<manuel1985>
Hi fellow Yoctonians! I'd like to add a recipe that provides a tool used to build another package. Is there something to take care of? Is the use of native.bbclass and '-native' suffix in the recipe customary or mandatory?
<TRO[m]>
#AWS #CDK version of #buildbot for #Yocto #linux - might be interesting for more people.
<TRO[m]>
We use it to test meta-aws with all supported releases and archs. (With shared SSTATE, polling changes of remote repos, kvm enabled testimage...)
<Ablu>
manuel1985: If you only ever use it as build tool for other packages then yes. https://docs.yoctoproject.org/singleindex.html#native describes the two options. If the package may also make sense to install as a target then you can follow the second method.
<rburton>
manuel1985: if you inherit native and don't call the recipe -native then you're just asking to confuse people
<rburton>
you _need_ to inherit native to build native code
Kubu_work has quit [Quit: Leaving.]
Guest62 has joined #yocto
<Guest62>
hello
<Guest62>
was wondering if anyone knows if there is already an existing recipe for libtorchtrt and co somewhere ...
<rburton>
unless the recipe name would be different
<rburton>
guess thats pytorch
<rburton>
that's a "challenge" to build
nemik has quit [Ping timeout: 245 seconds]
nemik has joined #yocto
Zaid has joined #yocto
nemik has quit [Ping timeout: 245 seconds]
nemik has joined #yocto
destmaster84 has quit [Quit: Client closed]
destmaster84 has joined #yocto
destmaster84 has quit [Client Quit]
zelgomer has quit [Ping timeout: 240 seconds]
zelgomer has joined #yocto
<mckoan>
what could be a lightweight browser alternative to chromium to be used with Yocto ?
Guest82 has joined #yocto
<manuel1985>
rburton: Thanks for the advice provided above
jetm has joined #yocto
<manuel1985>
Ablu too
<RP>
TRO[m]: We've often wondered what implementing an autobuilder on AWS infrastructure would look like but we don't have the money to be able to run such a thing!
philmd has joined #yocto
manuel_ has joined #yocto
manuel1985 has quit [Ping timeout: 245 seconds]
philmd has quit [Read error: Connection reset by peer]
jetm has quit [Read error: Connection reset by peer]
vladest has joined #yocto
nemik has quit [Ping timeout: 246 seconds]
nemik has joined #yocto
<Guest82>
Hello, I have a NodeJs project that I would like to include in my image. I am on Kirkstone and building linux-fslc-imx for an i.mx8 target.
<Guest82>
I have used devtool to create the recipe for me.
<Guest82>
Now I'm facing an issue where one of the dependency npm packages "leveldown" is failing in the do_package step, for example:
<Guest82>
ERROR: backend-1.0.0+gitAUTOINC+b46ad989d4-r0 do_package: QA Issue: File '/usr/lib/node_modules/backend/node_modules/leveldown/prebuilds/android-arm/node.napi.armv7.node' from backend was already stripped, this will prevent future debugging! [already-stripped]
<Guest82>
When i add "INSANE_SKIP:${PN} += "already-stripped" to the recipe i get:
<Guest82>
do_package_qa: QA Issue: Architecture did not match (ARM, expected AArch64) in /usr/lib/node_modules/backend/node_modules/leveldown/prebuilds/android-arm/node.napi.armv7.node
<Guest82>
It seems to me the prebuilt android-arm should not be used anyway in my case, so I should somehow set --build-from-source for leveldown? I have tried adding .npmrc with this command in the root but this did not help.
<Guest82>
Looking at npm.bbclass I could not figure out if I should tell it to do something specific.
<Guest82>
Would any of you know where to start debugging this?
<TRO[m]>
<RP> "TRO: We've often wondered what..." <- I mean setting up a build cluster in a few minutes is great. The running costs should be low, just the builder (on demand) instance type should be considered carefully...
<RP>
TRO[m]: we have around ~65 different "builders" active on any give test run and they use a fair bit of resources to run the builds so I suspect it would be expensive
<RP>
TRO[m]: it would be interesting to see what "oe-selftest -r reproducible" would 'cost' or "oe-selftest -a --skip-tests distrodata.Distrodata.test_checkpkg buildoptions.SourceMirroring.test_yocto_source_mirror reproducible -T machine -T toolchain-user -T toolchain-system -j 15"
<RP>
we run the latter four times on centos, debian, fedora and ubuntu
<RP>
" ninja: error: manifest 'build.ninja' still dirty after 100 tries, perhaps system time is not set"
<dl9pf>
RP: ok will take a look (smurray is on vacation)
<RP>
dl9pf: thanks
<TRO[m]>
RP: We can have this discussion offline ;)
Tyaku has joined #yocto
psi34 has joined #yocto
<RP>
TRO[m]: I've just had a lot of people tell me things about this over the years and in general people don't appreciate the scale of our servers/tests
rber|res has joined #yocto
<psi34>
Hi! I think I've found a bug with externalsrc and dependency handling. When I call clean on an externalsrc recipe then call prepare_recipe_sysroot for a recipe that depends on it, I get the following "The sstate manifest for task 'dummy_cleaned_task:populate_sysroot' (multilib variant '') could not be found.". If I build the dummy_cleaned_task
<psi34>
fully it works fine, also works fine when I convert the recipe to not use externalsrc. I think the "if task.endswith("_setscene"): bb.build.deltask(task, d)" causes the problem somehow or breaks an assumption elsewhere because if I disable this it also works. Has anybody encountered something similar?
alessioigor has quit [Quit: alessioigor]
alessioigor has joined #yocto
<RP>
psi34: it is probably worth filing a bug for that
alessioigor has quit [Quit: alessioigor]
alessioigor has joined #yocto
Zaid has quit [Quit: Client closed]
<JerryM>
if I (r)dpend on a package with a certain PACKAGECONFIG is it possible to check this?
Minvera has joined #yocto
<tgamblin>
Do we have a style/formatting guide for our Python scripts?
xmn has joined #yocto
rob_w has quit [Remote host closed the connection]
<rburton>
tgamblin: personally, what black says
Net147 has quit [Quit: Quit]
Net147 has joined #yocto
Net147 has joined #yocto
Net147 has quit [Changing host]
psi34 has quit [Quit: Client closed]
<tgamblin>
rburton: thanks!
<rburton>
(nobody has ran it over the existing code)
Xagen has joined #yocto
<tgamblin>
rburton: One of the other BL devs was asking but I didn't have an answer other than maybe pylint. I've suggested they look at black
Xagen has quit [Client Quit]
<rburton>
black is super-opinionated but there's a lot of zen from that
Xagen has joined #yocto
<RP>
zeddii: FWIW other builds all worked even on that worker
sakoman has joined #yocto
<zeddii>
RP: that is definitely strange. ninja just went insane.
<LetoThe2nd>
zeddii: i want the above quote out of context.
* zeddii
will have to avoid japan if the quote gets out
<RP>
zeddii: I doubt it is meta-virt related :/
alessioigor has quit [Quit: alessioigor]
alessioigor has joined #yocto
StephanKlatt[m] has joined #yocto
amitk has joined #yocto
fray has joined #yocto
alessioigor has quit [Quit: alessioigor]
alessioigor has joined #yocto
<manuel_>
Does the devtool shall have userspace qemu enabled?
<manuel_>
Today a collegue of mine was adding a recipe for a tool he wants to include in the docker build
<manuel_>
It's a native recipe but he failed to mark it as -native and choose the wrong architecture as well
<manuel_>
(the arch of the target, not of the build host)
<manuel_>
to my surprise, he could execute that binary in the devtool shell. That was really puzzling me.
<rburton>
nope
<rburton>
it probably just used your native compiler
<rburton>
did you check the binary was actually the wrong arch?
<manuel_>
jepp, did so. I checked with `file`.
<manuel_>
The file (pandoc) was precompiled. It's written in Haskell from what I know.
<tlwoerner>
perhaps use MACHINE names for each combo?
DvorkinDmitry has joined #yocto
<DvorkinDmitry>
I need to overwrite the file, generated by another recipe. can't use .bbappend. What can I do?
<rburton>
why can't you use a bbappend? that's what they're for.
Guest82 has quit [Quit: Client closed]
<vvn>
tlwoerner: isn't what SOME_SETTING:the-base-board provides you?
<vvn>
RP: I like ${TOPDIR}/tmp-${DISTRO} because it keeps the same directory depth. Also deploy dir already splits per-machine artifacts, so I think it'd be convenient to explicit the distro in the build path too as a saner default setup.
Wenqing has quit [Read error: Connection reset by peer]
Wenqing has joined #yocto
nemik has quit [Ping timeout: 252 seconds]
nemik has joined #yocto
nemik has quit [Ping timeout: 246 seconds]
nemik has joined #yocto
nemik has quit [Ping timeout: 245 seconds]
nemik has joined #yocto
nemik has quit [Ping timeout: 245 seconds]
nemik has joined #yocto
<DvorkinDmitry>
rburton, case I need to keep configs and files for this image in one recipe. I just need to overwrite the file that installed previously by OE
<JPEW>
DvorkinDmitry: One way would be to bbappend the OE recipe to make it `rm` the file in do_install:append. Then your new recipe wouldn't conflict
<JPEW>
DvorkinDmitry: But.... you might want to consider why your new recipe _has_ to be the one to install the file; it's probably not best practice to do that
<DvorkinDmitry>
JPEW, I see. I have two images for the same machine: first image has normal nginx, another has nginx + custom htmls + custom php config + custom nginx. all I need is just to add myrecipe (with this files) into image B. Seems it is not possible?
florian_kc has joined #yocto
florian_kc has quit [Ping timeout: 264 seconds]
<JPEW>
DvorkinDmitry: Is that 2 different nginx's or is the "custom nginx" the config?
<DvorkinDmitry>
JPEW, in the first image I have to keep default nginx config (from OE), in another image I need to replace the default nginx config with customer's. Nginx binaries in both images are identical.
<JPEW>
ah. We override the base nginx to use drop files, then you can easily add in other config fragments from other recipes as necessary
goliath has joined #yocto
florian_kc has joined #yocto
<DvorkinDmitry>
JPEW, ok. I have a recipe that installs several php files into /var/www/localhost/html/ but index.html and the directory /var/www/localhost/html conflicts with the files installed by nginx*.bb. If I'll place it into another dir I have to change default nginx.conf to point to another dir anyway. right?
<JPEW>
ya
<DvorkinDmitry>
JPEW, and I run into conflict again...
<DvorkinDmitry>
I can't even replace the nginx setup in my .bbappend, case it is image-specific, not machine specific
<JPEW>
Ya, then I'd just remove it completely in a bbappend and make new recipe(s) to provide it
<DvorkinDmitry>
JPEW, probably it is the best idea I can imagine too. was thinking about it, but hoped there are some magical key exists for such a situation in OE
<JPEW>
DvorkinDmitry: Ya, that one is a little tricky
<Ablu>
DvorkinDmitry: Can you maybe do your config in a file under "/etc/nginx/conf.d/*.conf" or "/etc/nginx/sites-enabled/*"? /etc/nginx/nginx.conf includes those.
<DvorkinDmitry>
Ablu, I can. but I can't override default webserver settings for localhost,_default and * servernames then
<Ablu>
DvorkinDmitry: You could overwrite parts of it, but yeah, doing a clean cut may be better in some situations.
florian_kc has quit [Ping timeout: 272 seconds]
pbsds has quit [Ping timeout: 252 seconds]
pbsds has joined #yocto
<rburton>
sounds like patches welcome for the nginx recipe to package the configuration separately as a recommendation
<rburton>
so you can provide your own
<rburton>
easy
<rburton>
something like package daemon recommends daemon-config and has a subpackage daemon-default-config which rprovides daemon-config
<rburton>
then you can write your own package and install it explicitly, so the default config doesn't get pulled in
Kubu_work has joined #yocto
mvlad has quit [Remote host closed the connection]
<tlwoerner>
vvn: but i don't want SOME_SETTING:the-base-board, i want SOME_SETTING:a-specific-add-on-board
<tlwoerner>
maybe i just need to add my add-on boards to OVERRIDES or something
florian_kc has joined #yocto
<vvn>
tlwoerner: yep, either inherit a new machine config from the base board, or extend MACHINEOVERRIDES and update your recipes with a few :a-specific-add-on-board
Haxxa has quit [Quit: Haxxa flies away.]
Haxxa has joined #yocto
|Xagen has joined #yocto
Xagen has quit [Ping timeout: 246 seconds]
Piraty has quit [Quit: -]
Piraty has joined #yocto
alessioigor has quit [Remote host closed the connection]
tin_ has joined #yocto
leon-anavi has quit [Quit: Leaving]
vladest has quit [Ping timeout: 272 seconds]
Kubu_work has quit [Quit: Leaving.]
<RP>
tlwoerner: MACHINEOVERRIDES may help
|Xagen has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<zelgomer>
what does this mean? "foo-1.0-r0 do_prepare_recipe_sysroot: The sstate manifest for task 'bar:populate_sysroot' (multilib variant '') could not be found."
<zelgomer>
i sometimes get workspaces into this state, i don't know how, and i don't know how to recover from it except to delete and recreate
florian_kc has quit [Ping timeout: 272 seconds]
<abelloni>
JPEW: are you planning to merge meta-mingw master-next in master?
<JPEW>
abelloni: Ya, does it fix the problem?
<abelloni>
yes
<abelloni>
I started a build with master and I got the problem
<JPEW>
Done
<abelloni>
I then remembered I did my other builds with master-next :)
kscherer has quit [Quit: Konversation terminated!]