |
| Robin_Watts: I see we ticked off zeniko again :( | 10:00.02 |
Robin_Watts | yeah, my bad. | 10:01.29 |
kens | DUmb user questions :-( | 10:05.25 |
Robin_Watts | tor8: I liked _ARM because ARM is an acronym, and because it fitted with ARCH_ARM | 10:15.16 |
tor8 | Robin_Watts: then _ARM is fine | 10:15.26 |
Robin_Watts | Cool. | 10:15.33 |
| If you could nod through the fax fix, I'll get those pushed and go looking for what else I broke, | 10:15.56 |
tor8 | Robin_Watts: where is that fix? I don't see it. | 10:16.49 |
Robin_Watts | oh, sorry. | 10:17.01 |
| Ok, it's there now, sorry. | 10:18.13 |
kens | OK, off to ride, back later. | 10:19.16 |
Robin_Watts | tor8: http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=06ab03422110773744d6028b2d128702d2fbcfce - single character change. | 11:07.42 |
tor8 | Robin_Watts: *nods* | 11:15.17 |
Robin_Watts | Thanks. | 11:16.00 |
| Gah. alex_x6 is stuck updating test files again. | 11:35.22 |
paulgardiner | Robin_Watts, tor8: trivial little commit on paulg/master | 12:01.28 |
Chumbawamba | Hi folks | 12:09.51 |
Robin_Watts | Mid90sAnarchistBand: Morning. | 12:11.01 |
Chumbawamba | I'm currently using Ghostscript to reduce the filesize of my pdf's and converting their colorspace to sRGB. (Most input pdf's are CMYK). But for some reason no everything is convert to sRGB correctly | 12:11.05 |
| unable to convert colorspace to srgb, is the error I get | 12:11.21 |
| -sDEVICE=pdfwrite -dNOPAUSE -dBATCH -dPDFSETTINGS=/ebook -dColorConversionStrategy=/sRGB -dProcessColorModel=/DeviceRGB -sOutputFile=sdfsdf.pdf original.pdf | 12:11.56 |
| Anyone have a clue? | 12:12.11 |
Robin_Watts | Chumbawamba: You probably need kens, and he's out for a couple of hours. | 12:12.31 |
Chumbawamba | Okay | 12:13.15 |
| Another question, totally different: Is it possible to run Ghostscript multiple times simultanious? | 12:21.00 |
Robin_Watts | yes. | 12:26.14 |
Chumbawamba | I tried once, but the outputfiles got mixed up | 12:27.56 |
| This happened when convert the pdf to multiple images | 12:28.11 |
Robin_Watts | paulgardiner: looking now. | 12:39.52 |
paulgardiner | ta | 12:40.03 |
Robin_Watts | Urm... | 12:40.38 |
| So, you only ever mark leaf nodes as being dirty? | 12:40.59 |
| paulgardiner: ^ | 12:45.43 |
paulgardiner | Yes. The dirty flag means this annotations appearance stream is out of date. Only leafs of the field hierarchy are annotations. | 12:47.05 |
Robin_Watts | pushed. Thanks. | 12:47.22 |
paulgardiner | Brill | 12:47.33 |
Robin_Watts | Chumbawamba: The output files should only get mixed up if you specify the same names. | 13:03.34 |
| If I run: "gs -o out_foo%d.pdf -sDEVICE=pdfwrite foo.pdf" and "gs -o out_bar%d.pdf -sDEVICE=pdfwrite bar.pdf" at the same time then there should be no problems. | 13:04.44 |
Chumbawamba | Okay thanks, I will look into that | 13:05.12 |
| Highest priority is still the CMYK -> sRGB conversion. The pdf reader in my iPad app can't handle CMYK for some reason :S | 13:08.05 |
Robin_Watts | Should use a decent PDF reader app for the ipad then :) | 13:17.02 |
| <cough>MuPDF</cough> | 13:17.13 |
Chumbawamba | I'll give that a go, thanks | 13:21.59 |
| kens: Are you the guy that knows everything about CMYK -> sRGB conversion? :P | 13:48.44 |
kens | Umm, no, that wqould be Michael | 13:48.55 |
| But if you have a question I'll try to answer it | 13:49.08 |
Chumbawamba | I'm trying to convert my pdf's from CMYK to sRGB with Ghostscript | 13:58.54 |
| but it's just not working | 13:59.03 |
kens | Oh, well I don't think that's really possible right now | 13:59.13 |
Chumbawamba | I use this: "-dColorConversionStrategy=/sRGB -dProcessColorModel=/DeviceRGB -sColorConversionStrategyForImages=sRGB" | 13:59.36 |
kens | We don't support COlorConversionStrategyForImages | 13:59.57 |
Chumbawamba | Oh :S, I tought I read that somewhere on StackOverflow | 14:00.31 |
kens | StackOverflow is not omniscient ;-) | 14:00.47 |
Chumbawamba | Thats true :P | 14:01.01 |
kens | THe fact that we don't support it is documented I believe | 14:01.03 |
| Anyway, I'm working on overhauling the colour management in pdfwrite at this moment | 14:01.20 |
| What you want should be possible, but not for a few months | 14:01.37 |
Chumbawamba | :( | 14:03.05 |
| How come that the images that you can extract from the pdf with ghostscript, are rgb? | 14:05.40 |
| Isn't there some conversion in there either? | 14:05.47 |
kens | Yes, there are lots fo conversions | 14:06.08 |
| sRGB isn't really one of them | 14:06.17 |
| RGB is not the same as sRGB | 14:06.31 |
Chumbawamba | oh ok | 14:07.45 |
| but converting cmyk to rgb should be possible than? | 14:08.04 |
| then* | 14:08.09 |
kens | Yes | 14:08.16 |
Chumbawamba | -dColorConversionStrategy=/RGB -dProcessColorModel=/RGB -sColorConversionStrategyForImages=RGB | 14:08.45 |
| Should that do the trick? :P | 14:08.52 |
kens | set -dUseCIECOlor -sProcessColorModel=DeviceRGB -sColorConversionStrategy=RGB | 14:08.53 |
| The 'ForImages' switch is ignored | 14:09.18 |
Chumbawamba | oh yea | 14:09.40 |
kens | Oops type that should be -dUseCIEColor | 14:09.41 |
Chumbawamba | MMm, the conversion worked, but it's still not working on my iPad :( | 14:38.35 |
kens | How do you mean its 'not working' onyour iPad ? | 14:38.55 |
Chumbawamba | purple is now blue | 14:38.58 |
kens | Using what viewer ? | 14:39.09 |
Chumbawamba | Standard viewer from Apple | 14:39.59 |
| library | 14:40.15 |
kens | sounds like a viewer bug then | 14:40.21 |
Chumbawamba | There is no support for CMYK in that viewer, that's why I need to convert them | 14:40.50 |
kens | How does it view on Acrobat on a desktop machine (or Poppler, or whatrever) | 14:40.55 |
| No support for CMYK is just crap, you should get a new viewer, ogh wait, you cna't.... | 14:41.18 |
Chumbawamba | On desktop it works fantastic | 14:41.47 |
kens | THen in my opinion, iots the viewer | 14:42.21 |
Chumbawamba | yea I guess | 14:42.37 |
| Any other colorspaces that I can try? | 14:42.46 |
kens | Grey ? :-) | 14:42.55 |
| Does hte viewer support ICCbased spaces ? | 14:43.20 |
Chumbawamba | -dUseCIECOlor -sProcessColorModel=DeviceRGB -sColorConversionStrategy=RGB Makes it ICC right? | 14:44.16 |
kens | No | 14:44.22 |
Chumbawamba | That didn't work either | 14:44.22 |
kens | Makes it RGB | 14:44.26 |
| -dUseCIEColor -sColorConversionStrategy=LeaveColorUnchanged makes it ICC | 14:45.36 |
Chumbawamba | Acrobat says that the colorspace is ICCbased | 14:45.37 |
kens | You have an image in there ? | 14:45.51 |
Chumbawamba | yea | 14:46.02 |
| and vectors | 14:46.05 |
| and text | 14:46.10 |
kens | I'm not sure about the image it may end up in ICC space | 14:46.18 |
| Bascially, the current version of pdfwrite isn't intended for what you are trying to do, and isn't really up to the job. And you are trying to work around a deficiency in a viewer, so even if you do get ti to 'work' it may still not produce the expected/desired result on the broken viewer | 14:47.36 |
| As per your RGB output | 14:47.44 |
Chumbawamba | Mm I see | 14:49.34 |
| You know of any other programs that should do the trick, kens? | 14:53.32 |
kens | TO be honest, not really. You could try Quite a Box of Tricks, and I think Dwight Kelly at ETS had a tool which would do this, but these are both commercial utilities | 14:54.30 |
| We do recognise it as something useful to do, which is why I'm working on it | 14:54.56 |
Chumbawamba | Ok, thank you kens :) | 14:57.51 |
kens | you're welcome | 14:57.58 |
Chumbawamba | Learned alot | 14:57.59 |
Robin_Watts | marcosw: I put alex_x6 back into the cluster yesterday, as alexcher_ said he'd done a manual update on the test files. | 15:09.58 |
| I then took it out again this morning as it was stuck updating test files :( | 15:10.12 |
marcosw | Robin_Watts: okay then. | 15:11.26 |
Robin_Watts | tor8: ping | 15:20.38 |
tor8 | Robin_Watts: hey. | 15:21.09 |
Robin_Watts | http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=92e95c9f19bfe54077389ecde805dafc960a62c6 | 15:21.12 |
tor8 | *nod* | 15:21.23 |
Robin_Watts | http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=17db20e94e9973ed76b033fbe30afc3fed20e5b6 | 15:21.25 |
| Thanks. | 15:21.28 |
| Is that a nod for both? | 15:23.30 |
tor8 | yes! | 15:24.44 |
| a bit early nod, but yes to both :) | 15:24.53 |
Robin_Watts | Thanks :) | 15:25.10 |
| marcosw: Can we make the cluster tests for mupdf include tests_private/pdf/sumatra please? | 15:52.27 |
henrys | zeniko wrath ;-) | 15:55.34 |
Robin_Watts | watts incompetence -> zeniko wrath. He's been right on all counts so far :( | 15:55.55 |
| and he's being quite polite about it :) | 15:56.10 |
henrys | at your rate of production I would be very surprised if there weren't a few problems. | 15:58.57 |
Robin_Watts | Infinite monkeys... | 15:59.20 |
paulgardiner | MuPDF saving is broken. :-( Worked last time I tried it, but that was ages ago. | 16:15.14 |
| On the bright side, the Windows and Android app are now equally capable of corrupting a file | 16:16.03 |
Robin_Watts | bah. | 16:16.11 |
paulgardiner | If it was working at the time we swapped to git submodule for thirdparty, I may be able to use git bisect | 16:16.59 |
| Broken before that point. :-( | 16:22.51 |
| Is it possible to locate in the history where thirdparty changed, and do we have archives of the old thirdparty versions? | 16:25.00 |
Robin_Watts | paulgardiner: Can you not just bisect back further than that and not update submodules? | 16:25.26 |
paulgardiner | Wont build, will it? | 16:25.45 |
Robin_Watts | I suspect the differences required to make it build are very small. | 16:26.23 |
| and you can stash/stash pop at each bisect stage. | 16:26.34 |
| i.e. bisect back until it fails. Then make it build and test. | 16:27.02 |
paulgardiner | I thought we swapped to leaving the version numbers out of the thirdparty path names | 16:27.12 |
Robin_Watts | Then stash, then continue the bisect, then stash pop at the next point. | 16:27.21 |
| and so on. | 16:27.26 |
| hopefully you'll only need to do the editing once. | 16:27.34 |
| junction ? | 16:27.39 |
marcosw | Robin_Watts: I've worked on that, but the tests_private/pdf/sumatra filenames contain spaces and other characters which confound the cluster scripts. I've tried escaping them but there a lots of places and I keep giving up :-( | 16:27.40 |
Robin_Watts | marcosw: Ah! | 16:27.48 |
marcosw | if we remove ', (, ), and space we'd be good to go :-) | 16:28.18 |
Robin_Watts | Should we just rename them within svn then ? | 16:28.18 |
marcosw | that would be the easiest solution, but not sure if that would mess up tor8. Does he expect the names not to change? | 16:28.52 |
Robin_Watts | marcosw: tor8 isn't working from the SVN copy, AIUI. | 16:29.21 |
| tor8? | 16:29.42 |
paulgardiner | Surprised mupdfclean still works. It has to be a change to pdfwrite. Maybe the way it's called has changed, and I'm still using the old way. | 16:30.29 |
marcosw | we should also remove (or disable) pdfref17.pdf, that file takes a long time to process. | 16:30.32 |
kens | is happy, Nexus 7 just arrived | 16:31.25 |
Robin_Watts | paulgardiner: pdfwrite doesn't tend (so much) to change the contents of objects in the same way you do. | 16:31.57 |
| pdfclean, I mean. | 16:32.04 |
| so it may not be your fault. It may be a limitation in the saving. | 16:32.19 |
henrys` | kens:wow you got one too.. I just got it for mupdf demo'ing don't know what I'd use it for otherwise, except as a "kindle" | 16:32.30 |
paulgardiner | Hmmm. Maybe it's erroring out because I'm getting tiny files saved | 16:33.15 |
marcosw | I would remove ( and ), change space to _, and remove ' by changing don't to does not, etc. Shall I have at it and if tor8 complains we can revert? | 16:34.03 |
henrys` | it seems like something so broken would be more easily found by just debugging the current release. | 16:34.07 |
| but maybe not | 16:34.20 |
paulgardiner | henrys: Yeah. I'll try that first. | 16:34.43 |
Robin_Watts | marcosw: Sounds good to me. | 16:35.00 |
henrys` | marcosw:why can't it just work? Surely this is a perl solved problem. | 16:35.01 |
Robin_Watts | henrys`: Death by 1000 different quoting schemes. | 16:35.25 |
kens | henrys : bought one for STella to web surf on the sofa :-) | 16:37.27 |
henrys` | hmmph I have a perl rename script using just @ARGV and I've tried it all kinds of things spaces, commas etc it just works I'd be curious to see a file that fails under the cluster script | 16:37.32 |
Robin_Watts | henrys`: The problem is that the cluster feeds lists of names into lots of different things. If it was a matter of reading the names in 1 place then it'd be more easily solved. | 16:38.25 |
| For instance the cluster makes a list of commands to execute on the host that it then sends to each host. Those commands contain funky characters etc. | 16:39.33 |
marcosw | henrys`: the problem isn't perl, it's when the jobs are spanned via fork and a system call, we use bash because of the way redirection works and it's technically possible to escape everything but I've already spent more hours than I'm willing to admit working on it and every time I think I'm done I find something still not working. | 16:39.37 |
| last week I was pretty close but then realized that bmpcmp would fail in various ways and got discouraged. | 16:40.09 |
henrys` | I see yes I was just thinking of the perl and not the system stuff, but it would be nice to have it working. | 16:41.12 |
marcosw | if we had started out with spaces, etc. in file names presumably all the code would have been written to allow for this but now it's really messy to change. | 16:41.54 |
Robin_Watts | tor8, paulgardiner: 3 small commits for review... | 16:53.20 |
| http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=ca19703811bd41bd2638d7d8678d1702aa56844f | 16:53.30 |
| http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=3ac9a4da1943e93a0e3e9d64b56228c7c4383b62 | 16:53.41 |
| http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=317a4aeac50a35af8020dbfe10a66d9f2d7e3b50 | 16:53.52 |
paulgardiner | First one's fine. I can't see anything obviously dodgy in the second, but I can't say I've understood every line. | 17:00.28 |
Robin_Watts | oops. I think I just spotted a leak in the second. | 17:02.27 |
paulgardiner | 3 is interesting. What are you doing there? | 17:04.18 |
Robin_Watts | The NDK has a bug whereby if I compile certain files with profiling the compiler dies. | 17:05.28 |
| so I moved the 2 effected files out into Core2.mk | 17:05.45 |
paulgardiner | Think I'm close to finding the saving problem, btw. Pretty sure a stream has had it's length set to a num object rather than an int. Almost certainly my doing: I had a day when I was confusing ints and nums. | 17:05.46 |
Robin_Watts | but I figured out how to fix one of them, so that commit moves it back from Core2. to Core. | 17:06.29 |
paulgardiner | Ah right. Nice. | 17:07.02 |
| Hmmm. Well I hadn't made the num/int mistake where I thought I might have. | 17:10.22 |
kens | Night all | 17:17.26 |
tor8 | Robin_Watts: fz_transform_pixmap(fz_draw_device *dev, I dislike passing the draw device there | 17:30.21 |
| would much prefer if you passed the cache specifically | 17:30.39 |
| caches, even | 17:30.43 |
Robin_Watts | tor8: howso? It's a static function, right? | 17:30.49 |
tor8 | oh, right! I'm confusing it with the next one in the chain. | 17:31.30 |
Robin_Watts | ah, ok. | 17:31.40 |
tor8 | we've rejigged these too often for me to keep up :) | 17:31.48 |
Robin_Watts | possibly that should lose the fz_ prefix to make that clearer. | 17:32.05 |
tor8 | Robin_Watts: I've gone back and forth on that endlessly... keep or lose the fz_ prefix for static functions | 17:32.45 |
Robin_Watts | tor8: Did you see the discussion earlier about sumatras file naming in SVN ? | 17:36.42 |
| The presence of spaces, commas, parens and quotes upsets the cluster. | 17:37.04 |
| We're tempted to just rename the files once and for all and be done with it. | 17:37.19 |
| Would that affect you in any way? | 17:37.24 |
tor8 | Robin_Watts: no, missed that conversation. won't affect me. | 17:41.28 |
Robin_Watts | tor8: fab. | 17:41.38 |
henrys` | happy birthday robin_watts I think I forgot to tell you yesterday. | 18:07.30 |
Robin_Watts | henrys: Thanks. | 18:07.49 |
| Roger Waters is touring the Wall again next year, this time to stadiums rather than arenas. | 18:08.26 |
| I hope I'm still that game when I hit 68. | 18:08.54 |
henrys` | a friend of mine went to see Crosby Stills and Nash or some subset of that - one of them is past 70 | 18:10.15 |
Robin_Watts | yeah. The Wall is a big undertaking though, and he's very front and centre in it. | 18:11.30 |
henrys` | true... so without gillmore? | 18:11.55 |
Robin_Watts | indeed. (Though Gilmour and Mason joined in for 1 night of the last tour in order to boost the DVD sales :) ) | 18:12.34 |
henrys` | that's really the part of floyd I liked - an underrated guitarist I believe. | 18:13.12 |
Robin_Watts | henrys`: I live his guitar sound, but I've been unwhelmed by his solo stuff, sadly. | 18:13.53 |
| s/live/love/ | 18:14.01 |
| but I did like the post waters floyd albums (and indeed the post floyd waters albums) | 18:14.31 |
henrys` | I didn't go for the solo stuff at all. I should say gillmour's work with pink floyd I thought was brilliant. | 18:16.44 |
Robin_Watts | I don't know who they had playing Gilmours part when Chrisl and I saw the Wall last year, but he did a brilliant job. | 18:19.41 |
henrys` | I think I have a pcl font selection breakthrough and I'm really excited - what a sad life I have ... | 18:20.39 |
Robin_Watts | hehe. | 18:20.47 |
| Forward 1 day (to 2012/11/22)>>> | |