mario-goulart has quit [Ping timeout: 264 seconds]
Thorn has joined #yocto
BoJalling has quit [Ping timeout: 276 seconds]
BoJalling has joined #yocto
jclsn has quit [Ping timeout: 250 seconds]
jclsn has joined #yocto
yssh has joined #yocto
BoJalling has quit [Ping timeout: 276 seconds]
BoJalling has joined #yocto
sakoman has quit [Quit: Leaving.]
seninha has quit [Quit: Leaving]
kevinEathan has joined #yocto
AKN has joined #yocto
Thorn has quit [Ping timeout: 265 seconds]
kevinEathan has quit [Quit: Leaving]
kevinrowland has quit [Ping timeout: 260 seconds]
BoJalling has quit [Ping timeout: 276 seconds]
BoJalling has joined #yocto
goliath has joined #yocto
mario-goulart has joined #yocto
BoJalling has quit [Ping timeout: 246 seconds]
BoJalling has joined #yocto
Thorn has joined #yocto
goliath has quit [Quit: SIGSEGV]
rob_w has joined #yocto
<JaMa>
hmm why does oe-selftest -r eSDK.oeSDKExtSelfTest.test_image_generation_binary_feeds seemingly test every possible selftest including 2023-03-13 06:16:05,962 - oe-selftest - INFO - test_yocto_source_mirror (buildoptions.SourceMirroring.test_yocto_source_mirror)
<JaMa>
when I was running selftest with -a and now I've killed it when it was running every test when selected only this one with -r, what's going on? :/
<JaMa>
overwrote the console log from that :) is there something like cooker log? I was saving the output with tee -> yes there is 14M build/tmp/log/oe-selftest-results-20230312020444.log
<JaMa>
in the complete selftest with -a both failed due to 2023-03-12 12:08:41,286 - oe-selftest - INFO - ERROR: setUpClass (eSDK.oeSDKExtSelfTest) [ (subunit.RemotedTestCase)
<JaMa>
ERROR: Task (virtual:native:/OE/build/poky/build-st-537178/tmp/work/qemux86_64-poky-linux/core-image-minimal/1.0-r0/selftest-esdk-s_oamu60/layers/poky/meta/recipes-support/attr/attr_2.5.1.bb:do_fetch) failed with exit code 'setscene ignore_tasks'
<JaMa>
I'll check why these 2 tests don't work with -r later (after finishing fixing other tests which failed in complete run and can be started individually with -r)
<RP>
JaMa: the autobuilder does filter some of the tests it runs, e.g. the source mirror and reproducibility ones are run separately iirc. I don't remember how the esdk tests work though
mvlad has joined #yocto
<JaMa>
they themselves look very simple, this one just calls devtool build <image>, but somehow oe-selftest triggered every test instead, will debug it later (just said that the builder was re-testing everything again whole night instead of just the tests which were failing)
<JaMa>
said/sad
<JaMa>
runqemu script looks like another worm hole wrt IMAGE_LINK_NAME
<RP>
JaMa: it is weird it would try and run them all. I'll try and get to the bottom of that weird tinfoil issue
<JaMa>
should I send my patches for various selftest issues for mickledore (without changing the naming of deploy artifacts - just unblocking someone to change the default and still use selftest)?
invalidopcode1 has quit [Remote host closed the connection]
invalidopcode1 has joined #yocto
Schlumpf has joined #yocto
AKN has quit [Ping timeout: 276 seconds]
seninha has joined #yocto
azcraft has joined #yocto
<RP>
JaMa: I did have some questions/worries about some of them I need to look further into but no harm in sending them and we can take a look
BoJalling has quit [Ping timeout: 246 seconds]
BoJalling has joined #yocto
florian_kc has joined #yocto
frieder has quit [Read error: Connection reset by peer]
leon-anavi has joined #yocto
yssh has quit [Ping timeout: 260 seconds]
Guest82 has joined #yocto
yssh has joined #yocto
<Guest82>
Hello, I'm trying to figure out what it would take to integrate certbot into an image so it can create certificates for me from letsencrypt.org.
<Guest82>
I could not find a package that provides that, would any of you be aware of an existing way to do this?
yssh has quit [Ping timeout: 260 seconds]
AKN has joined #yocto
Thorn has quit [Ping timeout: 264 seconds]
<JaMa>
INFO - FAIL: test_git_mirrors (fetch.Fetch.test_git_mirrors) is easy to explain, but more difficult to fix, it sets (PRE)MIRRORS:forcevariable = "" and expects to trigger fetch failure, but PREMIRRORS:prepend in my local.conf defeats it
<JaMa>
I can switch from prepend to =+ in my local.conf but that will solve it only in my environment
<RP>
JaMa: I've wondered if we should avoid local.conf and only copy in the specific things we know we might need
<RP>
JaMa: there are other variables which cause issues like this
<JaMa>
right, but on the other hand I've PREMIRROR and shared SSTATE_DIR in local.conf explicitly for other tests to re-use them instead of fetching/building locally unless explicitly defined by the test
<JaMa>
so to me it looks more like :forcevariable-for-real-this-time-I-know-what-I-am-doing-in-this-case :)
<RP>
JaMa: We've copy some key things to a variable set in the standard way for test purposes
<JaMa>
but it would still need to accept some other .conf file, right? e.g. if I really want to run selftest with IMAGE_NAME_SUFFIX set to empty, now I would put it in local.conf and assume selftest will respect that
<JaMa>
if it picks only specific key things, then I could modify the defaults in e.g. image-artifacts-names.bbclass before the tests, but that's not the same as testing that people can easily set them in local.conf
BoJalling has quit [Ping timeout: 260 seconds]
BoJalling has joined #yocto
kalj has joined #yocto
frieder has joined #yocto
nemik has quit [Ping timeout: 246 seconds]
starblue has quit [Ping timeout: 250 seconds]
nemik has joined #yocto
starblue has joined #yocto
nemik has quit [Ping timeout: 265 seconds]
nemik has joined #yocto
rob_w has quit [Quit: Leaving]
<RP>
JaMa: I've found the "tinfoil" issue. "d.getVar = Mock(return_value=commonlicdir)" is never undone
<RP>
JaMa: the selftests are meant to be self contained so I'm not sure it should depend on local.conf
Thorn has joined #yocto
nemik has quit [Ping timeout: 265 seconds]
nemik has joined #yocto
<JaMa>
RP: hehe, that's nasty 'this is mock value' would be easier to spot and grep for :)
<JaMa>
RP: and where would you define the changes you want selftest to test if not local.conf?
alessioigor has joined #yocto
nemik has quit [Ping timeout: 246 seconds]
nemik has joined #yocto
<RP>
JaMa: I'd normally just apply the actual changes and then test
<RP>
JaMa: most of the changes I make/test can't be switched from local.conf
<JaMa>
> then I could modify the defaults in e.g. image-artifacts-names.bbclass before the tests, but that's not the same as testing that people can easily set them in local.conf
<JaMa>
but I agree that this is quite specific corner case and there are bigger issues in selftest to tackle
<JaMa>
It's just that I regulary don't use selftest, so it's useful for me to document it's current limitations or differences of my system related to that - I'll include it in pull-request (in case it would be useful to someone else as well)
<RP>
JaMa: definitely, I'm just offering my perspective too :)
<RP>
JaMa: my usage probably isn't "standard"
<JaMa>
:)
<JaMa>
I'm using pretty clean poky clone (as I normally don't use poky), but after seeing duplicated sstate-cache in build-st*, I've added SSTATE_DIR in my local.conf and realized that DL_DIR and my default PREMIRROR might be useful as well :)
<JaMa>
in 99.9% test cases it was, only fetch.Fetch.test_git_mirrors seems to fail because of that
<JaMa>
also using -K -B <test-name> when debugging individual tests, instead of reusing the same build-st for many tests (not sure how often this is used/tested on AB)
<RP>
JaMa: the oeqa code could do with improvement in places so I'm not entirely surprised. rburton and I have tried over time to make it better
<RP>
JaMa: patches very welcome
AKN has quit [Ping timeout: 246 seconds]
<JaMa>
I plan to send one, but really don't know what would be the right regex which would suit everybody as these commits pretty much contradicts each other AFAIS
<JaMa>
Saur[m]: Saur: as author of the initial regex and that comment, do you have some opinion above ^^
alessioigor has quit [Quit: alessioigor]
alessioigor has joined #yocto
<JaMa>
would you accept eSDK module rename to esdk to pass current assumption in the regex?
davidinux has joined #yocto
<JaMa>
that looks like easiest solution, testing this now
<RP>
JaMa: I was hoping we could just allow upper and lower in it...
madisox_ has quit [Ping timeout: 250 seconds]
madisox_ has joined #yocto
invalidopcode1 has quit [Remote host closed the connection]
invalidopcode1 has joined #yocto
AKN has joined #yocto
sakoman has joined #yocto
BoJalling has quit [Ping timeout: 255 seconds]
BoJalling has joined #yocto
astlep5504 has joined #yocto
<marka>
RP, I see the python stdout change busted the bblogging.BitBakeLogging.test_python_exec_func_shell_loggingC and bblogging.BitBakeLogging.test_python_exec_func_shell_loggingD while I am guessing you had it staged
<marka>
I will take a look today, sorry about that, I had not run the whole testsuite as I probably should have
rsalveti has joined #yocto
<RP>
marka: np, I'm happy we're fixing that issue! :)
<RP>
marka: this is why we have the tests. I suspect there is a test case in that suite you can enable after the fix too
kscherer has joined #yocto
<marka>
ah, I had not looked to see if there were workarounds in the tests, will definitely review them
<marka>
good distraction while working through mundane paperwork
astlep5504 has quit [Remote host closed the connection]
<JaMa>
eSDK rename sent together with loader improvements to prevent starting all tests, now esdk tests relatively quickly fail with "ERROR: Task attr-native.do_fetch attempted to execute unexpectedly" which is separate issue
<RP>
JaMa: I'm still thinking it might be better just to accept uppercase there :/
<JaMa>
RP: uppercase was there before and explicitly removed in 2020
<RP>
JaMa: wasn't that to make 4 level modules work though?
<Saur[m]>
JaMa, RP: I haven't been following the discussion here today, but if jump into the middle of it, it definitely looks as if the changes after my initial regexp was added haven't taken my comment into account. However, if the comment is tweaked a little (`do not contain` to `do not start with`), and then that comment is actually true, then the original regexp should be changed to (untested)
<JaMa>
Saur[m]: it's probably easier to read there than IRC history
<JaMa>
"wasn't that to make 4 level modules" yes, that's what the commit message says, but why did he change it to accept 1-2 and not 3-9 or 0? why not \w. if only the dot was missing?
Guest82 has quit [Quit: Client closed]
<RP>
JaMa: I don't know. I'd love to simplify this if we could though, I doubt it needs to be this complex
<JaMa>
agreed, but renaming 1 file seems safer to me
<RP>
JaMa: until someone creates another "badly" named one
<RP>
evens chance I'll forget and be the one to do it :/
<JaMa>
yes, but now it will show proper warning and won't trigger 10+ hour long testing
<Saur[m]>
RP: With `eSDK.py` renamed to `esdk.py`, my original regexp can be restored.
<JaMa>
no uppercase assumption was explicitly mentioned there since 2017 and eSDK only abuser
<RP>
Right, but nothing stops someone calling a module that. I really don't like code with assumptions that aren't easily discovered
<Saur[m]>
Are there actually any limitations on what a module can be called, e.g., that it has to start with a lower case letter, or is that just a convention?
<JaMa>
WARNING - module 'eSDK.oeSDKExtSelfTest.test_install_libraries_headers' was skipped from selected modules, because it doesn't match with module name assumptions: package and module names do not contain upper case characters, whereas class names do
<JaMa>
I agree, but ^ makes it relatively easy to discover
<RP>
It looks like python module naming is supposed to be lowercase
<JaMa>
and honestly there are bigger issues to fix in oeqa
<RP>
If we're going to fix this, lets at least fix things properly so they can't/don't regress :/
<RP>
Can we make the thing hard error if a module has a name with caps in?
<JaMa>
it's hard error if all selected modules are skipped, but I don't mind making it error instead of warning or triggering OEQATestNotFound even when only one of them is skipped
<JaMa>
but the continue from you looked like in some cases you might want to just skip and continue with other tests
<RP>
JaMa: I was more thinking of having it error if any module with caps was found
<JaMa>
I see
<RP>
that stops my concern about weird errors in corner cases
<JaMa>
yeah, I was surprised that all modules are loaded when only some are selected, but you're right we can error out a bit sooner there
<JaMa>
will return to it after finishing debugging the other issues in testimage and runqemu
invalidopcode1 has quit [Remote host closed the connection]
invalidopcode1 has joined #yocto
BoJalling has quit [Ping timeout: 250 seconds]
BoJalling has joined #yocto
xmn has joined #yocto
<JaMa>
ah bitbake -c testimage <image1> <image2> doesn't work when you have just one preconfigured tap device
frieder has quit [Ping timeout: 265 seconds]
Amynka has quit [Ping timeout: 265 seconds]
Amynka has joined #yocto
alessioigor has quit [Quit: alessioigor]
alessioigor has joined #yocto
goliath has quit [Quit: SIGSEGV]
frieder has joined #yocto
alimon has joined #yocto
alessioigor has quit [Quit: alessioigor]
alessioigor has joined #yocto
Poppy has joined #yocto
<Poppy>
There is a simple way to avoid installation conflicts between files ? One recipe create user account and home directory, an other recipe populate files in /home/user/. Yocto complain (file /home/user conflicts between attempted installs of recipe1 and recipe2)
<landgraf>
Poppy: make sure mode (access rights/bits) are the same
BoJalling has quit [Ping timeout: 268 seconds]
BoJalling has joined #yocto
ptsneves has quit [Ping timeout: 276 seconds]
kalj has quit [Ping timeout: 260 seconds]
frieder has quit [Remote host closed the connection]
<Poppy>
landgraf thx for the tips, but same issue with all 755 mode.
GNUmoon2 has quit [Ping timeout: 255 seconds]
mkazantsev has joined #yocto
GNUmoon2 has joined #yocto
nemik has quit [Ping timeout: 256 seconds]
nemik has joined #yocto
rfuentess has quit [Remote host closed the connection]
mckoan is now known as mckoan|away
goliath has joined #yocto
<Saur[m]>
Poppy: At least with RPM it works for multiple packages to install the same *directory* as long as the mode, owner and group are identical. However, it is a bit unclear from your question if the error is related to the attempted installation of a file or a directory. Contrary to directories, files can only be installed from one package.
<Poppy>
Saur[m] Error is about the user home directory: Transaction test error:
<Poppy>
file /home/user conflicts between attempted installs of user-accounts-conf-1.0-r0 and ssh-userkeys (one of the recipe is used to create the home directory, the other populates public keys inside /home/user/.ssh
<Poppy>
Saur[m] Maybe some more info, all home directories must have root:root rights to be used with ftp/sftp
<rburton>
Poppy: does user-accounts-conf create the user?
<Saur[m]>
Poppy: Are you using RPM, IPK or Debian packages to create the rootfs?
<Poppy>
rburton yes the recipe inherit useradd and set USERADD_PACKAGES / USERADD_PARAM / GROUPADD_PARAM like the example
<rburton>
Poppy: does ssh-userkeys depend on that recipe so that it also knows about the new user?
<Saur[m]>
It also need to RDEPENDS:${PN} on it so that the user is actually created in runtime. Alternatively, both recipes should create the user themselves if you do not want them to depend on each other.
<Saur[m]>
s/runtime/rootfs/
<Saur[m]>
Assuming you are using RPM, you can use `rpm -qvf tmp/deploy/rpm/<architecture>/<package name>` to list the contents of the two recipes. The lines for the user directory should be identical.
<Poppy>
Saur[m] you are probably on the right way : Postinstall scriptlets of ['ssh-userkeys'] have failed. If the intention is to defer them to first boot,
<Poppy>
then please place them into pkg_postinst_ontarget:${PN} (). I have to read the doc a little bit more :)
<zelgomer>
JaMa: ah-ha, thanks. is it related to some magic here? 'do_image_complete[sstate-inputdirs] = "${IMGDEPLOYDIR}"' and '...[sstate-outputdirs] = "${DEPLOY_DIR_IMAGE}"' ?
<zelgomer>
my problem is that i want to add a task after do_image_complete to generate an additional file which tars up the rootfs image, initramfs, kernel, and some other boot files, which is used for distribution. i'm trying to decide how to best implement this
<JaMa>
last 2 selftest failures in esdk.oeSDKExtSelfTest seem related to uninative disabled in my poky build (due to my gentoo having newer glibc and gcc than uninative) and then eSDK somehow trying to use universal attr-native.do_deploy_source_epoch and fails to find it causing the class SetUp to fail with ERROR: Task attr-native.do_fetch attempted to execute unexpectedly
Haxxa has quit [Quit: Haxxa flies away.]
<JaMa>
after doing the same test in ubuntu-23.04 docker container I found that file-native detects some dependencies from host and createrepo-c-native is broken by it as well, but after fixing that (patch sent in few minutes) it passed both tests
<JaMa>
zelgomer: IMAGE_FSTYPE are a bit more complicated to understand, but once implemented correctly it's pretty powerful
<zelgomer>
i'm not sure if i can use IMAGE_FSTYPE because there seems to be no dependency on the kernel. i had this implemented as an IMAGE_POSTPROCESS_COMMAND before and it would occasionally fail because the kernel wasn't deployed yet
<zelgomer>
i'm a bit confused about this
<JaMa>
we use IMAGE_FSTYPE for packages with kernel, but the code is not public
<zelgomer>
i see
<JaMa>
check rpisdimg in meta-raspberrypi that might provide some clues as well (even when many users now use wic.rpi instead)
<zelgomer>
thanks. so far at a glance i see several constructs that i had attempted, but maybe if i stare long enough i'll see where my attempts went wrong :)
<zelgomer>
we don't want to use wic, we want to distribute tar
<JaMa>
maintaining world timezones is much more difficult than what I've expected earlier today (before reading e.g. https://mm.icann.org/pipermail/tz/2021-May/030089.html ) and yes I'm reading tz@iana.org just because I have a patch for tzdata-native to build with gcc-13
vladest1 has joined #yocto
vladest has quit [Read error: Connection reset by peer]
vladest1 is now known as vladest
azcraft has quit [Ping timeout: 250 seconds]
florian_kc has joined #yocto
azcraft has joined #yocto
<mischief>
zelgomer: for us we have an image class. image class defines a new CONVERSIONTYPES. in CONVERSION_CMD we just run tar
<mischief>
and CONVERSION_DEPENDS uses virtual/kernel:do_deploy so we just pull kernel out of deploy dir
<mischief>
maybe not the correct way but works for us for quite some time now
<zelgomer>
mischief: i'm not trying to create a tar of the root filesystem, though. i want to create a tar which contains the kernel, a few other boot files, and a root filesystem image file. do you think the CONVERSIONTYPES method is still a good approach in that case?
<JaMa>
mischief: I consider CONVERSIONTYPES as part of IMAGE_FSTYPE :)
nemik has quit [Ping timeout: 255 seconds]
nemik has joined #yocto
nemik has quit [Ping timeout: 250 seconds]
nemik has joined #yocto
<mischief>
zelgomer: a 'good approach'? no idea. it works for us though
zwelch has quit [Ping timeout: 248 seconds]
<mischief>
zelgomer: 'i want to create a tar which contains the kernel..' yes, that's what we do too, except it's only kernel+rootfs. but of course you could add whatever else as long as the task has the right dependencies on the deploy dir i believe
<zelgomer>
got it, thanks. i will play around with it
BoJalling has quit [Ping timeout: 276 seconds]
BoJalling has joined #yocto
<vladest>
Hello
<vladest>
any idea how to make this fixed:
<vladest>
ERROR: Nothing PROVIDES 'tslib-native' (but virtual:native:/hdd/rpiyocto/meta-qt6/recipes-qt/qt6/qtbase_git.bb DEPENDS on or otherwise requires it)
<vladest>
?
<JaMa>
vladest: don't enable tzlib for native qtbase
<JaMa>
it's not in default PACKAGECONFIG so it's probably your config somewhere, use bitbake-getvar if you don't know where
zwelch has joined #yocto
alessioigor has quit [Quit: alessioigor]
seninha has quit [Ping timeout: 256 seconds]
invalidopcode1 has quit [Remote host closed the connection]
invalidopcode1 has joined #yocto
BoJalling has quit [Ping timeout: 276 seconds]
BoJalling has joined #yocto
kscherer has quit [Quit: Konversation terminated!]
<paulg>
Setting up an old abandoned turd with ubu-20 so it can build Yocto.
<paulg>
It nags me because "diffstat" isn't installed.
<paulg>
I love diffstat - used it all the time prior to git.
<paulg>
But how in the hell is that *required* for a build?
<paulg>
That is a developer/autopsy tool.
leon-anavi has quit [Quit: Leaving]
<rburton>
i _think_ its quilt that wants diffstat
<rburton>
feel free to patch it out if you can be bothered
<rburton>
but you can use quilt-native interactively in a devshell, so breaking it might not be a good idea
<paulg>
yeah. I'll add it to my "to-do" list once I've retired and got a dump-truck of free time to go fix annoyances... :-P
<mischief>
you coul grab the yocto prebuilt stuff if you don't want to modify the host
<mischief>
had to use that recently for amazon linux 2, which has a dreadfully ancient copy of tar installed
<paulg>
mischief, yep - thanks, I've been using buildtools for a while - since the python update requirement made ubu-18.04 unfit anymore - as many others have no doubt experienced.
GNUmoon2 has quit [Remote host closed the connection]
<paulg>
I'm not complaining as a user who can't build - I'm complaining as an old fart who hates to see creeping requirements and general bloat.
<paulg>
If that makes sense to people....
GNUmoon2 has joined #yocto
starblue has quit [Ping timeout: 250 seconds]
starblue has joined #yocto
starblue has quit [Ping timeout: 276 seconds]
starblue has joined #yocto
starblue has quit [Ping timeout: 250 seconds]
starblue has joined #yocto
BoJalling has quit [Ping timeout: 255 seconds]
BoJalling has joined #yocto
kevinrowland has joined #yocto
seninha has joined #yocto
goliath has quit [Quit: SIGSEGV]
astlep5504 has quit [Remote host closed the connection]
astlep5504 has joined #yocto
starblue has quit [Ping timeout: 276 seconds]
starblue has joined #yocto
astlep5504 has quit [Remote host closed the connection]
azcraft has quit [Quit: Leaving]
starblue has quit [Ping timeout: 246 seconds]
starblue has joined #yocto
BoJalling has quit [Ping timeout: 268 seconds]
BoJalling has joined #yocto
GNUmoon has joined #yocto
GNUmoon2 has quit [Ping timeout: 255 seconds]
Ram-Z has quit [Ping timeout: 260 seconds]
LocutusOfBorg has quit [Ping timeout: 255 seconds]