| | <<<Back 1 day (to 2015/12/27) | 20151228 |
| Robin_Watts | chrisl: For when you get back... building gs on casper, we hit problems with xcb_get_setup being missing. | 15:45.32 |
| | marcosw: You here? | 16:38.14 |
| marcosw | Robin_Watts: I am now | 17:52.33 |
| Robin_Watts | marcosw: Hi | 17:53.33 |
| | So, I was going to ask what the results of that graphics alpha bits test run was, but I think I found the email and the answer was "201 new SEGVs". | 17:54.08 |
| | I've been looking at some fuzzing bugs today, and some of them seem to have gone away. | 17:54.34 |
| marcosw | yup. and a bunch of bitmap regressions. I can send you those if you want :-) | 17:54.39 |
| Robin_Watts | Is it hard to rerun? | 17:54.44 |
| marcosw | no. just takes ~12 hours. | 17:54.56 |
| Robin_Watts | the fuzzing stuff takes just 12 hours? | 17:55.05 |
| marcosw | sorry, thought you meant the alphabits stuff | 17:55.22 |
| | the fuzzing takes days, but I won't rerun it. I'll just run the files that have open bugs. | 17:56.11 |
| Robin_Watts | I need to look into it some more before I can ask for the graphicsbits test to be rerun sensibly. | 17:56.33 |
| marcosw | wow, there are only 10 open fuzzing bugs other than the ones I opened last month. | 17:58.51 |
| CrazyTB | Hello, where can I submit a feature request or a feedback regarding the Android version of MuPDF? | 18:02.22 |
| Robin_Watts | CrazyTB: You can ask here, and if we're around, we might be able to offer comments immediately. | 18:03.13 |
| | Or you can open an enhancement request on bugs.ghostscript.com | 18:03.27 |
| CrazyTB | Okay then. I assumed people were on holidays and not on IRC. :) | 18:03.35 |
| marcosw | Robin_Watts: if you have time could you look at <http://bugs.ghostscript.com/show_bug.cgi?id=695822>? FIxing it would make the weekly regression testing less painful. | 18:03.37 |
| CrazyTB | Several buttons on Android UI have the feature of showing a tooltip (with the button name) upon long-pressing. | 18:04.00 |
| | I wish the same would be true for MuPDF. For instance, I don't know what the chain icon means. | 18:04.19 |
| Robin_Watts | The Chain icon turns on whether link following is enabled or not. | 18:04.40 |
| | CrazyTB: If you wanted to submit a patch, we'd consider it. | 18:04.57 |
| marcosw | it might be a c-list issue, but showed up as a regression with the temp file cleanup commit that Paul did in 2014. | 18:04.59 |
| Robin_Watts | marcosw: I'll have a quick look, sure. | 18:06.24 |
| CrazyTB | Robin_Watts: given I have zero experience with Android programming, I don't think I'm going to be the one writing it. | 18:06.36 |
| Robin_Watts | CrazyTB: open an enhancement request on bugs.ghostscript.com and it won't be forgotten. | 18:07.14 |
| marcosw | thx. | 18:07.33 |
| Robin_Watts | marcosw: Could you rerun the fuzzing files that I can't reproduce (or maybe all the outstanding ones?) please? | 18:11.02 |
| | marcosw: I get an exit value of 0. | 18:20.01 |
| marcosw | Robin_Watts: sure | 18:20.25 |
| Robin_Watts | (release build, just trying a debug build now) | 18:20.47 |
| marcosw | you mean for 695822? | 18:21.00 |
| Robin_Watts | Yes. | 18:21.08 |
| marcosw | shouldn't require the debug build. | 18:21.52 |
| | Robin_Watts: hmm, I get an exit value of 0 as well. but there are still many errors in the weekly regression test results. let me check. | 18:23.01 |
| CrazyTB | Robin_Watts: Submitted it as http://bugs.ghostscript.com/show_bug.cgi?id=696485 , thanks! | 18:25.11 |
| marcosw | right, the problem is indeterministic. The recent test run had 59 errors out of 8835 tests. | 18:25.50 |
| | I'll find you a test case that fails with the current master. | 18:26.24 |
| | Robin_Watts: try this command line: ./head/bin/gs -I./gs/lib -sOutputFile='| lzop -f -o ./temp/tests_private__ps__ps3cet__23-12W.PS.pbmraw.72.0.lzo' -dMaxBitmap=400000000 -sDEVICE=pbmraw --saved-pages-test -r72 -Z: -sDEFAULTPAPERSIZE=letter -dNOPAUSE -dBATCH -K1000000 -dJOBSERVER %rom%Resource/Init/gs_cet.ps - < ./tests_private/ps/ps3cet/23-12W.PS | 18:32.30 |
| | sorry it's so long, i took it directly from the regression test. | 18:32.54 |
| Robin_Watts | yeah, doesn't go wrong for me, but then I can't run that exact line. | 18:35.15 |
| | Are we sure it's not the | lzop stuff that's causing the problem? | 18:36.03 |
| | Is there an example that goes wrong on peeved or peeves ? | 18:36.46 |
| | marcosw: So... the bug report says that we have overlapping source and dest in memcpy, which is a no-no. | 18:40.56 |
| | So I look it up in the source, and it's a memmove. | 18:41.03 |
| | Which is fine to allow overlapping src/dst. | 18:41.20 |
| | and in fact, the example given in the bug has src == dst. | 18:42.17 |
| | so it should be a noop. | 18:42.25 |
| | OK, on peeved I can reproduce the exit == 255 | 18:46.18 |
| mvrhel_laptop | Robin_Watts: I am finally getting back to the clist/alpha bits/transparency bug | 19:12.02 |
| Robin_Watts | mvrhel_laptop: Hi. Oh, fun. | 19:12.15 |
| mvrhel_laptop | I do see it changing the color space to gray for the softmask during the clist reading | 19:13.18 |
| | but I need to see if that is occurring in the band with the issue | 19:13.39 |
| Robin_Watts | mvrhel_laptop: What matters is not that it is changing the color space, but that it is not changing the clist_color_depth. | 19:14.46 |
| mvrhel_laptop | if it changes the color space it certainly should adjust the color depth | 19:15.24 |
| | that would be odd | 19:15.29 |
| | if it is not | 19:15.38 |
| Robin_Watts | (I think it's clist_color_depth that is being used to figure out how many bytes to read, but I might be wrong. Whatever thing it *is* using, it is not changing the depth. | 19:15.39 |
| mvrhel_laptop | ok I will figure it out | 19:15.50 |
| Robin_Watts | I put a data breakpoint on the depth value, and it never changed, IIRC. | 19:16.01 |
| mvrhel_laptop | Robin_Watts: so I believe the issue is that when the copy alpha is occurring and the pdf14 clist device is the target, we are forwarding the command and that is getting things confused. the pdf14clist device has the proper color depth but its target is the "final" device and is not a device that changes its colors | 19:54.07 |
| Robin_Watts | mvrhel_laptop: ok... | 19:54.40 |
| mvrhel_laptop | I may need to add a copy_alpha proc for the pdf14 clist devices | 19:54.45 |
| | to ensure we get this correct | 19:54.53 |
| | I had added the copy alpha for the regular pdf14 devices a while back | 19:55.13 |
| | but did not for see this | 19:55.19 |
| Robin_Watts | That sounds plausible (to my uneducated ear at least) | 19:55.22 |
| mvrhel_laptop | foresee | 19:55.23 |
| | I think this is the issue anyway. I can't see how it could work in its present state | 19:56.07 |
| | ick this is more complex than I realized | 20:10.31 |
| | I am going to need to dig a bit more | 20:25.06 |
| | lunch | 20:25.15 |
| rayjj | finally finished the laptop upgrade from H*ll | 20:38.16 |
| | Since my new laptop came with Windows 10/7 Pro, and my old laptop only had Home, I wanted to use the new OS | 20:39.04 |
| | the problem started with a smaller HD on the new system (500GB) and the old laptop has a 1Tb SSD I wanted to use. | 20:39.54 |
| | So I backed up my old laptop HD. That took several attempts (over days) because my backup 5Tb Seagate was running slow at times, then getting I/O errors. I thought the I/O errors might be the s/w I was using since the drive was only 1 yr old. | 20:43.38 |
| | then it hung, and after rebooting, the drive wouldn't show up at all (seeing if I can recover anything from it will be a separate exercise) | 20:44.38 |
| | Went to the store, bought a new 1Tb backup drive and overnight (by now Xmas eve) got it backed up. Used PCmover to move apps and data as much as possible, leaving out two big folders "Users" and "Artifex" so that it would fit | 20:46.17 |
| | bit the bullet and removed the Windows partition (and the Dell Recovery partition I didn't need any more) from the SSD and tried copying the Lenovo partition and Windows partition over (another overnight process) | 20:48.44 |
| | wouldn't boot, despite fiddling with the "Active" partition and eventually using a tool to "rewrite" the MBR. | 20:49.28 |
| | I used the Lenovo restore (from Restore DVD's I had made -- oh, yeah, that took a while) to "restore" the default factory configuration to the SSD. | 20:51.10 |
| Robin_Watts | rayjj: Attempting to 'move' programs without reinstalling is a dumb thing to do, IMHO. | 20:52.06 |
| | Reinstall apps from scratch, copy data via the network (or from a hard drive attached via USB). | 20:52.49 |
| rayjj | It booted, but now PCmover wouldn't transfer from my backup because my license had already been used to transfer once. Laplink's 24/7 transfer tech support was not available -- spent up to 45 minutes on hold with no update about queue status and a *VERY* irritating 5 second "music" loop. | 20:53.44 |
| Robin_Watts | Crap in the registry is the bane of windows systems. Moving to a fresh install and reinstalling is the only way to avoid it spiralling out of control. | 20:54.09 |
| rayjj | Robin_Watts: I used it last time, and even though I have to update some licenses manually, it was a LOT faster than installing all my apps over again | 20:54.22 |
| | Robin_Watts: particularly since I am in St. Louis and my install disks (for those that weren't downloaded) are in CA | 20:55.44 |
| Robin_Watts | Still sounds like a recipe for long term pain to me. | 20:56.14 |
| rayjj | Laplink tech support email was not available over the weekend, but I decided to copy the partition from the 500Gb HD that had been (mostly) transferred from my original SSD contents and it booted and runs :-) | 20:57.33 |
| | so after one more overnight copy of the two large folders from my backup of the SSD, I am up :-) | 20:58.18 |
| | Robin_Watts: I do agree with the registry "bloat" that happens. I've used a registry cleaner in the past (after making my own backup of the registry) and it helped a bit | 20:59.31 |
| | it cut it in half | 20:59.42 |
| | now, at least I am up and running -- I have to get licenses from Adobe, and I'll see what else. I also have to get used to the new KB | 21:01.32 |
| | hmm... it appears that VS 2005, 2008 and 2015 all complain that a component is missing. I'll report that to Laplink. They handled 2005 last time, not sure if I had 2008 yet. | 21:03.15 |
| | so I am up and limping. Well, at least I cleared my email and can surf the net ;-/ | 21:03.52 |
| | need to reboot... | 21:13.58 |
| | mvrhel: for the logs. With the color space num_components (depth) issue, I was seeing the cist reader getting the colorspace num_components and depth right given the pdf14 compositor commands. | 23:21.11 |
| | mvrhel: the clist *writer*, however was just pushing the compositor command and the devn_encode_color was using the target color_info.num_components which was *not* being modified. | 23:23.11 |
| | mvrhel: ordinarily, line 8316 of base/gdevp14.c sets the clist (writer) color_info.num_components, but the devn_encode_color (target of the clist device) doesn't get this device. | 23:29.17 |
| | Forward 1 day (to 2015/12/29)>>> | |
IRC Logs