00:09
librasteve_ left
08:42
sena_kun joined
13:08
[Coke]__ joined
13:10
[Coke]_ left
|
|||
Geth | HTML-Strip/main: 2e0300cf86 | (Elizabeth Mattijsen)++ | .travis.yml We don't do Travis anymore |
15:49 | |
rakudo/azure_improvements: 7d3fecb1c2 | (Timo Paulssen)++ | azure-pipelines.yml Grab a bunch of debug output using strace and ldd |
18:47 | ||
rakudo/azure_improvements: 0843566917 | (Timo Paulssen)++ | azure-pipelines.yml build nqp and rakudo with --no-silent-build for helpful(?) details |
18:52 | ||
rakudo/azure_improvements: f04161d152 | (Timo Paulssen)++ | azure-pipelines.yml fix paths to binaries for debug output |
18:57 | ||
rakudo/azure_improvements: 2e7ef049c7 | (Timo Paulssen)++ | azure-pipelines.yml more and more debug output |
19:04 | ||
rakudo/azure_improvements: 3da5de9924 | (Timo Paulssen)++ | azure-pipelines.yml replace "perl6" with "raku" |
|||
rakudo/azure_improvements: ff35d53556 | (Timo Paulssen)++ | azure-pipelines.yml put debugspam in a new artifact |
19:20 | ||
rakudo/azure_improvements: c48d2ebf10 | (Timo Paulssen)++ | azure-pipelines.yml look at only one, correct, path for debugspam |
19:30 | ||
ab5tract | lizmat: The latest release has a few major (if hack-ish) quality of life improvements for working on `rakudo.git` -- github.com/ab5tract/comma-plugin/r...2.0-beta.2 | 19:49 | |
Geth | rakudo/azure_improvements: 17f25f06b5 | (Timo Paulssen)++ | azure-pipelines.yml use absolute path as --prefix instead of relative |
19:50 | |
rakudo/azure_improvements: 1add5e0096 | (Timo Paulssen)++ | azure-pipelines.yml output actual debugspam files, not left-overs(???) |
|||
rakudo/azure_improvements: 4a01a33a8c | (Timo Paulssen)++ | azure-pipelines.yml fix output dir for debugspam |
19:59 | ||
timo | i sure hope nobody has to pay for all my pipeline runs, i'm just assuming we're getting them for free so i can retry a lot | ||
Geth | rakudo/azure_improvements: c25d2c3912 | (Timo Paulssen)++ | azure-pipelines.yml win and relocatable mac/linux work fine, skip them for now |
20:02 | |
rakudo/azure_improvements: 4a24205d62 | (Timo Paulssen)++ | azure-pipelines.yml have to skip win completely, or the stage has 0 jobs and breaks |
20:03 | ||
rakudo/azure_improvements: f1bfa83284 | (Timo Paulssen)++ | azure-pipelines.yml turn on debug mode for pipelines! |
20:16 | ||
rakudo/azure_improvements: 70bb8372ea | (Timo Paulssen)++ | azure-pipelines.yml try to put debugspam from different jobs/stages together |
|||
rakudo/azure_improvements: 693e4b1e34 | (Timo Paulssen)++ | azure-pipelines.yml need to bump cache version key after changing --prefix= parameter |
20:18 | ||
rakudo/azure_improvements: 1d20d10f12 | (Timo Paulssen)++ | azure-pipelines.yml extra spesh tests depend on relocatable linux build |
20:23 | ||
rakudo/azure_improvements: 7a9570ccef | (Timo Paulssen)++ | azure-pipelines.yml i think linux runners have multiple cores so -j3 yay |
|||
rakudo/azure_improvements: 4bd8618565 | (Timo Paulssen)++ | azure-pipelines.yml prove -j and TEST_JOBS up to 3 for non-windows why not |
20:33 | ||
rakudo/azure_improvements: 2354259c3c | (Timo Paulssen)++ | azure-pipelines.yml oops, need to *build* relocatable, not just test |
|||
rakudo/azure_improvements: 2fd4c61793 | (Timo Paulssen)++ | azure-pipelines.yml i could have sworn you could just add to an existing artifact, but no |
20:45 | ||
timo | patrickb: you know about the azure pipelines, right? what i'm doing here isn't costing any raku person extra money, is that correct? | 20:49 | |
patrickb | timo: They are completely free. | 20:50 | |
timo | can i get some kind of access to the azure devops thing so i can cancel or rerun my own stuff from there? | ||
patrickb | Go ahead, I marvel at your endurance! | ||
Let's see | 20:51 | ||
Geth | rakudo/azure_improvements: bbb5517aac | (Timo Paulssen)++ | azure-pipelines.yml normal variables are not expanded in this property |
20:56 | |
timo | i think the problem for a long time has been that you get a file not found for /blah/blah/doesnotexist/../install which makes sense now that i think about it and probably shouldn't be different? | 21:09 | |
ab5tract | That makes sense to me. Though I have no idea what the context is :) | 21:17 | |
Geth | rakudo/azure_improvements: 7b0f572b15 | (Timo Paulssen)++ | azure-pipelines.yml don't rebuild rakudo before running spectests |
21:26 | |
ab5tract | timo: ^^^^ and I thought the pipelines at work were overly aggressive in rebuilding :O | 21:28 | |
timo | well, this is a Make thing i think | 21:35 | |
after bundling up the install/ folder and the rakudo/ folder in a PipelineArtifact and extracting it in the follow-up stage, "make spectest" has decided that the "all" target also has to be re-built | 21:36 | ||
i'm now having Make spit out why it thinks that, maybe i can find out the exact reason and prevent it. maybe i just have to "touch" the right files or whatever | |||
make: *** unknown debug level specification 'why'. Stop. | 21:37 | ||
love that | |||
requires GNU Make 4.4, that version of make isn't even in the absolute newest ubuntu, much less 20.04 | 21:42 | ||
Geth | rakudo/azure_improvements: 6918d6e8ed | (Timo Paulssen)++ | azure-pipelines.yml make isn't a new enough version to support --debug=why, so --trace instead |
21:47 | |
ab5tract | Good idea. Trying to determine what’s missing by examine the make target definitions would be quite a headache | 21:51 | |
22:04
sena_kun left
|
|||
Geth | rakudo/azure_improvements: 41c94d053b | (Timo Paulssen)++ | azure-pipelines.yml rakudo-m needs to be chmod+x'd after extracting from pipeline artifact |
22:04 | |
timo | sounds like --trace is good enough for now | ||
Makefile:1212: update target 'gen/moar/nqp-version' due to: /home/vsts/work/1/install/bin/nqp-m | 22:05 | ||
^- i'm assuming that's because of modification date or so? | |||
Geth | rakudo/azure_improvements: d89e02936d | (Timo Paulssen)++ | azure-pipelines.yml more debug output to find rebuild reasons before spectests |
22:08 | |
ab5tract | ohhh | 22:19 | |
IIRC, that's definitely enough for make | 22:20 | ||
Geth | rakudo/azure_improvements: 0fd1ed8014 | (Timo Paulssen)++ | azure-pipelines.yml get extra precise time output in ls, sleep briefly between extracting artifacts |
22:23 | |
timo | with just ls -l the date&time was the same, so i need --time-style=full-iso :D | 22:24 | |
ab5tract | lol | 22:27 | |
make is nothing if not meticulous | |||
I sincerely hope there is some flag that allows a bit of variance, or disables time concerns entirely for a given run | 22:28 | ||
timo | well, i don't have to worry about "make m-spectest" rebuilding rakudo a second time if i just remove the "all" dependency from spectest_update | 22:29 | |
with a little call to sed | |||
788057 20K -rwxr-xr-x 1 vsts docker 18K 2024-09-15 22:28:52.874520793 +0000 ../install/bin/nqp-m | 22:31 | ||
787066 4.0K -rw-r--r-- 1 vsts docker 40 2024-09-15 22:28:32.036734674 +0000 nqp-version | |||
ab5tract | got 'em :) | 22:35 | |
timo | ok it's using the timestamps from when it was extracted, so i just have to extract the two artifacts in the opposite order | ||
the other jobs don't have to do it like that because they invoke prove directly instead of "make some-kind-of-test" | 22:37 |