Previous day | Jump to hour: 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | Next day

Seconds: Show Hide | Joins: Show Hide | View raw
Font: Serif Sans-Serif Monospace | Size: Small Medium Large

Click in the nick column to highlight everything a person has said.
The Logo icon identifies that the person is a core developer (has commit access).

Notice: Only Gecko based browsers prior to FF4 support the multipart/mixed "server push" method used by this log reader to auto-update. Since you do not appear to use such a browser, this page will simply show the current log, and not automatically update.

#rockbox log for 2019-07-20

00:27:20 Quit lebellium (Quit: Leaving)
00:36:20 Quit ZincAlloy (Quit: Leaving.)
00:36:21***Saving seen data "./dancer.seen"
00:36:57foolsh__builtin: Well wolf3d and duke3d work very well on fuze+, even sound, I'll give quake a test run
01:34:32__builtinGood to know
01:34:47__builtinFor quake make sure you get the latest patch set
01:35:32foolshIt's weird, cherry picking Quake into my local branch causes lots of warnings touching SDL wold3d duke3d and of course quake
01:35:49__builtinDon't cherry pick it alone
01:36:13__builtinIt relies on its parent commits
01:36:30__builtinBest bet is probably to checkout
01:36:35foolshah well of course
01:37:13__builtinActually... don't checkout
01:37:33__builtinCherry pick g#1794 and then g#1832
01:37:36fs-bluebotGerrit review #1794 at : Add proper float formatting to vuprintf by Michael Sevakis
01:37:36fs-bluebotGerrit review #1832 at : Quake! by Franklin Wei
01:37:56__builtinIt won't compile if you pull in g#2105
01:37:57fs-bluebotGerrit review #2105 at : system-arm: log PC address via ipod6g piezo on fault (HACK) by Franklin Wei
01:38:03foolshYeah I thought as much when I looked at the related changes
02:00:01__builtinok, quake seems to consistently run at 8.6FPS
02:00:06__builtinlet's see if I can optimize that
02:36:22***No seen item changed, no save performed.
03:09:01 Join krabador [0] (~krabador@unaffiliated/krabador)
03:09:18foolsh__builtin: Quake on fuze+ compiles with a shit metric ton of warnings but no errors, starts normally, exits at first screen with message "Error: Draw_Pic: bad coordinates" then goes to a screen that says "Plugin returned error #leak-file-handles"
03:09:26foolshI'll get you more to go on later
03:10:25__builtinsounds like an issue with the screen size
03:11:53__builtinif you want a quick but hacky fix, go to vid_sdl.c and set BASEWIDTH=320 and BASEHEIGHT=200
03:11:58__builtinSDL will scale it automatically
03:12:10__builtin(it'll be slow as all heavens, of course, but it ought to run)
03:30:47foolsh__builtin: actually it runs at a decent speed on fuze+, crashed once something about "drawing plane" and buttons are buggy it's stuck in auto fire mode, wasted all my ammo right at the beginning :) but how ever over all impressive feat for a DAP
03:31:56__builtinif you press the fire button a few times it should get unstuck
03:40:22 Join massiveH [0] (
03:41:26foolshWas thinking for wold3d and duke3d SOUND_ENABLE could be #ifdef'ed out for targets with low ram but with color lcds, instead of just omitting them
03:44:05fs-bluebotBuild Server message: New build round started. Revision 3e2b50e, 280 builds, 9 clients.
03:44:40__builtinfoolsh: sure, for wolf3d
03:44:58__builtinWhat would the threshold be for sound?
03:47:11 Quit krabador (Remote host closed the connection)
04:36:23***Saving seen data "./dancer.seen"
04:37:54fs-bluebotBuild Server message: New build round started. Revision 5d05b9d, 280 builds, 9 clients.
04:54:49foolsh__builtin: I can spend sometime decreasing the allocated ram on my fuze+ til I hit the limit
04:55:04__builtinSounds good
04:55:18__builtinit's going to be a very rough estimate no matter what
05:13:25 Quit TheSeven (Disconnected by services)
05:13:34 Join [7] [0] (~quassel@rockbox/developer/TheSeven)
05:23:48fs-bluebotBuild Server message: New build round started. Revision b0e9bb1, 280 builds, 9 clients.
05:24:47__builtinthat ought to mitigate any worries about core memory usage
05:51:18fs-bluebotBuild Server message: New build round started. Revision 9c17734, 280 builds, 10 clients.
06:36:27***Saving seen data "./dancer.seen"
07:17:22 Quit massiveH (Quit: Leaving)
08:25:41 Join dys [0] (
08:31:51 Quit [7] (Ping timeout: 264 seconds)
08:35:16 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
08:36:31***Saving seen data "./dancer.seen"
09:54:57 Join lebellium [0] (
10:01:05 Join ZincAlloy [0] (~Adium@2a02:8108:9440:dfc:54c1:7e76:1ad8:2d7d)
10:29:25 Quit Huntereb (Ping timeout: 258 seconds)
10:36:32***Saving seen data "./dancer.seen"
10:53:23 Join Huntereb [0] (
11:41:24 Quit paulk-leonov (Ping timeout: 268 seconds)
11:43:44 Join paulk-leonov [0] (
11:49:09fs-bluebotBuild Server message: New build round started. Revision 55eb1c5, 280 builds, 10 clients.
12:12:37fs-bluebotBuild Server message: Build round completed after 652 seconds.
12:12:38fs-bluebotBuild Server message: Revision 6392e05 result: 2500 errors 0 warnings
12:21:59amdjWho do we fire foam missiles at, for breaking the build?
12:36:33***Saving seen data "./dancer.seen"
12:41:53fs-bluebotBuild Server message: New build round started. Revision 14d91ca, 280 builds, 11 clients.
12:52:20 Quit ZincAlloy (Quit: Leaving.)
13:02:47Bilgusit depends whens the last time you committed?
13:03:33Bilgusspeachy, but technically its my fault for pushing it
13:04:28gevaertsSome of the people who broke the build once haven't been seen in years
13:06:31*gevaerts doesn't know if there's a causal link
13:09:06Bilgusso the issue is genlang isn't being run before the apps / plugins are built therefore lang_enum.h is missing
13:09:52BilgusI tried moving genlang up in the list in root.make but its still not happening at the right time
13:11:34Bilgushmm mentions lang.h so maybe I can co-opt it
13:16:22 Quit dys (Ping timeout: 268 seconds)
14:15:22 Join ZincAlloy [0] (~Adium@2a02:8108:9440:dfc:7896:baad:ba3c:acfd)
14:36:37***Saving seen data "./dancer.seen"
15:13:34 Join dys [0] (
15:32:52 Join speachy [0] (d102414d@
15:33:29speachyWell, time to figure out why Make isn't generating proper dependencies for the language stuff. :/
15:33:45Bilgusuh yeah I just hacked a fix to get it to build lol
15:34:06Bilgusits not getting generated till half way through the plugins
15:35:08Bilgusin lang.make DUMMY := $(shell touch $(BUILDDIR)/lang_enum.h) lol
15:37:01speachyJust added lang_enum.h to the clean target too
15:37:20speachyI want to say I ran into this exact problem a while back and fixed it.
15:37:47speachythe dependency list doesn't have any reference to lang_enum.h, which is .. wrong IMO.
15:40:14Bilgusmake is not my favorite language
15:42:00 Quit speachy (Remote host closed the connection)
15:42:48 Join speachy [0] (d102414d@
15:44:45speachyI think I know what's going on..
15:45:08speachywhen the dependencies are generated, PLUGIN is not defined.
15:45:45speachylet's see..
15:50:28speachybetter, but the puzzles plugin remains pathological
15:52:16speachybecause it explicitly includes rbcompay.h on the cmdline..
15:56:07speachyso mkdepfile won't pick it up
16:01:36speachyfound the problem. testing..
16:03:00Bilguswell you certainly got it faster than I
16:06:42fs-bluebotGerrit review #2121 at : build: Fix puzzles build due to missing dependency by Solomon Peachy
16:06:55*speachy loves make
16:07:27speachyit's nothing more than a dependency resolver, so if something doesn't work, it's because the deps aren't specified correctly.
16:08:47fs-bluebotBuild Server message: New build round started. Revision 3cb272c, 280 builds, 11 clients.
16:09:11 Quit michaelni (Ping timeout: 245 seconds)
16:09:13speachyokay, let's see how much red is left after this..
16:09:16gevaertsThe issue tends not to be really make but how messy the problem is
16:09:48speachythat's why everything that seeks to replace make ends up being worse, because the problem isn't with make itself. :)
16:10:01*gevaerts nods
16:11:43*speachy hears his build server's fans spin up
16:13:00*gevaerts fondly remembers the time when he had a build client on his phone :)
16:13:07speachytime to go start cooking brunch. from-scratch biscuits & gravy.
16:22:43 Join michaelni [0] (
16:36:39***Saving seen data "./dancer.seen"
16:39:40 Quit pixelma (Quit: .)
16:39:40 Quit amiconn (Quit: - Chat comfortably. Anywhere.)
16:40:43 Join amiconn [0] (jens@rockbox/developer/amiconn)
16:40:43 Join pixelma [0] (marianne@rockbox/staff/pixelma)
16:41:07speachya lot less red, yay.
16:42:25*speachy will fix the rest after the guests leave.
16:58:15*gevaerts is annoyed by that sim snprintf warning
17:02:39 Quit foolsh (Ping timeout: 272 seconds)
17:12:24fs-bluebotBuild Server message: New build round started. Revision 3665eec, 280 builds, 11 clients.
17:25:06 Quit Huntereb (Quit: See ya!)
17:26:58 Join Huntereb [0] (
17:35:11 Quit Huntereb (Quit: See ya!)
17:38:32 Join Huntereb [0] (
17:41:00 Quit noobineer (Remote host closed the connection)
18:36:41***Saving seen data "./dancer.seen"
19:01:45fs-bluebotBuild Server message: New build round started. Revision d1b8b39, 280 builds, 11 clients.
19:03:21speachyokay, it broke on non-RTC targets. that's fixed −− the strings needed to be moved from rtc-only context to global. English fixed, but I don't know if there's an automagic way to fix up the other languages too.
19:03:31speachyreally don't want to do it by hand.
19:05:35__builtinspeachy: what are you trying to do?
19:07:27__builtinis it just what you did in HEAD, but for all languages?
19:07:48__builtinlooks like an awk script should be able to do that
19:09:44gevaertsThere's tools/ that might do that, but I'm not sure
19:11:11speachyyes, waht I did in head.
19:11:46speachyfor about 18-odd strings, make the rtc: target the * target.
19:12:06*__builtin is working on it
19:12:11__builtinseems simple enough
19:23:06__builtinspeachy: is this what you want?
19:24:50speachyit looks like you updated the entire rtc: subset rather than just the stuff I changed to get the build fixed.
19:25:46__builtinah, just those?
19:26:09speachythough it's possible I may have missed some.
19:26:23__builtinwhat are the IDs you want changed?
19:27:54speachyso 19 in all.
19:29:41speachyenough to be able to voice the date and time, and configure the time format.
19:33:27speachylooks good
19:33:35speachyonly scanned the first (cz)
19:35:24speachythe errors went to warnings. which may be resolved by the update you're cooking
19:36:43speachybut the fm rec went green too, but player went red. wonder if that's a builder quirk or something.
19:36:57fs-bluebotGerrit review #2123 at : Move date/time strings out of RTC context by Franklin Wei
19:38:23speachylooks sane to me
19:40:35fs-bluebotBuild Server message: New build round started. Revision ee659d4, 280 builds, 11 clients.
19:40:59speachyand the fans spin back up...
19:42:09*__builtin realized that his buildclient was running as root
19:42:19__builtinso no sparta this time
19:44:06 Join petur [0] (~petur@rockbox/developer/petur)
19:44:08speachyhmm. I think I'll set up another build client
19:45:52*__builtin is working on a docker image to containerize the whole thing
19:46:00__builtinit's a real pita getting a dev environment set up
19:46:47speachythe only annoyance is building the toolchains
19:47:51__builtinonce I get this working, it should be a matter of 1 or two shell commands
19:48:54__builtinright now the mipsel build is failing because of a conflict installing documentation
19:52:10speachywould be really nice to tie the builders into gerrit so all patches can get build-tested before committing.
19:54:20__builtinI had that idea many moons ago... got shot down for one reason or another
19:54:37__builtinfirst thing that comes to mind is arbitrary code execution by anyone who pushes a gerrit patch
19:55:47speachytrue that.
19:56:32__builtinI guess it could be made developers-only
19:56:42__builtinbut even then it'd be hard to actually implement
19:57:01__builtin(and would require action by the Swedes, who hvae largely removed themselves from the project)
19:58:16speachygerrit rules could help a lot with that distinction. only blessed users could get stuff autobuilt (or trigger an autobuild after a review)
19:58:42__builtinyeah, something like that
19:58:44speachybut yeah, with the current development velocity it's hard to justify that effort
19:59:19__builtinwe don't lose that much from fixing the things after they're pushed
20:00:28speachyall green except for the player targets!
20:01:03__builtincrap, I didn't actually kill my client
20:02:36speachyI'm going to commit g#2073 and g#2074 once I get the second builder going.
20:02:39fs-bluebotGerrit review #2073 at : FS #7912: Voice enable the chessbox plugin by Solomon Peachy
20:02:39fs-bluebotGerrit review #2074 at : FS #7705: Voice enable battery bench plugin by Solomon Peachy
20:12:49speachyokay, now to do a manual build to make sure it's sane..
20:14:20fs-bluebotBuild Server message: New build round started. Revision 4edaf3f, 280 builds, 11 clients.
20:14:54speachyokay, two builders going now
20:15:00 Quit michaelni (Ping timeout: 245 seconds)
20:16:51__builtinhmm... who/what builds the latex manuals?
20:17:30gevaertsMy clients do, although there seems to be something wrong with the html one
20:17:47gevaertsAs in, the build succeeds, but uploading the log fails with http 500
20:19:01__builtinwhich packages do I need for that?
20:19:35gevaertsSeveral :)
20:20:30__builtinmight be best to stay away then
20:20:33gevaertsI don't know how current is
20:20:39gevaertsNo, go for it!
20:20:56gevaertsIt's one of the things where we have very little redundancy
20:21:30speachyI'll get my builders set up for the manuals too
20:22:06speachyblows up for me with 'multirow.sty' not found
20:22:43speachy(fedora, texlive installed and working for other things)
20:23:04gevaertson debian stretch, that's in texlive-latex-extra
20:24:40speachyfedora it's texlive-multirow
20:25:57__builtinalright, I've got a docker image for sh, m68k and arm going
20:26:01__builtinand sdl/latex
20:26:14gevaertsStill missing mips and android then :)
20:26:27__builtinyeah, those are annoying
20:26:45speachyboth of mine have mips-native and mips-hosted
20:27:03speachydon't recall those being particualrly difficult since the move to gcc 4.9
20:27:47__builtinthey seem to conflict with other toolchains
20:27:54gevaertsoh, and application arm
20:30:59 Quit speachy (Remote host closed the connection)
20:31:26 Join speachy [0] (d102414d@
20:36:44***Saving seen data "./dancer.seen"
20:37:21speachyon the next build my builders will do the latex manuals
20:39:12 Join michaelni [0] (
20:40:40gevaertsGreat :)
20:41:18speachythe archos builds are being flaky. from one to the next it's inconsistent.
20:44:44fs-bluebotBuild Server message: New build round started. Revision 78892b5, 280 builds, 11 clients.
20:45:23Bilgusspeachy you dun blowed up the rockbox!!
20:46:37speachystrings. :/
20:46:38Bilgus:P its good for 'everyone' to get a little hands on there was a couple months there that nothing even got committed
20:47:36speachythough looking at the recent commits, Quake added 3-4k by itself. :)
20:48:35gevaertsWe had years of debate on whether adding doom was a mistake, and then __builtin came along :)
20:48:51__builtinspeachy: to the core?
20:49:25__builtinThe increase is limited to quake targets which already have massive ram
20:49:45speachyif one believes the deltas on the dev builds page
20:50:20speachyeven clip+ etc saw a 4.4k bump, and there's no way that thing runs quake..
20:51:11Bilgusspeaking of I fixed all those new warnings expept 1 the rtc one as wodz said it should be sanitized properly and I've no clue what properly is
20:51:26Bilgusone of you wanna take a hack at it
20:51:40speachyah, I saw the subsquent build shrunk. the printf float thing.
20:52:30gevaertsYes, but that's still 4K to core for a plugin that I think is of fairly limited appeal
20:52:31Bilgus#g1994 way my try
20:52:42Bilgusahem was my try
20:53:48__builtinI should be able to move it into the plugin
20:53:52BilgusIDK geaverys I've seen a bunch of people very interested in those damn games lol
20:54:16BilgusI must be drunk typing
20:54:25gevaertsBilgus: yes, those people are loud :)
20:54:56speachyeh, the only game anyone needs is snake!
20:55:21Bilgusand the damn bullet proof nokia to run it on
20:55:40BilgusSo I thake it no takers on g#1994?
20:55:42fs-bluebotGerrit review #1994 at : Fix Truncation warning screens.c->set_time_screen by William Wilgus
20:55:55gevaertsAnyway, I don't know how important 4K still is on devices that are relevant today
20:56:40gevaertsBilgus: oh, I didn't see that you were working on that...
20:57:11gevaertsI did 3665eecf earlier today to fix that, with a different approach
20:57:19speachy4k probably isn't relevant for anything but the old HWCODEC Archos units..
20:57:42BilgusI went thru the last time I fixed all the new warnings and fixed all the warnings for the next gcc version for when someone updated the build system again
20:58:35Bilgusgeaverts a different approach was needed as wodz said the rtc should be sanatized properly but Ive no idea what properly should be
20:58:37gevaertsI didn't like the "fix numbers locally" approach because it's extra code just because the compiler doesn
20:58:45speachyI will say this though −− full accessibility/voice is probably Rockbox's best "killer feature"
20:58:54gevaertst know those numbers have a limited range
20:59:24Bilgusyeah speacht rb is the only accessible dap firmware for the blind
21:00:10gevaertsBilgus: as I said, I did 3665eecf earlier today because I didn't know you were working on it. I went for increasing the buffer for hosted builds (where presumably 40 bytes extra stack is not very much)
21:00:50BilgusI'm up for whatever as long as that damn warning goes away lol
21:01:10gevaertsFeel free to revert that (or should loudly so I revert it) if I'm on the wrong track there
21:02:32gevaertsspeachy: it definitely was a killer feature, but as with all else, smartphones have probably taken over for most people
21:02:47Bilgusup to you wods doesn't like my approach either but I've grown tired of the yellow lol
21:03:00gevaertsOK, up to wodz then :)
21:03:12Bilgusthe blind don't like touchscreens
21:03:27gevaertsSome might not
21:03:41Bilgusfor that matter neither do I, I really like tactile feedback
21:03:54speachyexactly why the fuze+ was so gawdawful.
21:03:57*__builtin grumbles
21:04:04__builtin"killer feature" = "quake"
21:04:11Bilgusyeah I own one and its my trucks DAP
21:04:23speachyit's hard to play quake on a 160x120 monochrome OLED screen. Just saying. :P
21:04:47BilgusI really like lua as that allows the people to make their own stuff :p
21:04:49gevaertsOne of my coworkers is blind. He basically is a cyborg with an integrated iphone and apple watch
21:05:18gevaerts__builtin: usually people take the "killer" in "killer feature" as something other than "shooting things" :)
21:05:23speachyI cheered inside when the screen on my fuze+ broke.
21:10:18Bilgusgevaerts, aren't you the one with the DAP international museum?
21:10:41gevaertsI don't have *that* many
21:10:58BilgusI bet gevaerts wouldn't cheer if one of his lesser favorite targets got broken
21:11:07speachykeep seeing connection stalled messages when the builders are trying to upload the manuals.
21:12:11gevaertsYes, there's something weird going on there. Tracing stuff said that the cgi on the log upload returned http 500
21:12:30gevaertsI have *no* idea why
21:13:36*gevaerts asks zagor
21:25:54gevaertsspeachy: do you see a hard reason not to just include lang_enum.h unconditionally in plugin.h? It's just an enum declaration, that shouldn't change anything, and I'm fairly sure the issue is caused by the dependency generation not doing the PLUGIN define
21:29:02gevaertsWhen that's done, the player build breaks because the lrcplayer plugin uses get_codepage_name, which was moved to HAVE_LCD_BITMAP in the talk support commit
21:30:33speachyI don't see why not, no
21:31:07*speachy was about to build an archos toolchain to start digging into this mess.
21:31:34gevaertsok, I'll commit that then. I expect just the lrcplayer issue to remain after that, which I declare to be yours :)
21:31:57gevaertsAlthough that's just moving get_codepage_name() back to where it was I think
21:33:32fs-bluebotBuild Server message: New build round started. Revision 4d2874f, 280 builds, 11 clients.
21:33:58gevaertsThat's the #ifdef PLUGIN fix. We'll see if it's *just* get_codepage_name() stuff leftover
21:34:21__builtinspeachy: are the authors of fs#7705 and fs#7912 in docs/CREDITS?
21:34:22fs-bluebot Add speech to battery bench (patches, closed)
21:34:22fs-bluebot Voice enable chessbox plugin (patches, closed)
21:35:06speachythe get_codepage_name thing had been wtapped in a HAVE_LCD_BITMAP in the original talk patch, I moved it up to consolidate it.
21:35:21speachyI believe so, but I'll double-check.
21:36:16speachyyes, all names were already in CREDITS
21:37:18__builtinagh, I grepped the wrong file...
21:39:34gevaertsspeachy: I suspect that was an accident originally. The main get_codepage_name() function is in firmware/common/unicode.c, and doesn't have any such restriction, so just moving it back out should be fine. I'll move it
21:40:10gevaertsI'll keep plugin API version the same. It only changes for the Player, and that one hasn't built yet with the new API version
21:40:29*gevaerts will wait for the current build round to finish first
21:40:56speachysounds good
21:41:09gevaertsThese big changes are always an adventure :)
21:42:00speachymeanwhile twenty years later the Archos Player is still rearing its head. :)
21:42:22gevaertsIt keeps people on their toes!
21:42:47gevaertsWe don't actually know for sure if there's still a working one anywhere in the world
21:42:58*__builtin wonders if anyone has actually tested git HEAD on one of them
21:43:13speachymine literally fell apart, as did the recorder I replaced it with.
21:43:35*gevaerts once got one for testing, but then he used a power supply with the wrong polarity
21:44:16speachyproably wouldn't be that hard to fix. likely just the power regulator that blew..
21:44:17gevaertsI really don't understand why people didn't just add a few diodes to those things
21:44:39BilgusI saw one in the wild a few years back but its probably dead by now surely
21:44:52speachybecause a diode would add $0.05 to the BOM..
21:44:58Bilgusthe diode drops .5v and generates heat
21:45:18Bilgusthe rule of 5's apparently
21:46:41speachysince I'm mucking with my builders anyway, what would it take to add voice capabilities?
21:46:41BilgusI like the trick of a fuse on the powerline and a reversed biased diode across tyhe power rails so it blows the fuse and the device is dead but repairable
21:47:48gevaertsspeachy: you mean to tell you stuff? should explain all :)
21:48:37__builtinalright, sparta-builtin is back :)
21:48:42__builtinstill as root, but sandboxed
21:49:04speachythat doesn't mention anything about what generates the nightly voice files
21:49:13gevaertsOh, that sort of voice!
21:49:15speachyand is similarly quiet
21:50:04gevaertsI don't know exactly how that does it, but there's tools/release/
21:53:07gevaertsI think technically we can get away with adding a new toolchain to, and then just add lines to the builds file on the server
21:59:04*__builtin can get all but arm-app built in Docker
22:01:05gevaertsGreat! As expected, just Player left :)
22:02:10gevaertsThat looks interesting
22:03:03gevaertsspeachy: looks like in your cleanup you managed to make two nested #ifdef HAVE_LCD_BITMAP blocks
22:04:46gevaertsOr not, that was already there
22:07:27fs-bluebotBuild Server message: New build round started. Revision 859f762, 280 builds, 11 clients.
22:14:43*__builtin is getting buildclient messages like these:
22:14:45__builtin2019-07-20 21:20:57 Server message: Fatal build error: Command not found. You have been temporarily disabled.
22:16:18__builtinlogs don't seem to have anything useful
22:16:30gevaertsYes, those are fun. Means you're missing something, but the server doesn't tell you what
22:17:10__builtinI can build manuals... manually
22:17:20gevaertsThe server saw "not found" in the build log somewhere
22:18:43__builtinand it seems to be fine automatically
22:19:25gevaertsDo a manual build and capture the output, and then grep for "not found" in there
22:19:42gevaertsThat should at least give some hints
22:22:26__builtinhere's my client image:
22:23:49__builtindamn, my client is too fast for me to manually type `tail' the on the log files
22:25:47gevaertsWhich build is it?
22:25:51gevaertspdf or html?
22:26:12gevaertsThat runs ../tools/configure −−target=iriverh120 −−type=s && make MAKEFLAGS= manual-zip|grep -v 'TeX4ht warning.*not found'
22:26:42gevaertsCan you run (../tools/configure −−target=iriverh120 −−type=s && make MAKEFLAGS= manual-zip|grep -v 'TeX4ht warning.*not found') > logfile 2>&1
22:27:01gevaertsand then put the logfile on a pastebin?
22:27:31__builtinno output on stdout
22:27:52gevaertsWell, everything should be going to that file if you run that
22:28:51gevaertsThat's exactly what the client runs, except it uploads it to the build server instead of to a pastebin
22:32:13__builtinThe grep -v ought to eliminate all the TeX4ht* lines, right?
22:32:32__builtinahh... it's wrapping onto a new line for some reason
22:32:42__builtinso the "not found" is slipping through
22:32:44gevaertsYes, it now also does that gere
22:33:04*gevaerts tries something
22:33:13__builtingrep multiline?
22:33:31gevaertsFirst COLUMNS=300
22:33:48gevaertsMaybe I can trick it into not wrapping. I think that would be the best solution
22:35:46__builtinit seems this build is the one left holding up the build round
22:36:03gevaertsI'm not convinced it's exactly this issue, but it might be
22:36:47***No seen item changed, no save performed.
22:44:10gevaertsOK, if you set max_print_line = 1000 in texmf.cnf (which for me is /etc/texmf/web2c/texmf.cnf, but I had to make a file in /etc//etc/texmf/texmf.d/ with that bit in it), it works
22:45:37gevaertsSo on debian-like systems, as root, "echo 'max_print_line = 1000' > /etc/texmf/texmf.d/linelength.cnf && update-texmf"
22:53:01gevaertsWait, is it now triggering on finding "not found" in "grep -v 'TeX4ht warning.*not found'"?
22:54:21gevaertsI've changed the build command to try to avoid that. We'll see at the next build...
23:17:49speachywow, a completely green build.
23:19:50*speachy is back from unpacking/organizing the garage..
23:19:54*__builtin doesn't understand gevaerts's patch
23:20:02__builtinhow does n->[n] change anything?
23:20:59gevaerts__builtin: easy :)
23:22:15gevaertsThe server (in looks for literally "not found", and the build command included that. Now it has "[n]ot found", which in regular expression terms is exactly the same, but won't trigger the server
23:22:57__builtinah, it was triggering on the build command itself?
23:23:10gevaertsIt's the same thing you do when running "ps -ef|grep some_command", which will find your grep line, so if that's inconvenient you do "ps -ef|grep [s]ome_command"
23:23:22gevaertsI think so. I'm not entirely sure
23:23:31__builtindid you get around the tex wrapping too?
23:24:02speachywell, I can kick off another build by voice-enabling another plugin
23:24:08gevaertsYes, see above. I didn't find how to disable that for the command, but I did find how to do it system-wide
23:24:24__builtinoh, duh
23:24:45*__builtin should generate some screenshots for quake and wolf
23:25:08__builtinseeing as I now have a decently fast machine, it shouldn't take a week as it used to :)
23:25:31gevaertsIt will
23:25:48gevaertsI mean, if you start those plugins, you might find yourself distracted :)
23:26:25__builtinthink I could port mesa and quake 3?
23:27:32gevaertsDepends. What framerate are you aiming for?
23:28:01gevaertsRAM isn't an issue if you have sufficient disk to swap things out to, and you have time
23:28:22speachyheh, some of the newer targets have a 3d engine in their SoC.
23:39:53fs-bluebotBuild Server message: New build round started. Revision 5a84239, 280 builds, 12 clients.
23:40:21speachylet's see if the iriver manual build succeeds this time..
23:46:51__builtingigabeatswps fails on my client
23:46:56__builtinsame "command not found" error

Previous day | Next day