| <<<Back 1 day (to 2013/09/03) | 2013/09/04 |
kens | chrisl ping | 08:19.43 |
chrisl | kens: pong | 08:20.14 |
kens | did you want to read book 2 of the merchant princes ? | 08:20.26 |
kens | can't remember the title right at the moment | 08:20.46 |
chrisl | Hmmmm, do I......? | 08:20.48 |
| TBH, I rather resent the assumption in the first book that I *will* read the rest.... | 08:21.32 |
kens | A lot of people seem to feel that way | 08:21.48 |
| Though its not unusual in books deliberatley written as a series | 08:22.13 |
chrisl | I guess not, I just don't expect it from Stross | 08:23.21 |
kens | Well most of his other works are pretty stand alone, even the Laundry series is really a set of individual novels | 08:24.07 |
chrisl | I think I'll leave it - maybe next time | 08:24.39 |
kens | OK no problem, more room in my bag that way :-) | 08:24.50 |
chrisl | Well, I was going to bring the first book back to you, so.... | 08:25.16 |
kens | OK must leave room for that, thanks | 08:25.30 |
Robin_Watts | sebblonline: I'm still looking into the leaks. It's in our transparency stack handling code, and it looks a bit vicious. | 08:26.29 |
chrisl | I've just started rereading a Greg Bear book - I want to see if it's as good as I remember it | 08:26.31 |
kens | which one ? | 08:26.41 |
chrisl | Darwin's Radio | 08:26.48 |
kens | Ah, don't think I;ve read that one | 08:26.57 |
chrisl | I think it's the best of his I've read, by some distance. If I finish it over weekend, you can take it - or I'll bring it to the next meeting for you | 08:27.49 |
kens | Great, thanks! | 08:28.01 |
chrisl | Off to squash | 08:32.03 |
sebblonline | Robin_Watts: Hey | 08:53.34 |
| thanks so far. | 08:53.40 |
| so the document is vicious? :) | 08:53.51 |
Robin_Watts | Its to do with nested smasks, I think. | 09:02.34 |
| I can't promise to solve it before I have to travel thursday -> monday. | 09:02.57 |
| paulgardiner, kens, chrisl_away, tor7: just a reminder to check in for flights. | 11:22.38 |
| chrisl: does zoltans mail mean we need to redo the 9.10 commercial release? (Just add the missing dir to the archives?) | 11:31.31 |
| henrys said: "I'm from colorado my bike costs more than my car." | 11:43.21 |
| I've had cars like that too. | 11:43.25 |
chrisl | Robin_Watts: it just means I need to fix the commercial release script, and run it again..... | 11:53.42 |
kens | is relieved it wasn't his fault this time | 11:58.59 |
Robin_Watts | kens: online checkin time if you haven't done so yet. | 12:01.11 |
kens | Robin_Watts : was about to do so, just got back | 12:02.15 |
zeniko | Robin_Watts: there's another round of patches on zeniko/mupdf for when you find some time | 12:11.12 |
Robin_Watts | zeniko: Thanks. I will try to look, but I'm mired in a bug at the moment, and I've got to pack for travelling later today. | 12:11.45 |
| We'll pretty much all be away from thursday -> monday. | 12:12.05 |
| So bear with me if they don't all get looked at before next week. | 12:12.52 |
zeniko | No worries, next week would be fine just as well | 12:13.40 |
| That bug of yours about a softmask leak? | 12:13.53 |
Robin_Watts | zeniko: indeed. | 12:16.59 |
sebras | what happened to all the tags?! | 12:17.08 |
Robin_Watts | it looks to me like we're not handling the transparency stack quite right in the softmask case. | 12:17.27 |
zeniko | Robin_Watts: IIRC, I've hit and fixed such an issue at one time | 12:17.36 |
Robin_Watts | but maybe the code is just really cunning and I'm too thick to understand it now. | 12:17.39 |
| zeniko: Ah. | 12:17.45 |
| which tags? | 12:17.52 |
zeniko | had to move the line "gstate->softmask = save->softmask;" from end_softmask to begin_softmask | 12:18.19 |
sebras | Robin_Watts: I'm doing git fetch --all --prune and each repository I fetch from is removing and re-adding tags in the git repo. | 12:18.25 |
Robin_Watts | sebras: I have no idea if that's right or not :) | 12:19.09 |
zeniko | Robin_Watts: and on top of that clear the softmask in begin_softmask if it was for an xobject | 12:25.09 |
| (which apparently was needed at least for Bug6901014_CityMap-evince.pdf) | 12:25.47 |
Robin_Watts | zeniko: Thanks. I'll keep all that in mind. | 12:28.19 |
| Fundamentally it looks like the graphics calls we are making aren't properly nested though, but when I try to fix that I hit asserts and things. So I need to try to understand it a bit better. | 12:28.54 |
| It may be that I end up agreeing with your fix, but I need to work it through in my head. | 12:29.13 |
henrys | no devices folder okay release testing gets some agenda exclamation points | 13:33.24 |
chrisl | henrys: if we add much more testing, it'll take a month to do a release! | 13:38.33 |
henrys | chrisl: I think there is a lack of staff pulling down the code and going through it carefully, that can overlap with any other testing going on. | 13:39.55 |
chrisl | henrys: well, to make that worthwhile, I'd have to do two lots of archives for every release candidate, too. | 13:40.55 |
henrys | chrisl: I am just seconding your agenda submission about release candidate testing. | 13:42.49 |
chrisl | henrys: sure, but in this case it was *only* the commercial release that had the problem, and I currently don't make release candidates for those | 13:43.24 |
| henrys: in this case, I think I need to rejig that part of the commercial release script to exclude certain directories, rather than include an explicit list. | 13:45.27 |
henrys | chrisl: hmm tricky the other way (include all) could do something we didn't want also, but that is probably unlikely. | 13:48.01 |
chrisl | henrys: it's much smaller list of directories we want to exclude than include, and almost never changes. I figure the smaller, more static list is easier to maintain. | 13:50.03 |
henrys | chrisl: makes sense | 13:50.34 |
chrisl | I'll make a note to do that next week. | 13:50.52 |
mvrhel_laptop | good morning | 13:52.39 |
kens | Morning Michael | 13:52.48 |
henrys | howdy mvrhel_laptop | 13:53.50 |
| mvrhel_laptop: I added your agenda items - do you have more specifics about cloud printing? Or did you want to just talk about it there? | 13:54.55 |
unK2013 | Hi all | 13:56.53 |
| need help on mupdf 1.2 | 13:57.03 |
| trying to detect the first char of the line is a space or a tab.... can someone help me understand how to do it? in the dev_text.c | 13:58.07 |
kens | : unK2013 you need Robin_Watts (hellooo Robin) | 13:59.06 |
unK2013 | kens, thanks... can he be expected around this time? | 13:59.27 |
kens | unK2013 : he's here, I see his nick in the list, but maybe away from keyboard | 13:59.50 |
| We are travelling to a staff meeting tomorrow, so he may be packing | 14:00.02 |
chrisl | (or hiding!) ;-) | 14:00.05 |
henrys | unK2013: probably packing | 14:00.11 |
unK2013 | henrys, Ummm .... thanks! | 14:00.40 |
henrys | unK2013: maybe a bad time to get help but Robin_Watts or tor7 are likely to answer your question when available. | 14:01.32 |
unK2013 | henrys, Sure, I will come back again sometime to catch them up! Thanks to all the guys here!\ | 14:02.20 |
Robin_Watts | unK2013: Hi | 14:17.05 |
mvrhel_laptop | henrys: nothing specifics. we can talk about it there | 14:17.12 |
| Robin_Watts: are you working on a soft mask issue in gs or mupdf? | 14:17.23 |
Robin_Watts | mupdf | 14:17.29 |
henrys | going out to breakfast today, bbiab | 14:18.13 |
Robin_Watts | henrys: order bacon. | 14:18.23 |
| unK2013: The first char of a line is unlikely to be a space or a tab. Are you trying to detect indents? | 14:18.53 |
henrys | I gotta watch my figure... | 14:19.02 |
Robin_Watts | henrys: and an isotonic milkshake :) | 14:19.16 |
unK2013 | Robin_Watts, yes trying to detect indents | 14:21.22 |
| Robin_Watts, should we just compare x1 for lines? | 14:21.45 |
sicos | hi | 14:21.56 |
ghostbot | niihau | 14:21.56 |
Robin_Watts | The first step should be to understand the way PDF sends text. Essentially, it says "put this char here". | 14:22.03 |
| Sometimes they don't even bother sending spaces. | 14:22.18 |
| certainly they never send "tabs". | 14:22.23 |
sicos | this probably sound like a stupid question.. but is it possible with ghostscript to input a pdf file with text and images and output a pdf file that only contains images | 14:22.32 |
Robin_Watts | They needn't even send chars in the obvious order. | 14:22.38 |
| unK2013: I would strongly advise you to start from mupdf 1.3 rather than 1.2 | 14:23.00 |
| as 1.3 has advances in this area. | 14:23.08 |
tor7 | sicos: you could use gs to render to images and then another tool (like imagemagick) to create a pdf from these images | 14:23.15 |
kens | sicos, and do what with the text ? | 14:23.34 |
Robin_Watts | sicos: You could write a device for gs that 'filters' the text operations out. | 14:23.37 |
unK2013 | Robin_Watts, I tried 1.3 but for some reason it is dumping core | 14:24.03 |
sicos | convert all the text to images | 14:24.04 |
kens | Robin_Watts : wouldn't fancy makingthat work with pdfwrite | 14:24.10 |
unK2013 | Robin_Watts, 1.2 remained stable that way | 14:24.16 |
sicos | something like convert the pdf to png and put those back in to convert it back to pdf | 14:24.25 |
| so the pdf only contains images | 14:24.31 |
Robin_Watts | unK2013: Your code + mupdf 1.3 is dumping core? or vanilla 1.3 is dumping core? | 14:24.43 |
kens | sicos use GS to render to an image ofrmat and then 'somethign' else to convert the image to PDF< or to PostScript and send the PostScript to GS for conversion | 14:24.47 |
Robin_Watts | mupdf 1.3 changed the text extraction structures, so calling code will need to be updated. | 14:25.05 |
sicos | ok.. already thought about that but just wondered if it was possible to do it directly | 14:25.23 |
kens | sicos no, not really, we go to quite a lot of effort *not* to do that :-) | 14:26.45 |
sicos | hehe .. the only reason why I did ask this is because we use streamserve from OpenText | 14:28.31 |
| And OpenText did something stupid in that software package | 14:28.43 |
| When streamserver (a print tool) want to print a pdf it starts to read the pdf.. when it finds a font that it doesn't regocnize it stops processing the pdf | 14:29.37 |
kens | Hmm, not ideal, a substituion is the usual result. But it may not be possible depending on the font. Best practice is to embed fonts, always | 14:30.20 |
sicos | Yeah that is what I did with ghostscript.. font substituion .. | 14:30.58 |
| Also part of the fonts that are used | 14:31.13 |
| But streamserve doesn't understand that | 14:31.28 |
kens | Its possib le that PDF->PDF willl embed the substitute fonts, but it depends | 14:31.29 |
sicos | So I'm going to solve it by converting the PDF to PNG and then back to PDF again | 14:32.05 |
| that way there are no more fonts since everything is an image | 14:32.18 |
kens | seems reasonable | 14:32.20 |
sicos | but I need to go.. thanks for the chat | 14:33.26 |
| bye | 14:33.28 |
Robin_Watts | mvrhel_laptop: I pushed my changes to the antidropout downscaler this morning - your polarity stuff was just what I needed, thanks. | 15:04.29 |
mvrhel_laptop | Robin_Watts: ok great! | 15:06.01 |
Robin_Watts | Crumbs. The cluster is turbotastic at the moment. | 16:31.34 |
chrisl | 24 nodes! We might need to split the dashboard into two columns.... | 16:33.13 |
Robin_Watts | chrisl: That's without the aws ones that marcos can add in. | 16:33.40 |
kens | Its dead quick, has been for weeks | 16:33.51 |
Robin_Watts | splitting the dashboard into 2 columns would be bad when the description column gets wide. | 16:34.11 |
chrisl | We could drop the job names from the "waiting to complete" message | 16:34.54 |
Robin_Watts | sebblonline: ping | 16:39.39 |
sebblonline | pong | 16:40.14 |
Robin_Watts | sebblonline: http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=515c0106845a7dafd71abba6b29305e39c88380e | 16:40.54 |
| That should solve it, I think. | 16:40.58 |
| tor7, paulgardiner, sebras: Fix on robin/master for review please. | 16:41.09 |
sebblonline | great, thanks alot for the help and fast responses | 16:42.11 |
| i will try that | 16:42.17 |
kens | Well I asked for my boarding pass as a PDF by email, won't do that again. | 16:56.57 |
chrisl | Mine came in seconds this time - very unusual | 16:57.49 |
kens | Mine hasn't arrived at all | 16:57.58 |
| Been 5 hours and nothing | 16:58.08 |
| Its not a huge deal, jsust means I have to head to the desk, but I'm going back to printing it to a PDF myself | 16:58.47 |
Robin_Watts | kens: yeah, given that when people are doing their boarding passes they might be logged into hotel computers and only have a few minutes, you'd think they'd get their asses in gear. | 16:59.02 |
| kens: You can still log in and print it now, I believe. | 16:59.16 |
kens | Robin_Watts : its says '20 minutes' | 16:59.23 |
| I'll try logging in and pritning it, wouldn't let me earlier (error 500) but we'll see | 16:59.39 |
sebras | Robin_Watts: any particular fix, or all of them on robin/master? | 16:59.57 |
chrisl | kens: Last time mine took 16 hours, I think, but I was able to log in a print it directly from the site | 17:00.12 |
sebras | scratch that.. I forgot to pull... | 17:00.27 |
kens | Then I'll do that, thanks both of you | 17:00.29 |
Robin_Watts | sebras: ah, phew. there should just be 1. | 17:00.38 |
sebras | Robin_Watts: 300Mb! wow! I'm sure you appreciated every minute of the download time... ;) | 17:02.13 |
| Robin_Watts: I'll take a closer look later. need to run. | 17:02.35 |
kens | worked a treat, thanks Robin_Watts chrisl | 17:03.24 |
chrisl | Good - as I said, the mail will probably arrive over night...... | 17:03.42 |
kens | So I'm off to pack, see you all tomorrow | 17:03.46 |
chrisl | Yeh, see you then | 17:03.54 |
mvrhel_laptop | henrys: you are correct about the profile call | 17:53.08 |
| gs_setdevice_no_erase will get us the correct default profile | 17:53.34 |
| and there is no need for pcl, xps, and friends to call gsicc_init_device_profile_struct | 17:54.08 |
| Do you want to fix this or do you want me to take care of it? | 17:54.49 |
| I can go ahead and do it | 17:57.18 |
| cluster testing fix now | 18:06.17 |
| looks ok. pushing fix henrys | 18:37.45 |
| bbiab | 18:38.18 |
henrys | mvrhel_laptop: sorry I was away from irc⦠looks like it's add done though | 19:48.28 |
| s/add/all | 19:48.39 |
Leo | hi | 20:13.36 |
ghostbot | hey | 20:13.36 |
Leo | I can ask a question? | 20:14.47 |
henrys | Leo:go ahead you may have to come back later for an answer though | 20:16.59 |
Robin_Watts | plinnell: hi. did you open a bug about mupdf having a build problem on SUSE? | 21:38.40 |
zygew | i am tired of reading code all my lone,any chance to change the background color of the mupdf? | 21:45.19 |
| canvas | 21:52.16 |
Robin_Watts | zygev: should be easy enough, sure. | 21:52.39 |
zygew | lol | 21:53.04 |
Robin_Watts | zygew: Look in platform/x11/pdfapp.c | 21:53.46 |
| look for calls to fz_clear_pixmap_rect_with_value | 21:54.00 |
| lines 684 | 21:54.26 |
| and fz_clear_pixmap_with_value on line 784 | 21:54.43 |
| Those are the lines that clear the background color. | 21:54.53 |
zygew | lucky | 21:55.17 |
Robin_Watts | The problem is that you need to set the background to rgbargbargbargba etc. | 21:55.28 |
| currently we set them all to 255, so solid white. | 21:55.40 |
| you probably need to set it to r,g,b,255 | 21:55.54 |
| so you might need to roll your own pixmap clearing functions, but you'll see they aren't exactly hard. | 21:56.11 |
| actually, those functions are set up to set the value to v,v,v,255 | 21:57.12 |
| so you can have a greyscale background :) | 21:57.21 |
| Or you can just use the 'i' key to invert the colors that are rendered which might be easier. | 21:57.40 |
zygew | i will give it a try | 21:58.25 |
| latter on, | 22:00.21 |
| seems like it is homebrew coffee.. | 22:05.15 |
| Forward 1 day (to 2013/09/05)>>> | |