| <<<Back 1 day (to 2015/03/12) | 20150313 |
mvrhel_laptop | aha! found the damn issue with the overprint. the overprint compositor is not getting set idle. right before the image enumeration there is a graphic state pop which changes the overprint state from true to false. | 06:08.58 |
| unfortunately, the current color in the graphic state at that point in time is a pattern. | 06:09.14 |
| and | 06:09.43 |
| static int | 06:09.44 |
| gx_set_overprint_Pattern(const gs_color_space * pcs, gs_state * pgs) | 06:09.46 |
| { | 06:09.47 |
| return 0; | 06:09.49 |
| } | 06:09.50 |
| does not make the overprint compositor idle | 06:10.03 |
| so the overprint compositor stays active during the image fill | 06:10.30 |
| the comment above the function says | 06:11.35 |
| Set the overprint compositor for a Pattern color space. This does nothing; | 06:11.46 |
| for patterns the overprint compositor is set at set_device_color time. | 06:11.52 |
| will need to think about this, but at least I have made a bit of progress. burnt up a day on this | 06:13.03 |
| night all | 06:13.23 |
chrisl | Morning kens | 11:01.50 |
kens | Morning chrisl | 11:02.03 |
chrisl | Did you get a look around St John's? | 11:03.14 |
kens | I don't think there was much to see, it was bitterly cold and I was exhausted, so no not really | 11:03.38 |
| |'ve seen quite a bit of the airport | 11:03.54 |
chrisl | I don't think you missed much! | 11:04.02 |
kens | THey were offering a 3 hour tour of 'the historic harbour' for $50, that seemed to be the highlight | 11:04.49 |
chrisl | Was the hotel nice, at least? | 11:05.36 |
kens | It was absolutely fine. THe only place to eat was a strange Italian place, but to be honest that was OK | 11:05.56 |
jogux | does the iOS mupdf rely on the system fonts? | 13:19.11 |
Robin_Watts | I don't believe so. | 13:22.29 |
| Certainly windows/linux/android don't. | 13:22.50 |
jogux | cool | 13:23.04 |
Robin_Watts | Morning mvrhel_laptop. | 15:22.15 |
| I pushed that commit. | 15:22.20 |
mvrhel_laptop | Robin_Watts: ok thanks. I will pull it and get things changed in the gsview project | 15:22.47 |
| to grab the libs from the right place | 15:23.00 |
| Robin_Watts: so I think I found the overprint issue. I think this problem has always existed in gs | 15:23.26 |
| or at least there has always been the potential for such a problem | 15:23.43 |
Robin_Watts | yeah, I saw the comments in the logs. | 15:24.46 |
| sounds like the usual kinda crap :( | 15:24.53 |
mvrhel_laptop | yes | 15:24.57 |
| the file is large and super sensitive to any changes which made it tricky to see where things were going wrong | 15:25.24 |
| luckily the clist was not involved though ;) | 15:25.40 |
Robin_Watts | aw.... :;) | 15:27.38 |
fredross-perry | Yesterday I was looking into bug #695040. I believe I found a fix that involves freetype. How should this be handled? Does Artifex maintain freetype? Thanks. | 15:55.01 |
kens | No FreeType is a 3rd party library | 15:55.14 |
| We supply fixes upstream to Werner | 15:55.23 |
| Bets hting is to pass it through Chrisl | 15:55.31 |
| Or me, but today better be chris, I'mnot well | 15:55.41 |
fredross-perry | OK. I also noticed that the freetype thatâs submoduleâd to mupdf is over 2 years old? | 15:56.31 |
Robin_Watts | fredross-perry: If there is a good reason to update it, we can do that. | 15:56.54 |
kens | Can't commetn there, I thougth you were talking about GS | 15:56.55 |
chrisl | It would be good to check whether it's already solved in newer freetype versions - I guess I can do that if you want | 15:57.42 |
fredross-perry | no thanks. In fact it is. Thatâs my concern. | 15:58.09 |
| do we have a process for folding in/testing newer thirdparty code into mupdf? | 15:58.44 |
kens | Then it sounds like MuPDF should upgrade | 15:58.48 |
chrisl | We do tend to be a little lazy with freetype updates because they *always* result in rounding differences to the outlines, and thus cause a *lot* of diffs in the test suite. Hence we tend to wait for a compelling reason | 15:59.33 |
fredross-perry | I am happy to upgrade, if I can get some guidance regarding testing and such. | 15:59.34 |
Robin_Watts | fredross-perry: Yeah, we pull the changes into the submodule, then we do a mupdf commit that changes the submodule pointer. | 15:59.48 |
fredross-perry | not a git question, more of an org question. | 16:00.16 |
chrisl | fredross-perry: pull in the newer version, do a clusterpush/bmpcmp and look at the differences | 16:00.28 |
Robin_Watts | fredross-perry: We pull the changes into the submodule git thing (that has to happen on golden). | 16:00.37 |
| Then the commit to the mupdf repo can be done on your personal repo. | 16:00.58 |
| That can then be tested, and reviewed. | 16:01.04 |
| When we're all happy, that goes to golden. | 16:01.11 |
fredross-perry | But do we just grab the head, or some other version that Werner might ID? | 16:02.32 |
kens | Ordinarily, the latest release | 16:02.45 |
| FT releases fairly often | 16:02.51 |
Robin_Watts | (When I say "that has to happen on golden" I mean "that has to happen on our public facing thirdparty repos (such as git.ghostscript.com:/home/git/thirdparty/freetype.git)" | 16:03.01 |
| yeah, latest release, I think. | 16:03.13 |
fredross-perry | how is the latest release identified? for example, FT has a 2-months-ago tag called VER-2-5-5. | 16:04.18 |
kens | Give me a second, I'll look for the lats release annoucnement | 16:04.36 |
| Looks like CDecember was the last release | 16:05.22 |
henrys_ | Robin_Watts: you really like your MS keyboard right, is it quiet? | 16:05.22 |
Robin_Watts | I do, but... | 16:05.39 |
| 1) You can't buy this model any more. | 16:05.45 |
kens | CUrrent release seems to be FreeType 2.5.5 | 16:05.46 |
chrisl | fredross-perry: yes, use the git tag VER-2-5-5 | 16:06.04 |
Robin_Watts | 2) It's not the most clacky thing known to man, but it's not silent. | 16:06.05 |
| 3) it's PS/2 only. | 16:06.09 |
| I thought you had a nice gamer keyboard? Is that too noisy? | 16:06.38 |
kens | You can get PS/2 to USB dongles | 16:06.41 |
Robin_Watts | kens: You can, indeed. | 16:07.05 |
fredross-perry | I thought it was now unlawful to own a PS/2 kbd | 16:07.25 |
kens | Ah you cna also get bluetooth dongles if you want it wireless | 16:07.30 |
chrisl | fredross-perry: not unlawful, just socially unacceptable ;-) | 16:07.58 |
henrys_ | Robin_Watts: love the keyboard but it's loud as hell | 16:08.05 |
Robin_Watts | http://www.amazon.com/Microsoft-Natural-Ergonomic-Keyboard-4000/dp/B000A6PPOK/ref=sr_1_1?ie=UTF8&qid=1426262845&sr=8-1&keywords=Microsoft+natural+keyboard | 16:08.26 |
| That's probably the latest version. | 16:08.33 |
kens | is very tempted | 16:08.42 |
chrisl | is not sure about a keyboard that looks like it melted | 16:09.12 |
Robin_Watts | fredross-perry: I have a KVM that only takes PS/2 (and one of the computers that's connected to the KVM is a very strange thing with a non-standard mouse interface that I have a PS/2 converter attached to) | 16:10.05 |
henrys_ | right I want a keyboard not a Dali sculpture ;-) | 16:10.06 |
kens | I find them way more comfortable than regular qwerty designs | 16:10.26 |
Robin_Watts | henrys_: hold your hands out in front of you, in their most comfortable position. | 16:10.36 |
kens | I had one way back when MS first intrduced them | 16:10.37 |
| The Amazon reviews suggest this one is quieter than previous models | 16:11.10 |
Robin_Watts | You should find that they aren't straight out, but rather face inwards to meet. | 16:11.18 |
henrys_ | I know they're better ... I don't know if I can change my ways though | 16:11.29 |
Robin_Watts | The bend in the keyboard therefore feels entirely natural. | 16:11.33 |
chrisl | One problem is that I regularly combine using the mouse and hotkeys, so I end up using the keyboard one handed | 16:11.56 |
kens | I wasn't convinced until I tried one, I was very sad when my old one self-destructed | 16:12.05 |
fredross-perry | robin: whatever you need to tell yourself. | 16:12.06 |
Robin_Watts | henrys_: The downside to buying one of these is you might start to look like one of those nutters that has a standing desk :) | 16:12.07 |
| henrys_: I suspect it would take you, oooh, maybe 2 minutes to get used to it. | 16:12.53 |
fredross-perry | If I had ever learned to type properly, the MS kbd might work for me. | 16:12.56 |
henrys_ | Robin_Watts: I'm thinking about it... | 16:13.08 |
| I assume you don't use wireless, slow for gaming? | 16:13.28 |
Robin_Watts | henrys_: Wired FTW :) | 16:13.59 |
| I've never found anything where wireless was better than wired (except for the convenience of WiFi, but you pay for that in unreliability) | 16:14.36 |
henrys_ | Robin_Watts: minimizing wires become a priority with the up and down desk. | 16:15.48 |
Robin_Watts | http://www.amazon.com/Microsoft-Ergonomic-Keyboard-Business-5KV-00001/dp/B00CYX26BC/ref=sr_1_1?ie=UTF8&qid=1426263421&sr=8-1&keywords=Microsoft+natural+keyboard+wireless | 16:18.17 |
mvrhel_laptop | henrys: still planning on another gsview beta release hopefully today. I have the page range working for printing. these gs overprint devicen issues have been distracting | 16:22.51 |
| getting ready to test a fix for the overprint issue now | 16:23.09 |
henrys_ | mvrhel_laptop: unfortunately that's a high priority customer. | 16:23.45 |
mvrhel_laptop | the overprint issue is? | 16:23.59 |
| or the devicen you mean | 16:24.10 |
henrys_ | mvrhel_laptop: do you have all the slides from the last trip or does Miles handle all that? I'd like to review them today before the trip. | 16:24.18 |
mvrhel_laptop | henrys: miles has all of those | 16:24.38 |
henrys_ | mvrhel_laptop: I saw a boatload of stuff from Ray which I was referring to. | 16:24.44 |
mvrhel_laptop | oh that is another one | 16:24.51 |
| I have not seen those bugs yes | 16:24.59 |
| I just knew the file | 16:25.06 |
| and what he is describing sounds to me like it is not going to be a bug | 16:25.20 |
| they can't get CMYK rendering results for that file when doing RGB rendering | 16:25.35 |
| not going to happen | 16:25.39 |
| AR has huge differences if you change the profile from the output CMYK intent to sRGB | 16:26.05 |
| likewise we will too | 16:26.18 |
henrys_ | mvrhel_laptop: okay well that's good, so you're mostly working on the color customer? | 16:27.51 |
| mvrhel_laptop: I wonder if gsview shouldn't wait till after the release anyway so you have our latest release inside. | 16:28.22 |
mvrhel_laptop | henrys_: yes good point. but we can internally play with the beta | 16:28.57 |
| I want to at least get that out | 16:29.06 |
| keeping figures crossed this overprint cluster test works. this is a really odd one | 16:30.12 |
| yea! that fixed the issue and no regressions | 16:48.14 |
kens | great! | 16:48.22 |
henrys_ | mvrhel_laptop: nice | 16:48.31 |
mvrhel_laptop | no progressions either. I wonder if we should add my simplified file to the regression tests | 16:48.43 |
| psdcmyk would have shown the issue | 16:49.09 |
henrys_ | marcosw, mvrhel_laptop: I was just talking to marcosw about having an upload test file on the regression dashboard. I think that would be easy to do. plenty of examples on the web that use a simple php to do the uploading and an html submit form. | 16:55.45 |
Robin_Watts | henrys_: we'd need to upload it, and insert it into the right place in the SVN hierarchy. | 16:57.55 |
| It'd be easy enough to knock something up I suspect. | 16:58.39 |
henrys_ | Robin_Watts: you'd have to say what sort of file it is. | 16:59.14 |
| problem now is I don't think anybody checks in test files because the damn thing is too huge to checkout. | 16:59.49 |
| well at least I would contribute more files if there were a simple way to do it. I'm not sure about others. | 17:01.39 |
Robin_Watts | henrys_: Let me have a play later. | 17:01.47 |
| (web based dashboard etc is largely my baby now) | 17:02.00 |
henrys_ | Robin_Watts: okay thanks | 17:02.23 |
Robin_Watts | got the memory usage for torture-test.pdf down from 95Meg to 70Meg on mupdf. | 17:27.11 |
| oops: http://www.zdnet.com/article/panda-antivirus-mistakenly-flags-itself-as-malware-breaks-pcs/ | 17:28.43 |
mvrhel_laptop | grrr. it doesnt matter how much I do, I always find something more to fix on gsview. in any event, it is close enough for me to throw it up to casper for people to beat on | 18:18.50 |
| printing seems reasonably fast now though | 18:19.04 |
| a couple things to clean up with respect to it | 18:19.31 |
| with printable area | 18:19.38 |
| vs. page size | 18:19.52 |
| and some feedback while printing pages | 18:20.09 |
| progress bar does for printing is not working nor does it go away :) | 18:20.24 |
| fredross-perry: where do you have the linux and mac installers? | 18:22.40 |
| I am going to send out an email asking people to beat on these a bit | 18:22.58 |
fredross-perry | in my public html area on Casper | 18:24.35 |
mvrhel_laptop | oh ok great | 18:24.42 |
fredross-perry | so Iâll expect a barrage of bug reports soon. | 18:27.03 |
mvrhel_laptop | yes. I hope so. | 18:27.09 |
| Robin_Watts: ok email sent on where to find the installers | 18:29.06 |
Robin_Watts | mvrhel_laptop: OK. I promise I will look. | 18:29.17 |
mvrhel_laptop | thanks! I expect your usually no stone left unturned review ;) | 18:29.44 |
Robin_Watts | mvrhel_laptop: I have a policy of finding *something* on every review, just to prove I looked. If I'm complaining about your comment style, you know you're doing well :) | 18:30.58 |
mvrhel_laptop | :) | 18:32.13 |
Robin_Watts | mvrhel_laptop: Can you mail me an example file that should be added to the tests please? | 19:16.06 |
mvrhel_laptop | Robin_Watts: you mean for our regression suite? | 19:21.29 |
Robin_Watts | yes. | 19:21.34 |
| I have an upload script I need to test :) | 19:21.47 |
mvrhel_laptop | Robin_Watts: the cutdown file in http://bugs.ghostscript.com/show_bug.cgi?id=695863 | 19:23.15 |
Robin_Watts | Ta. | 19:23.34 |
| Hmm. Not sure that testing an upload script with a 66Meg file is a way to preserve my sanity :) | 19:24.33 |
| The non cutdown file was only 10Meg. Can I not add that instead? :) | 19:25.36 |
| mvrhel_laptop: OK, want to give it a whirl? | 19:49.42 |
| If you look on the regression dashboard, there should be a 'Add test files' link. | 19:50.01 |
| Do the obvious and see how it works for you. | 19:50.10 |
mvrhel_laptop | it is a zipped non-cutdown | 19:50.48 |
| I have a smaller one I think. | 19:50.56 |
| hold on | 19:50.58 |
| let me make sure this one has the problem it is only 1000K | 19:52.33 |
| 1m | 19:52.57 |
| ok. yes, this smaller file does the job | 19:56.34 |
| Robin_Watts: | 19:58.27 |
| so the button submit query might need to be changed with respect to its text | 19:58.47 |
| upon pressing the button I got | 19:58.52 |
| bug695863.pdf uploaded as pdf/uploads/bug695863.pdf | 19:58.58 |
| Executing: | 19:58.59 |
| /home/robin/sauce/svnadd/svnadd /var/tmp/CGItemp55920 file://localhost/home/svn-private/ghostpcl/trunk/pdf/uploads/bug695863.pdf | 19:59.01 |
| svn: E000013: Can't open file '/var/tmp/CGItemp55920': Permission denied | 19:59.02 |
| Success! | 19:59.04 |
| not sure if it worked ;) | 19:59.10 |
Robin_Watts | it didn't work. | 19:59.43 |
| Ok, so why can't it read that file... | 20:00.00 |
| Oh, crap, I see. | 20:00.18 |
| just a mo. | 20:00.26 |
mvrhel_laptop | ok | 20:00.29 |
henrys_ | Robin_Watts: marcosw was saying it would be difficult to add multiple files but I think you can use a multiple argument to html's "input" | 20:01.16 |
| Robin_Watts: I could live without that feature though | 20:01.52 |
Robin_Watts | Michael: Try again now ? | 20:02.39 |
mvrhel_laptop | ok | 20:02.42 |
Robin_Watts | henrys_: I am not doing multiple files :) | 20:03.08 |
mvrhel_laptop | ok. that looks like it worked | 20:03.22 |
| bug695863.pdf uploaded as pdf/uploads/bug695863.pdf | 20:03.26 |
| Executing: | 20:03.28 |
| /home/robin/sauce/svnadd/svnadd /var/tmp/CGItemp55808 file://localhost/home/svn-private/ghostpcl/trunk/pdf/uploads/bug695863.pdf | 20:03.29 |
| Adding /var/tmp/CGItemp55808 Committed revision 3437. | 20:03.31 |
| Success! | 20:03.32 |
Robin_Watts | excellent. | 20:03.38 |
mvrhel_laptop | very cool | 20:03.46 |
Robin_Watts | Now we just need to add the 'uploads' directories to the clusters testing. | 20:03.52 |
mvrhel_laptop | thanks for setting that up Robin_Watts | 20:03.53 |
Robin_Watts | no worries. | 20:03.58 |
| henrys_: If you want to give it a whirl, please do. It should put .pcl and .xl files into the appropriate directory. | 20:04.25 |
| henrys_: Should I be looking for other extensions for pcl too? | 20:04.38 |
| like .bin maybe? | 20:04.43 |
henrys_ | Robin_Watts: I wonder if the user should pick a file type from a menu? | 20:06.05 |
Robin_Watts | henrys_: Ahem. | 20:06.12 |
| By which I mean "you mean like the menu on the upload page?" | 20:07.28 |
henrys_ | Robin_Watts: sorry just got in from my run, didn't look at it first | 20:07.43 |
Robin_Watts | Ah :) | 20:07.52 |
| It can either try to guess from the extension or you can select one. | 20:08.19 |
| If there are 'common' extensions I should add them. | 20:08.31 |
| Ok, the cluster should include those on cluster runs now. | 20:09.24 |
henrys_ | Robin_Watts: I assume it won't overwrite a file of the same name? | 20:09.27 |
Robin_Watts | henrys: yeah, svn barfs if you try to do that. | 20:09.42 |
henrys_ | Robin_Watts: I get Success for my file. | 20:09.43 |
| Robin_Watts: nice I hope folks will start adding more tests now that it's easy. | 20:11.14 |
Robin_Watts | I can see that a file was added. I can't tell where. | 20:11.35 |
| My svn foo is not strong. | 20:11.40 |
henrys_ | Robin_Watts: what's the first arg to svn add you have? | 20:12.38 |
Robin_Watts | svn import -m "Web upload of test file" <tmpfilename> <url_of_repo>/trunk/<type>/uploads/<filename> | 20:13.35 |
henrys_ | Robin_Watts: oh I'm reading the feedback after Executing: and before success. | 20:14.31 |
| svnadd /var/tmp etc. | 20:14.49 |
Robin_Watts | henrys_: Yeah, I have to use a setuid exe of my own. | 20:14.55 |
| svnadd does "svn import -m "Web upload of test file" ...." | 20:15.17 |
| I have to be a member of gs-priv to be able to write to svn. | 20:15.43 |
| but conversely I have to be www-data to get access to the tmpfile. Lots of fun :( | 20:16.10 |
| What's the URL you are seeing? | 20:16.40 |
henrys_ | 1file://localhost/home/svn-private/ghostpcl/trunk/pcl/uploads/2bars.pcl | 20:18.56 |
Robin_Watts | perfect. | 20:19.03 |
| ok, I think maybe my path packing commit should work now. | 20:21.27 |
mvrhel_laptop | ok. I need to get this printing going on another thread. The xps writing is blocking the ui when it occurs. | 22:44.54 |
| not the gs creation that is already on another thread. the xps printing is the problem. | 22:45.32 |
| they have a writeasync command but that throws an exception | 22:45.49 |
| but the write command is fine | 22:45.54 |
| what is weird is the write command will do the callback on the write async functions | 22:46.18 |
| of course, this thread needs to be a special type. doing a background thread will not work | 23:47.39 |
| will finish tomorrow or tonight | 23:47.50 |
| done for now | 23:47.53 |
| Forward 1 day (to 2015/03/14)>>> | |