| Thursday, July 29th, 2021, 00:08 UTC | ||
| [00:08:30] | peterbennett (peterbennett!~peter@mythtv/developer/peterbennett) has quit (Quit: Leaving.) | |
| [01:00:00] | amessina (amessina!~amessina@user/amessina) has quit (Quit: Konversation terminated!) | |
| [01:53:21] | markspieth (markspieth!~markspiet@mythtv/developer/markspieth) has quit (Ping timeout: 256 seconds) | |
| [01:53:33] | markspieth (markspieth!~markspiet@mythtv/developer/markspieth) has joined #mythtv | |
| [01:53:34] | Mode for #mythtv by ChanServ!ChanServ@services.libera.chat : +v markspieth | |
| [02:30:33] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7d83:6501:6ea1:20d:5872:d872) has joined #mythtv | |
| [02:53:55] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7d83:6501:6ea1:20d:5872:d872) has quit (Quit: Konversation terminated!) | |
| [05:00:34] | Warped (Warped!~Warped@user/warped) has quit (Quit: We Gone! Bye Bye) | |
| [06:09:25] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7d83:6501:6ea1:20d:5872:d872) has joined #mythtv | |
| [06:40:55] | markspieth2 (markspieth2!~markspiet@1.152.108.210) has joined #mythtv | |
| [06:40:55] | markspieth2 (markspieth2!~markspiet@mythtv/developer/markspieth) has joined #mythtv | |
| [06:40:55] | markspieth2 (markspieth2!~markspiet@1.152.108.210) has quit (Changing host) | |
| [06:40:55] | Mode for #mythtv by ChanServ!ChanServ@services.libera.chat : +v markspieth2 | |
| [06:42:55] | markspieth (markspieth!~markspiet@mythtv/developer/markspieth) has quit (Ping timeout: 256 seconds) | |
| [07:10:27] | markspieth (markspieth!~markspiet@202.169.118.74) has joined #mythtv | |
| [07:10:27] | markspieth (markspieth!~markspiet@mythtv/developer/markspieth) has joined #mythtv | |
| [07:10:27] | markspieth (markspieth!~markspiet@202.169.118.74) has quit (Changing host) | |
| [07:10:27] | Mode for #mythtv by ChanServ!ChanServ@services.libera.chat : +v markspieth | |
| [07:53:46] | stuarta: | gtb: your ubuntu-20_04 and centos8 builders appear to be offline |
| [08:00:07] | stuarta: | is this expected? |
| [08:18:37] | DevMythNotifyBot: | COMMIT: stuarta pushed 14 commit(s) to branch devel/http: https://github.com/MythTV/mythtv/compare/e4f2 . . . a3ea45a7c647 |
| [08:25:34] | markspieth (markspieth!~markspiet@202.169.118.74) has joined #mythtv | |
| [08:25:34] | markspieth (markspieth!~markspiet@mythtv/developer/markspieth) has joined #mythtv | |
| [08:25:34] | markspieth (markspieth!~markspiet@202.169.118.74) has quit (Changing host) | |
| [08:25:34] | Mode for #mythtv by ChanServ!ChanServ@services.libera.chat : +v markspieth | |
| [10:36:40] | gtb (gtb!~gtb@2001:470:69fc:105::1e7) has quit (Quit: Bridge terminating on SIGTERM) | |
| [10:39:42] | gtb (gtb!~gtb@2001:470:69fc:105::1e7) has joined #mythtv | |
| [11:38:57] | warpme_ (warpme_!uid391875@id-391875.brockwell.irccloud.com) has joined #mythtv | |
| [11:44:16] | warpme_: | guys: i have multiple reports on segfault with mythtv-master on intel systems. i can reproduce it on 2 systems i have (n3450 and z8500). backtrace from coredump: https://pastebin.com/mfKKqMrx Interesting is that i can't reproduce segfault under gdb or when logging is set to all. I can reproduce only when frontend is launched with minimal logging. maybe somebody can look in this? |
| [11:44:18] | glenb77 (glenb77!~glenb77@2603-8000-b440-c812-0000-0000-0000-0331.res6.spectrum.com) has joined #mythtv | |
| [11:54:07] | stuarta: | curious |
| [11:54:50] | stuarta: | if it can't be reproduced under gdb, then it's likely to be a race |
| [11:57:45] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7d83:6501:6ea1:20d:5872:d872) has quit (Quit: Konversation terminated!) | |
| [11:58:47] | paul-h (paul-h!~paul@pres-25-b2-v4wan-161467-cust2666.vm29.cable.virginm.net) has joined #mythtv | |
| [11:58:47] | paul-h (paul-h!~paul@mythtv/developer/paul-h) has joined #mythtv | |
| [11:58:47] | paul-h (paul-h!~paul@pres-25-b2-v4wan-161467-cust2666.vm29.cable.virginm.net) has quit (Changing host) | |
| [11:58:47] | Mode for #mythtv by ChanServ!ChanServ@services.libera.chat : +v paul-h | |
| [13:07:57] | warpme_: | stuarta: yes. i think so. i'm too weak in myth internals to intelligently add startup selective delays or threads serialisation to find where root issue is. My rough hypothesis about suybsystems is: startup of UI with ongoing theme displaying/caching. 1 per 10 tries fe starts ok and in such case i continues to work well. this tells me issue isn't display/gl related but purely in startup code. another observation: changing |
| [13:07:57] | warpme_: | themecache dir to ro makes things working.... |
| [13:09:37] | peterbennett (peterbennett!~peter@mythtv/developer/peterbennett) has joined #mythtv | |
| [13:09:37] | Mode for #mythtv by ChanServ!ChanServ@services.libera.chat : +v peterbennett | |
| [13:13:26] | peterbennett: | warpme_: The crash – one approach is to find out in which commit it started – if you know a good version bisect with bad version to find the problem commit. |
| [13:39:23] | stuarta: | warpme_: the crash appears to occur underneath libpng, in libz when it's trying to write out an image, so it's not really surprising that making the theme cache ro stops the issue, as it'll stop the write |
| [13:49:57] | peterbennett (peterbennett!~peter@mythtv/developer/peterbennett) has quit (Remote host closed the connection) | |
| [13:50:41] | peterbennett (peterbennett!~peter@mythtv/developer/peterbennett) has joined #mythtv | |
| [13:50:41] | Mode for #mythtv by ChanServ!ChanServ@services.libera.chat : +v peterbennett | |
| [13:51:00] | peterbennett (peterbennett!~peter@mythtv/developer/peterbennett) has quit (Client Quit) | |
| [13:52:23] | warpme_: | peterbennett: indeed. bisecting is idea. unfortunately i don't remember when it starts so it might be painful. interesting is that setting ro rights on themecache allows fe to start ok. Investigating this i see fe starts only when themecache has ro on empty dir. if themecache has files – it make crash |
| [13:52:56] | stuarta: | i emptied my themecache, made no difference |
| [13:58:19] | warpme_: | it looks like i was a bit wrong. it seems is enough to block writtes to themecache dir and fe starts ok. my themecache is on ramdisk. |
| [14:10:47] | paul-h: | having the themecache dir in a ramdisk kind of defeats the point of the themecache \0/ |
| [14:16:27] | warpme_: | paul-h: isn't that themecache has rescaled images to current screen res? so even if storgage has infinite speed – caching saves rescaling at startup?. (note: having ramdisk isn't exactly equal to non-persistence in my case) |
| [14:24:01] | paul-h: | The difference is if you have a persistent themecache stored on disk then you only have to rescale an image once for a theme. If you use a ramdisk then I would have thought you will have to rescale the images each time you load the FE |
| [14:25:09] | paul-h: | It will be most noticeable for image intensive themes that use coverart or posters for background images for example |
| [14:26:51] | paul-h: | We do already have a small in memory theme cache that stores the most used images if that is what you was thinking |
| [14:28:29] | warpme_: | paul-h: indeed – this is true when ramdisk is non-persistent. but when is it persistent then it serves it's role as expected i think. (to be pedantic i would say even with non-persistent ramdisk it will speedup a bit due much much lower access time & much higher read/write speed). |
| [14:30:08] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7d83:6501:6ea1:20d:5872:d872) has joined #mythtv | |
| [14:31:54] | paul-h: | The bt shows a segfault because of a null pointer pointing to a RatingSettings which is a settings screen in MythMusic |
| [14:33:32] | paul-h: | warpme_: the crash is just on startup of the FE? |
| [14:33:54] | peterbennett (peterbennett!~peter@mythtv/developer/peterbennett) has joined #mythtv | |
| [14:33:54] | Mode for #mythtv by ChanServ!ChanServ@services.libera.chat : +v peterbennett | |
| [14:33:59] | paul-h: | Is the bt the same each time? |
| [14:37:20] | paul-h: | My guess is the themecache stuff is just a red herring that is changing the timing of things just enough so the race condition doesn't occur |
| [14:55:18] | peterbennett (peterbennett!~peter@mythtv/developer/peterbennett) has quit (Quit: Leaving.) | |
| [15:26:07] | gtb: | stuarta: re: builders. The host system appears to have hung. I have power cycled the host. Unfortunately there was nothing in the logs to suggest the why, and I do not have KVM access to that system. |
| [15:26:28] | stuarta: | gtb: fair enough, it happens :-/ |
| [15:32:27] | warpme_: | paul-h: yes. i have the same hypothessis. bt is each time different. interesting it happens is only on intel. can't get it on x86 amd nor nvidia also can't get on 9 different armv8 platforms... |
| [15:37:12] | gtb: | stuarts: btw, last evening I wrote a quick Qt app to test the mariadb 10.6 TZ issues (fedora 34 host, mariadb 10.6 from mariadb.org directly). It seemed to do the right thing (report the correct status of the mariadb tz tables). No MythTV involved though. |
| [15:39:07] | stuarta: | gtb: thanks. wonder how we can narrow it down some more |
| [15:43:03] | gtb: | stuarta: re: builder. I don't like not knowing, but it is a 12+ year old system, and I am going to have to replace it anyway if I want to run EL9 VMs as it predates x86_64 v2 support. |
| [15:44:28] | stuarta: | i'd love to buy myself a server with which to run and test a lot of stuff |
| [15:45:43] | gtb: | stuarta: Re: TZ. Here is the quick and dirty test program I used: https://paste.centos.org/view/56d9e5a8 |
| [15:46:13] | gtb: | stuarta: re: TZ. What platform did you test on? arch? |
| [15:46:22] | stuarta: | yes, i have an arch vm |
| [15:47:36] | stuarta: | i'm going to build that test program here, can you share the .pro file too? save me having to whip one up |
| [15:47:58] | gtb: | Ok, I'll spin one up at some point. |
| [15:48:11] | stuarta: | if you don't have one handy, then i'll just roll one up |
| [15:49:09] | gtb: | It does not take much (my) time to spin up a VM, it is just arch was never one of the distros I spent much time worrying about. |
| [15:49:23] | stuarta: | nor me, it's only good as a canary for future breakages |
| [15:50:11] | gtb: | I did not use a pro file (I said quick and dirty). Manual compile: c++ -I /usr/include/qt5 -I /usr/include/qt5/QtGui -I /usr/include/qt5/QtSql -l Qt5Gui -l Qt5Core -l Qt5Sql test.cpp |
| [15:50:56] | stuarta: | pro file contains |
| [15:51:01] | stuarta: | QT += core gui sql |
| [15:51:04] | stuarta: | SOURCES += tzcheck.cpp |
| [15:51:06] | stuarta: | that's it |
| [15:52:09] | stuarta: | interesting, that fails here, returns "null value" ie. not loaded, yet they are |
| [15:52:41] | stuarta: | therefore this is a simplified "test case" app |
| [15:53:02] | gtb: | So (conjecture time) something about arch? Or its mariadb package? |
| [15:53:20] | ** stuarta shrugs ** | |
| [15:53:30] | stuarta: | could be either, or both |
| [15:53:45] | stuarta: | i don't think any other platform is on mariadb 10.6 |
| [15:54:14] | stuarta: | no, i don't see *any* in the smolt data |
| [15:54:35] | stuarta: | 10.[012345] and then the various mysql versions |
| [15:54:47] | ajshell1 (ajshell1!~ajshell1@50.234.215.82) has joined #mythtv | |
| [15:56:12] | gtb: | Since 10.6.3 "just" became GA earlier this month (the previous versions were considered early access), I suspect it will take a bit of time for it to make it elsewhere. |
| [16:00:36] | stuarta: | hence the arch canary |
| [16:06:47] | gtb: | As long as the canaries understand their role (and responsibility) is to experience pain and death...... |
| [16:07:21] | stuarta: | L-) |
| [16:07:25] | stuarta: | :-) |
| [16:07:39] | stuarta: | this is why nobody sensible runs arch |
| [16:10:55] | gtb: | Well, I run Fedora for personal use, which is [b]leading edge (with bandages available), but arch, being truer bleeding edge, tends to require tourniquets to save the body, but sacrificing entire appendages, and I sometimes want to actually get things done. |
| [16:12:46] | gtb: | But I have to say, I love arch's user documentation. Every other distro should want to produce docs that useful/great. |
| [16:13:06] | stuarta: | agree on both counts |
| [16:13:11] | ** stuarta wanders off ** | |
| [16:13:18] | hampton: | gtb: lol (re: canary) |
| [16:17:47] | peterbennett (peterbennett!~peter@mythtv/developer/peterbennett) has joined #mythtv | |
| [16:17:47] | Mode for #mythtv by ChanServ!ChanServ@services.libera.chat : +v peterbennett | |
| [16:18:36] | ajshell1 (ajshell1!~ajshell1@50.234.215.82) has quit (Quit: Connection closed) | |
| [16:26:50] | ajshell1 (ajshell1!~ajshell1@50.234.215.82) has joined #mythtv | |
| [16:38:25] | warpme_: | ok. i found it. it was cec init. with disabled: 25 consecutive starts all success :-) |
| [16:40:28] | peterbennett: | warpme_: Do you have an Intel system with cec support? Or is it failing even if there is no cec support? |
| [16:46:49] | warpme_: | peterbennett: oh no. i don't have any intel box with cec. afaik there is not so many x86 systems with cec. i had it enabled by way how i adding new frontend (have script automating pxe boot setup + populating user DB settings). As 11 from 15 platforms i have have working cec – my defaults are with cec on. unfortunately cec "on" on hw without cec seems to break some intel boxes.... |
| [17:07:41] | peterbennett: | warpme_: OK, I wonder when this started. Is it ok with older versions? |
| [17:11:16] | peterbennett: | I see cec is enabled, probably by default, on my laptop |
| [17:12:55] | peterbennett: | during startup of mythfrontend there are messages "CECAdapter: Failed to find any CEC devices." and "CECAdapter: Closing down CEC." |
| [17:13:22] | peterbennett: | Do you see there messages before it crashes? |
| [17:14:34] | peterbennett: | I do not see any crashes, but my master build is a bit out of date |
| [17:15:21] | ** peterbennett Wanders away ** | |
| [17:15:31] | peterbennett (peterbennett!~peter@mythtv/developer/peterbennett) has quit (Quit: Leaving.) | |
| [17:38:12] | gtb: | stuarta: Looks like something is causing my armv7 builder to not reboot [I was updating the kernel to the latest]. I am suspicious of a recent bcm.... firmware update that got applied a few days ago. I will work it later today. |
| [18:07:40] | stuarta: | gtb: cool thanks |
| [18:13:23] | DevMythNotifyBot: | COMMIT: linuxdude42 pushed 1 commit(s) to branch fixes/31: https://github.com/MythTV/mythtv/compare/7037 . . . d59053b747b6 |
| [18:24:51] | hampton: | Interesting that 1/3 of the time building the "Videos" dialog is spent in the naturalCompare function. I like the ordering, but I'm not sure I like it that much. |
| [18:26:54] | hampton: | Almost 4.5% of the total time is spent just in the QChar::isPunct() function, with all of the calls from naturalCompare. |
| [18:26:58] | hampton: | :-( |
| [18:29:05] | DevMythNotifyBot: | COMMIT: linuxdude42 pushed 1 commit(s) to branch master: https://github.com/MythTV/mythtv/compare/8a75 . . . 4d899b94bfd2 |
| [19:07:08] | Warped (Warped!~Warped@user/warped) has joined #mythtv | |
| [19:07:10] | gtb: | hampton: I seem to recall that Qt added some natural sort capability at some point. .... found the ref: https://doc.qt.io/qt-5/qcollator.html and https://forum.qt.io/topic/68910/natural-sort-using-qcollator which might be faster/slower than the projects implementation (only testing would tell). |
| [19:23:47] | DevMythNotifyBot (DevMythNotifyBot!~nodebot@cust32-dsl91-135-4.idnet.net) has quit (Ping timeout: 268 seconds) | |
| [19:23:53] | DevMythNotifyBot (DevMythNotifyBot!~nodebot@cust32-dsl91-135-4.idnet.net) has joined #mythtv | |
| [19:25:09] | MythBuild: | Build [#149](https://code.mythtv.org/buildbot/#builders/140/builds/149) of `31-freebsd12–64bit` 3completed successfully. |
| [19:50:54] | hampton: | That could possibly help, but I don't think it would handle sorting "season 2" < "season 10", which appears to be the reason for naturalCompare. It does have the idea of creating sort keys and comparing those keys repeatedly, which is what I was considering trying to implement for naturalCompare. |
| [19:54:22] | peterbennett (peterbennett!~peter@mythtv/developer/peterbennett) has joined #mythtv | |
| [19:54:22] | Mode for #mythtv by ChanServ!ChanServ@services.libera.chat : +v peterbennett | |
| [19:55:45] | hampton: | The algorithms appear to have some sort of combinatorial expansion issue, where 11,471 videos turns into 241,090 calls to naturalCompare, which results in 16,385,104 calls to QChar::isPunct, 14,681,948 calls to QChar::isSpace, etc, etc. Finding a way to parse the 11K strings once, and then have a simpler routine to compare keys should save lots of time. |
| [19:56:04] | hampton: | I just have to grok the nuances of naturalCompare so I can write said functions. |
| [20:03:28] | gtb: | Re: Season 2 < Season 10. As I understand it, that is what coll.setNumericMode(true) is supposed to do, but that is based on the docs, not actually testing it. |
| [20:04:13] | hampton: | ah, missed that when I scanned the docs. Thanks. |
| [20:04:27] | gtb: | In any case, anything you do to eliminate millions of calls to those tests will have to be an improvement (or convince people to save far fewer videos?) |
| [20:06:12] | hampton: | I'm not sure how that 11K number got so big. :-D |
| [20:07:50] | hampton: | The "Videos" page currently takes 5 seconds to load on my Intel NUC6 front end. The "Gallery" page takes far longer. I'm wondering (hoping) its the same issue. |
| [20:08:33] | gtb: | btw, the docs also do mention that getting the sortKey rather using compare directly is faster with a lot of strings, which sounds like what you were talking about (parse once). |
| [20:10:05] | ajshell1 (ajshell1!~ajshell1@50.234.215.82) has quit (Quit: Connection closed) | |
| [20:10:06] | hampton: | agreed. |
| [20:14:13] | hampton: | I'm definitely going to try replacing naturalSort with QCollator. I'd rather use a library than custom code if it does what we need. |
| [20:15:42] | peterbennett (peterbennett!~peter@mythtv/developer/peterbennett) has quit (Quit: Leaving.) | |
| [20:25:44] | peterbennett (peterbennett!~peter@mythtv/developer/peterbennett) has joined #mythtv | |
| [20:25:44] | Mode for #mythtv by ChanServ!ChanServ@services.libera.chat : +v peterbennett | |
| [21:25:51] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7d83:6501:6ea1:20d:5872:d872) has quit (Quit: Konversation terminated!) | |
| [21:44:58] | paul-h (paul-h!~paul@mythtv/developer/paul-h) has quit (Quit: Konversation terminated!) | |
| [21:47:13] | peterbennett (peterbennett!~peter@mythtv/developer/peterbennett) has quit (Quit: Leaving.) | |
| [22:32:33] | markspieth2 (markspieth2!~markspiet@mythtv/developer/markspieth) has joined #mythtv | |
| [22:32:34] | Mode for #mythtv by ChanServ!ChanServ@services.libera.chat : +v markspieth2 | |
| [22:51:31] | markspieth2 (markspieth2!~markspiet@mythtv/developer/markspieth) has quit (Read error: Connection reset by peer) | |
| [22:51:39] | markspieth (markspieth!~markspiet@mythtv/developer/markspieth) has joined #mythtv | |
| [22:51:40] | Mode for #mythtv by ChanServ!ChanServ@services.libera.chat : +v markspieth | |
| [22:53:23] | glenb77 (glenb77!~glenb77@2603-8000-b440-c812-0000-0000-0000-0331.res6.spectrum.com) has quit (Quit: glenb77) | |
| [23:35:26] | amessina (amessina!~amessina@user/amessina) has joined #mythtv | |
| [23:37:19] | amessina (amessina!~amessina@user/amessina) has quit (Read error: Connection reset by peer) | |
| [23:37:55] | amessina (amessina!~amessina@user/amessina) has joined #mythtv | |
| [23:55:30] | peterbennett (peterbennett!~peter@mythtv/developer/peterbennett) has joined #mythtv | |
| [23:55:30] | Mode for #mythtv by ChanServ!ChanServ@services.libera.chat : +v peterbennett | |
IRC Logs collected by
BeirdoBot.
Please use the above link to report any bugs.