camus has quit [Remote host closed the connection]
camus has joined #yocto
rob_w has joined #yocto
gsalazar has joined #yocto
amitk has joined #yocto
Schlumpf has joined #yocto
Wouter0100 has quit [Ping timeout: 258 seconds]
<kroon>
RP, the bitbake docs talks about the weak default value assignment operator, ??=, saying "the assignment is made at the end of the parsing process". To me that is not quite correct, a more correct description would be "a default value which the variable will expand to if no other value has been assigned". Does that sound right to you ?
<landgraf>
kroon: +1
<kroon>
at least to me, "end of the parsing process" sounds like after all .conf's and recipes have been parsed. but I can have an immediate expansion in the middle of a recipe, either via := or an "include ${foobar}" statement, or somet other way
<landgraf>
kroon: it was my impression as well when I was reading the doc before sending the patch. But I have qschulz's presentation bookmarked which has right example of usage :)
<kroon>
landgraf, oh yeah, it was your email that got me curious about the docs
<landgraf>
probably
Wouter0100 has joined #yocto
<kroon>
on a different topic, make does immediate key expansion, contrary to bitbake, but I suppose there are good reasons for that (?)
Wouter0100 has quit [Read error: Connection reset by peer]
Wouter0100 has joined #yocto
Tyaku has joined #yocto
olani has quit [Ping timeout: 258 seconds]
olani has joined #yocto
ThomasRoos has joined #yocto
florian has quit [Ping timeout: 258 seconds]
<landgraf>
kroon: magic ...
<qschulz>
oh no
<qschulz>
not updating my talk to include this
<qschulz>
RP: how do you feel about latest patch from amahnui[m] to add the obsolete header on the old docs? landgraf and kroon just used the old one probably without knowing and it drives me crazy how often this happens. Something I can help with? (except reviewing better in the first place :) )
<landgraf>
qschulz: old one is from google. it happens all the time
<qschulz>
it's "easy" now because the latest version is actually using the new override syntax so anything from www.yoctoproject.org/docs would actually show the old syntax (_ instead of :)
<kroon>
qschulz, what do you think of the ??= description ?
nemik has quit [Ping timeout: 246 seconds]
nemik has joined #yocto
<qschulz>
kroon: what I believe is that the whole operators/override syntax needs some rework to make it more accessible to beginners
<qschulz>
My talk was basically a way start this work and motivate myself, but I happened to switch jobs since then, barely working on Yocto anymore and too much work for my one man team (excuses, excuses)
camus has quit [Ping timeout: 258 seconds]
<qschulz>
kroon: but we're always open to contribution :)
camus has joined #yocto
<qschulz>
so just send a patch and we can discuss it there with other people :)
<qschulz>
you don't need my approval to send anything
<qschulz>
(i'm not even the maintainer :) )
<kroon>
qschulz, i dont mind fixing the documentation, but I thought I'd first check in here if I understood the ??= operator correctly
<kroon>
the parties who would know I assume would all be in here
florian has joined #yocto
<qschulz>
??= is basically "if not a single other "=" (+= =+ = := ?= =. .=) operator was used anywhere else in any of the files parsed for the recipe (.conf, .inc, .bbclass, .bb, .bbappend)
<qschulz>
and ONLY the last ??= being parsed will be used
<qschulz>
I do not know what happens if you have:
<qschulz>
X ??= foo
<qschulz>
Y := "${X}"
<qschulz>
X = bar
<qschulz>
I assume Y = foo and X = bar
<qschulz>
but haven't tested
nemik has quit [Ping timeout: 260 seconds]
nemik has joined #yocto
<landgraf>
qschulz:on kirkstone: Y="foo" X="bar"
<kroon>
qschulz, thats what happens, and what I also expected, but to me it doesn't match the documentation "the assignment is made at the end of the parsing process"
<qschulz>
kroon: I guess we can add something about the first := that mentions the variable
<kroon>
qschulz, I suggested "a default value which the variable will expand to if no other value has been assigned" above
<qschulz>
kroon: does not explain the := trick
<kroon>
qschulz, why not ? := is immedaite expansion
<qschulz>
kroon: of *another* variable
<kroon>
qschulz, I don't follow
<qschulz>
kroon: your sentence does not explain what happens when X ??= foo Y := "${X}" X = "bar"
<qschulz>
because we're talking about the effect of ??= operator on X
<qschulz>
but the effect relies on the behavior of other variables immediate expansion operator (:=)
<qschulz>
and where they are located in the parsing order
peoliye has quit [Quit: Client closed]
<qschulz>
kroon: I am very good at writing super complex sentences to cover all possible use cases, so much that they are unreadable :D
<qschulz>
so, we'll FOR SURE need examples to highlight this because I'm guessing whatever the sentence we'll come up with, it'll not be really clear to everybody, especially beginners
<qschulz>
also, I am not a native speaker
<qschulz>
and sometimes valid sentences are really hard to understand
<kroon>
qschulz, can you expand on what you mean with "but the effect relies on the behavior of other variables immediate expansion operator (:=)" ?
<RP>
qschulz: I would be happy to have that patch. Wasn't there an issue I replied about?
<kroon>
because I do think it is possible to write technical documentation that covers everything. the glibc manpages are very good at that
<RP>
kroon: I'd be ok with something like that. End of the parsing process isn't entirely correct
<qschulz>
kroon: you need to specify that the := operator of ANOTHER variable (Y) using THIS variable (X), will trigger ??= to be used/expanded even though there are other operators after in the parsing process
<qschulz>
or that := basically triggers "one" end of a parsing process for all involved variables
ptsneves has joined #yocto
ptsneves has quit [Client Quit]
<qschulz>
RP: you did highlight an issue on v16 (two writes, duplicating the content), but a v17 was sent the same day as your review
<kroon>
qschulz, that is what := does, it is immediate expansion, all of the right-hand-side will be expanded immediatley.
<qschulz>
RP: well, os.walk does not go into directories which are symlinks, but if we have symlinks of files, that's a different story
<jclsn[m]>
Is there a variable that holds the current architecture which is building? I need to manually write a do_compile and append the archtitecture
<jclsn[m]>
$KARCH I guess
<qschulz>
jclsn[m]: bitbake -e <some-recipe> would probably help you find it :)
florian_kc has joined #yocto
<jclsn[m]>
qschulz: I know there architecutre
<jclsn[m]>
I just want to automize building for arm or aarch64
<qschulz>
LetoThe2nd: I am no native speaker, and rostedt has years of experience doing this
<LetoThe2nd>
qschulz: of course i did NOT. I'm just the clown on the stage, I actually know a lot less than people think I do.
<LetoThe2nd>
qschulz: i believe in you! you can do 120 slides in 30 minutes!
<RP>
qschulz: only registered users can trigger
<RP>
qschulz: I triggered one
rburton has quit [Quit: Connection closed for inactivity]
<kroon>
qschulz, "for me, W ??= x followed by A := "${W}" would have made the W ??= x into W = x"
goliath has joined #yocto
<kroon>
qschulz, yes, that is tricky, which is why I tried to emphasize that ??= changes what the *expanded* value will be, if it is currently undefined
<kroon>
but it doesn't assign any value
<kroon>
in the common sense
<kroon>
qschulz, does that make sense to you ?
cmd has quit [Read error: Connection reset by peer]
cmd has joined #yocto
nemik has quit [Ping timeout: 256 seconds]
nemik has joined #yocto
nemik has quit [Ping timeout: 272 seconds]
nemik has joined #yocto
<mrybczyn[m]>
ernstp: the new image file has a bug when CVE_CHECK_REPORT_PATCHED != "1" and it was "1" before. In text mode only. Basically it takes the old result partial file, even if it does not correspond to the current build. I'm wondering how to solved it: either clean up partial files after build or write empty files if we have empty results
ThomasRoos has quit [Remote host closed the connection]
<mrybczyn[m]>
ernstp: for JSON to avoid that I created the index file
<jclsn[m]>
How is it possible that some libraries in the image are not landing in the sysroot of the SDK? I can use the library just fine, but it is not listed in the target manifest
<jclsn[m]>
I am building the SDK with bitbake image -c populate_sdk. From my understanding everything should be contained
<mcfrisk>
jclsn[m]: maybe the -dev package dependecy to real shared library package is removed somehow
<kriive>
I need to backport Go from master to dunfell. I already copied the necessary files and fixed the syntax. Is there a way to make some recipes build with the new Go and some other with the provided one?
<qschulz>
kroon: I understand with the example but that was something that was unexpected to me
<qschulz>
and also it's a bit too complex to be the only example
<qschulz>
I'm not sure what we should do with := operator examples
<kroon>
qschulz, the := operator examples are, although a little hard to follow, correct as far as I can tell. as for referencing ":=" under the "??="-section, I don't have a problem with that. I don't think people are expected to read the manual from start to finish ?
<jclsn[m]>
<mcfrisk> "jclsn: maybe the -dev package..." <- What is the dev packagae dependency? It is a development image, which requires all the core image packages
vladest has quit [Quit: vladest]
<qschulz>
kroon: yes. we expect people to read the whole thing
<qschulz>
it's just going to take them about 4 days without breaks
<qschulz>
:D
<qschulz>
(I did it once, to review the docs after the migration to Sphinx... never doing it ever again)
<qschulz>
kroon: the "difficult" part for me is that people will most likely look for something specific
<kroon>
qschulz, I've never read it from start to end. I use it to lookup specific things
<qschulz>
and this := thingy is a pretty big gotcha
<landgraf>
kroon: Having a comment that ":=" will be decribed below is good IMO
<landgraf>
sorry for nitpicking :)
<kroon>
landgraf, yes I agree, I'll add that
<Guest87>
anybody else having problems building ncurses on dunfell? Patch 3 is failing to apply on my setup
<qschulz>
Guest87: does it happen when doing a build from scratch (no tmp) with vanilla poky checked out in dunfell branch?
<Guest87>
qschulz: I will try that
<kroon>
landgraf, well, I don't wanna say "see below", in case someone decides to shuffle the sections in the future
<landgraf>
kroon: makes sense
<kroon>
landgraf, so I'll just add a comment "Immediate variable expansion"
<landgraf>
yup.
<kroon>
landgraf, don't think I can reference sections in code examples either
<landgraf>
reader should be smart enough to press Ctrl-F :)
<kroon>
*nod*
<qschulz>
kroon: can't in literal blocks in sphinx unfortunately
<qschulz>
(which is started with ::)
<kroon>
qschulz, ok. I will say, sphinx is a heck of a lot faster compared to what was there before, IIRC
<qschulz>
kroon: that's the consensus yes :)
agupta1 has joined #yocto
kroon has quit [Quit: Leaving]
agupta1 has quit [Ping timeout: 255 seconds]
<Guest87>
qschulz: I deleted my tmp directory, ncurses package still fails on dunfell, on commit 0ca0aec7aa3ad9bcb3f8936a7e876f34866f2f68
seninha has joined #yocto
GNUmoon has joined #yocto
<Guest87>
qschulz: nevermind it was related to a SRCREV_PN statement i had in a distro config
<qschulz>
Guest87: good you found the issue :)
agupta1 has joined #yocto
camus has quit [Remote host closed the connection]
camus has joined #yocto
agupta1 has quit [Quit: Konversation terminated!]
Schlumpf has quit [Quit: Client closed]
<jclsn[m]>
I have a really weird error I don't understand. I am setting a password with the extrausers class and am doing it exactly as in the documentation, but when I try to log in this password doesn't worrk. It just breaks the password
<jclsn[m]>
I just wanted to set some plain password. Is that not possible?
<qschulz>
nope
<qschulz>
not anymore
<jclsn[m]>
Since honister?
<jclsn[m]>
My colleague is on hardknott and can do it
<RP>
qschulz: the old docs are meant to redirect to the new. Can you collect up some example urls which aren't working and email halstead, cc me please
<jclsn[m]>
You are required to change your password immediately (administrator enforced).
<jclsn[m]>
Permission denied, please try again.
<jclsn[m]>
Trying to login via ssh if that makes a difference
camus1 has joined #yocto
camus has quit [Quit: camus]
camus1 is now known as camus
<jclsn[m]>
Well enough for today
<jclsn[m]>
Have a nice weekend guys!
kscherer has joined #yocto
Habbie has quit [Ping timeout: 260 seconds]
<jclsn[m]>
I think something is wrong in the shadow file
dz1 has quit [Quit: Leaving]
Tokamak has quit [Ping timeout: 246 seconds]
Tokamak has joined #yocto
Habbie has joined #yocto
pgowda_ has quit [Quit: Connection closed for inactivity]
Tokamak_ has joined #yocto
Tokamak has quit [Ping timeout: 256 seconds]
goliath has quit [Quit: SIGSEGV]
<sotaoverride>
how do I use a single recipe to insall multiple systemd services. What would the SYSTEMD_SERVICE:${PN} = "......" syntax look like in such a scenario?
<ptsneves>
if they are all in the main package you just add the services to that variable
<sotaoverride>
ptsneves so like so like SYSTEMD_SERVICE_${PN} = "var.mount1 var.mount2 var.mount3" or comma seperated. blanking out here a little with the syntax. Do i leave a spave after the last service name..
<sotaoverride>
?
<sotaoverride>
ptsneves:
<sotaoverride>
ptsneves: no commas I guess. I found this example, just no commas I guess I found this example meta-openembedded/meta-oe/recipes-extended/cfengine/cfengine_3.15.0.bb:SYSTEMD_SERVICE_${PN} = "cfengine3.service cf-apache.service cf-hub.service cf-postgres.service
<ptsneves>
no commas just spaces
<ptsneves>
yep, grepping the source is a good way to find these details
<ptsneves>
qschulz nowhere does it state it is a space separated list, or am i missing something
<qschulz>
ptsneves: not mentioned and could/should be
<qschulz>
probably in the variable definition though and not the class
<qschulz>
or let's say, the most important place where this needs to be defined is in the variable glossary
<ptsneves>
do you mean here?
<ptsneves>
by the way sotaoverride in yocto the convention is to space separate. Not great but that is how it is
<ptsneves>
which means do not create file names with spaces :D We are in the 90s still :)
<qschulz>
we have a few occurrences in the docs of "space-separated list" so one more won't hurt :)
<ptsneves>
sorry when i said here i forgot to paste the link. Yes it does not mention it. I could try to create a patch but i dont even know where the docs are anymore nor the mailing list. :(
<qschulz>
it's even more than "it won't hurt", we should actually fix it because it's not mentioned it's possible to have more than one element in there
<qschulz>
ptsneves: yocto-docs is the git repo
<qschulz>
docs at lists.yoctoproject.org is the ML
Minvera has quit [Remote host closed the connection]
Tokamak_ has quit [Ping timeout: 255 seconds]
Tokamak has joined #yocto
<halstead>
kanavin: I was wrong, it finished green.
clifonlintic[m] has joined #yocto
<yates>
abelloni: because my customer wants it?
<abelloni>
yates: very good reason ;)
<abelloni>
I guess you'll need a commercial solution
GillesM has joined #yocto
GillesM has quit [Remote host closed the connection]
kevinrowland has quit [Ping timeout: 252 seconds]
Vonter has joined #yocto
<manuel>
I would like to send a patch for a file in poky. The file has a different content on dunfell and master branch. How do I specify which branch my patch is based on? I'd like to see it on dunfell branch eventually.