<rob_w>
in scarthgap ... i am having trouble to get curl with openssl, openssl is builded and in my image but curl seems to not see/find it ? what is a proper way to enable this ?
<mcfrisk>
openssl works correctly with upstream meta layers like poky. Something in your setup breaks it. I'd compare "bitbake -e openssl" output from plain poky targets to your custom build target output. Same for curl.
ptsneves has joined #yocto
sotaoverride has quit [Ping timeout: 252 seconds]
sotaoverride has joined #yocto
<rob_w>
ok ..
Jones42_ is now known as Jones42
olani- has quit [Ping timeout: 260 seconds]
<mcfrisk>
rob_w: sometimes simple bbappends which try to change e.g. PACKAGECONFIG or DISTRO_FEATURES end up breaking things in odd ways
<rob_w>
i guess openssl is not build in a way to have others see it .. as i try to overrride curls PACKAGECONFIG but it doesnt see the proper headers etc
<mcfrisk>
rob_w: sounds like a possible root cause
<rob_w>
seems i am missing openssl headers and libs in sysroot ?? why
<mcfrisk>
rob_w: DEPENDS does not Clist openssl
<rob_w>
sorry , mcfrisk what do you mean by that
<mcfrisk>
I'd compare "bitbake -e" output of the recipes
<mcfrisk>
build sysroot of a recipe includes only binaries from recipes which are in DEPENDS
<mcfrisk>
rob_w: for example, curl recipe on poky sets PACKAGECONFIG with ??= which means that any changes to it via += will overwrite it completely
<mcfrisk>
which then also impacts curl recipes DEPENDS variable, e.g. openssl will be missing and not installed to curl sysroot. These are visible in bitbake -e output which can be diff'ed from pure poky vs your build config
<rob_w>
hmmm .. but i hardly change yocto or poky with my layers ..
<mcfrisk>
rob_w: yet things like curl and openssl are broken. please do the "bitbake -e" comparison to find out which variables are set differently. Sometimes things don't behave the way one expects.
<rob_w>
ok will try
Ad0 has quit [Ping timeout: 252 seconds]
aduskett has quit [Quit: Konversation terminated!]
aduskett has joined #yocto
aduskett has quit [Remote host closed the connection]
aduskett has joined #yocto
<rob_w>
help me here , only when a recipe has a DEPEND on a other recipe , only then will the target recipe tmp/work/*/recipe/sysroot* show the headers or libs of the DEPEND package ?
Ad0 has joined #yocto
<mcfrisk>
rob_w: yes
<rob_w>
thx
<rob_w>
so indenpendant from any PACKAGECONFIG, if i add DEPEND openssl, i should actaully those headers ?
<mcfrisk>
and DEPENDS can change from direct operations or indirectly also through PACKAGECONFIG, DISTRO_CONFIG etc
<rob_w>
see^
<rob_w>
hmm oh ok
<mcfrisk>
PACKAGECONFIG "openssl" adds the configure snippets to enable openssl support and also adds "openssl" to DEPENDS
<rob_w>
which i have or curl does have by nature
Saur_Home has quit [Quit: Client closed]
<mcfrisk>
yes. so something changed PACKAGECONFIG and removed "openssl" from there
<rob_w>
could this be a host issue ? like some -native thing missing ?
Saur_Home has joined #yocto
<mcfrisk>
possibly, but most build tools in yocto use openssl-native recipe to provide openssl related things so host openssl version rarely impacts build output
<rob_w>
right thx
Saur_Home has quit [Quit: Client closed]
Saur_Home has joined #yocto
aduskett has quit [Ping timeout: 276 seconds]
fray has quit [Ping timeout: 252 seconds]
aduskett has joined #yocto
sotaoverride has quit [Ping timeout: 255 seconds]
prabhakalad has quit [Remote host closed the connection]
<mcfrisk>
rob_w: "bitbake -e curl" from both environmens, then diff them. build tree tmp changes can be ignored. Check for DEPENDS and PACKAGECONFIG. If you have a working and broken custom build config, then capture bitbake -e output from both and check differences. That will explain if layer, distro, machine configs or bbappends broke things.
<rob_w>
did that but as my paths are different , the whole thing comes up changed in the diff
<mcfrisk>
rob_w: check for DEPENDS and PACKAGECONFIG variables in the diff
<rob_w>
...checking
<mcfrisk>
really good thing about the bitbake -e output is that it explains how the content of variable was formed
<rob_w>
yeah ,, i see that , its huge ouptput but it should bring me something ... has to.
<rob_w>
is this a thing saying --without-ssl if all 3 of the PACKAGECONFIGS are set ?
sotaoverride has quit [Ping timeout: 248 seconds]
<mcfrisk>
I think the opposite, if none of "gnutls mbedtls openssl" are not in PACKAGECONGIG, then --without-ssl is used
<rob_w>
right, it disables ssl
<Saur>
rob_w: See commit 6c737396c705e1388aff5f5a599c901a1a1760a2 (or 083623bdd356daa3e5000f261464b73ec8243f91 in poky) for the explanation to why it is needed.
sotaoverride has joined #yocto
<mcfrisk>
thus what is removing "openssl" or overwring PACKAGECONFIG. "openssl" is the default in curl recipe in poky
pbiel has quit [Quit: Leaving]
<rob_w>
yeah thx Saur
<rob_w>
will get there .. i kicked my server to rebuild from scratch and i go from there and see which adds what when i enable my layers , while checking -e outputs ... thx guys ... its a bummer and i bet its a stupid old option i carry around unknowingly
paulg has quit [Ping timeout: 252 seconds]
<rob_w>
i once saw some license magic ... like building pure gpl images etc ,, could this be a source of my issue too
<rob_w>
but then i need to read up on that license stuff
<rob_w>
i need to run to my homeoffice now ,.. thx and cya
rob_w has quit [Quit: Leaving]
cglab has joined #yocto
Saur_Home has quit [Quit: Client closed]
Saur_Home has joined #yocto
roussinm has joined #yocto
sakoman has joined #yocto
rob_w has joined #yocto
|Xagen has quit [Ping timeout: 276 seconds]
Daanct12 has quit [Quit: WeeChat 4.4.3]
yannd has quit [Remote host closed the connection]
cyxae has joined #yocto
cyxae has quit [Client Quit]
cyxae has joined #yocto
vthor has joined #yocto
vthor has joined #yocto
vthor has quit [Changing host]
florian_kc has joined #yocto
tlwoerner has quit [Ping timeout: 255 seconds]
<usvi>
rburton: thank you for your discussion earlier about the pitfalls of make. it lead me down quite the rabbit hole. I discovered also that basically all our makefiles are total garbage. I'm now fixing them
tlwoerner has joined #yocto
<rburton>
usvi: don't feel bad: most are :)
<tgamblin>
rburton: did you have a chance to look at my latest?
* tgamblin
is still hacking away at it
<tgamblin>
I'm now trying to use the full `nativepython3 -m installer ...` call from the pep517 install function inside do_install:append() to see if that solves it... still can't quite figure out why the buildpaths are sneaking in again after supposedly recompiling __config__.py
<tgamblin>
I wonder if it's because they're getting built into the .whl before I can strip them out
Saur_Home26 has joined #yocto
sotaoverride has quit [Ping timeout: 265 seconds]
sotaoverride has joined #yocto
Saur_Home has quit [Ping timeout: 256 seconds]
goliath has quit [Quit: SIGSEGV]
xmn has joined #yocto
<tgamblin>
doing it in do_compile:append() doesn't seem to help, so I think it's something to do with the mesonpy/pep517 classes
<tgamblin>
and/or some variables that need to be set in the recipe...
<usvi>
yea. I started to investigate why TARGET_CPPFLAGS did not work. it was the garbage makefile ignoring it. and yes, looking at "helpful" stackoverflow posts about how to create makefiles just makes me want to puke now
Xagen has joined #yocto
frieder has quit [Remote host closed the connection]
Saur_Home29 has joined #yocto
sakman has quit [Read error: Connection reset by peer]
<tgamblin>
it does look like maybe some of the -fdebug-prefix-map (and similar) flags aren't set up correctly...
goliath has joined #yocto
Xagen has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Xagen has joined #yocto
mjm has quit [Ping timeout: 252 seconds]
mjm has joined #yocto
rob_w has quit [Quit: Leaving]
rob_w has joined #yocto
Minvera has quit [Remote host closed the connection]
<rob_w>
mcfrisk: .. found it , surely in side my layers, artifact back from my rocko layers when i hacked ssl there, just overlooked it 20 times ... thx again for the hints n help
florian_kc has joined #yocto
rfuentess has quit [Remote host closed the connection]
<rburton>
tgamblin: its that the pyc contains the full path to the source
<rburton>
but installer writes a path with the install prefix stripped, as it should
sotaoverride has quit [Ping timeout: 246 seconds]
<tgamblin>
rburton: right. I'm not sure how to solve that. It's probably a simple fix, but I'm probably overthinking it now that I've looked at it so much
<rburton>
pycompile's main entry point doesn't let you fix it, but you can hit it with sticks
<rburton>
one minute :)
sotaoverride has joined #yocto
<rburton>
aha
<JPEW>
moto-timo, tlwoerner: Any idea why the github actions for bmaptool stopped running?
mckoan is now known as mckoan|away
goliath has quit [Quit: SIGSEGV]
<rburton>
tgamblin: change the compile_all line to nativepython3 -mcompileall -s ${D} ${D}${PYTHON_SITEPACKAGES_DIR}/numpy/__config__.py
<rburton>
compileall is a better version of py_compile and importantly has a strip-prefix option
<rburton>
tgamblin: the py_compile line, that is
Xagen has joined #yocto
<tgamblin>
rburton: that seems to work :) I'll run a reproducibility test and a buildall-qemu, and if both of those clear then I'll submit a patch
<rfs613>
per the bitbake manual I tried "bitbake-dumpsigs" but I only seem to have "bitbake-dumpsig" (without the final "s"), and it seems to require a bunch of arguments.
<tgamblin>
rburton: No luck :( ./usr/lib/python3.13/site-packages/numpy-2.1.3.dist-info/RECORD still seems to have different shasums for __config__.py. I'll come back and look at it again soon
rob_w has quit [Remote host closed the connection]
leon-anavi has quit [Quit: Leaving]
ederibaucourt has quit [Quit: ZNC 1.8.2 - https://znc.in]
<rfs613>
setscene seems to just piggyback on regular task ordering (DEPENDS), or is there another mechanism that i'm missing?
florian_kc has quit [Ping timeout: 252 seconds]
Saur_Home84 has joined #yocto
ederibaucourt has joined #yocto
Saur_Home29 has quit [Ping timeout: 256 seconds]
PhoenixMage has quit [Ping timeout: 252 seconds]
PhoenixMage has joined #yocto
ederibaucourt has quit [Ping timeout: 246 seconds]
Saur_Home37 has joined #yocto
Saur_Home84 has quit [Ping timeout: 256 seconds]
sotaoverride has quit [Ping timeout: 248 seconds]
sotaoverride has joined #yocto
PhoenixMage has quit [Ping timeout: 244 seconds]
ederibaucourt has joined #yocto
PhoenixMage has joined #yocto
Saur_Home84 has joined #yocto
Saur_Home37 has quit [Ping timeout: 256 seconds]
robwoolley has quit [Quit: Client closed]
Saur_Home82 has joined #yocto
Saur_Home84 has quit [Quit: Client closed]
<ak77>
huh. I stumble into all possible bugs... have to get into sending patches game. now i discovered why my rust executable isn't in a package, it's name ends with examples
<ak77>
(* executable built with rust, if name ends with executable, system treats it as examples directory)
<ak77>
(** argh, ends with "examples")
florian_kc has joined #yocto
Jones42 has joined #yocto
<ak77>
is this patch worthy ?
<rburton>
ak77: if you have a neat fix yes
<JPEW>
moto-timo, tlwoerner: Ugh... signing support in bmaptool is just really broken because it will now *only* work with the distro gpg python module (well, at least in CI since we can't test with any arbitrary python version anymore)
cyxae has quit [Remote host closed the connection]
cyxae has joined #yocto
goliath has joined #yocto
hexbrex has joined #yocto
geoff__ has quit [Read error: Connection reset by peer]
geoff__ has joined #yocto
sotaoverride has quit [Ping timeout: 260 seconds]
sotaoverride has joined #yocto
druppy has joined #yocto
ptatrai has joined #yocto
ptatrai has quit [Remote host closed the connection]
<RP>
ak77: the npm update class sounds interesting. I'd love to see the npm stuff in core tidied up with tests...