<thomasxweber>
mckoan: someone deletes the source from DL_DIR, SSTATE_DIR contains artifact from former build; so will the source file be downloaded?
leon-anavi has joined #yocto
<mckoan>
thomasxweber: in that case the task do_fetch won't be able to find the sources, therefore will download them again
<thomasxweber>
mckoan: thanks
florian has quit [Ping timeout: 245 seconds]
thomasxweber is now known as swirl
yannd has joined #yocto
<RP>
swirl: to be clear, if the sstate is present but the files in DL_DIR are not, you would have to rerun do_fetch to get them back. do_fetch would not run if everything needed was in sstate
<ak77>
while at it... how to remove files in DL_DIR that are not needed anymore ? e.g. no recipe references them anymore
<landgraf>
ak77: DL_DIR can be shared between machines/distro/etc
<ak77>
yes. ok. but during development, versions get bumped ... there are lefovers
<landgraf>
ak77: how do you know this "leftovers" are not used by different version? :)
florian has joined #yocto
guest20 has joined #yocto
<ak77>
ok. fair enough. another one, i have in memory that I tried to move DL_DIR, and updated it in local.conf, and it triggered rebuild. is that so ?
<ak77>
shouldn't be if, as RP said, sstate has everything it needs. ...
<ak77>
what if I move sstate ?
<ak77>
(black friday is near... :))
<RP>
ak77: were you using hash equivalence?
* ak77
scratches head...
<RP>
moving DL_DIR and SSTATE_DIR wouldn't matter. Losing a copy of the hash equivalence mapping in the build directory would
<guest20>
Can anyone please tell me how to resolve these issues
<guest20>
Problem 1: package packagegroup-core-standalone-sdk-target-1.0-r0.cortexa7t2hf_neon from oe-repo requires libstdc++-dev, but none of the providers can be installed
<guest20>
package space-monitor-1.0-r0.cortexa7t2hf_neon from oe-repo requires busybox-cron, but none of the providers can be installed
ehussain has quit [Ping timeout: 252 seconds]
aduskett has quit [Ping timeout: 252 seconds]
<rburton>
ak77: use find to delete files that haven't been accessed for six months, or something
<guest20>
rburton could you please help me with my question
aduskett has joined #yocto
prabhakalad has quit [Quit: Konversation terminated!]
prabhakalad has joined #yocto
sotaoverride has quit [Ping timeout: 260 seconds]
sotaoverride has joined #yocto
<Jones42>
Guest20: try to move step by step. find the providers of busybox cron, try to build them alone (bitbake busybox), check the logs in the "temp" directory for errors
<Jones42>
Guest20: 'find' and ''
<Jones42>
Guest20: 'find' and 'rgrep' (i like 'ack') are your friends here
<Jones42>
and while I'm already here: quick sanity check: Our customer messes with "IMAGE_INSTALL" in his layer.confs. There's no sane reason to do that, right?
<RP>
Jones42: it depends what you're trying to do but it would be a best practise, it is potentially very confusing
<Jones42>
RP: thanks for the "somewhat-confirmation" :-) - I'm tring to clean up a big project and there are a lot of cases where they just add the packages provided by the layer to IMAGE_INSTALL.
<rburton>
yeah that's literally one of the DO NOT DO THIS slides in my 'how to write a bsp' presentation this summer
<Jones42>
rburton: oh, great! I've seen the talk but must have forgotten that point.
<Jones42>
Good to have some "authority" I can refer to while I'm digging their whole structure over :-)
<Jones42>
also I've just figured out that their way of working also conflicts with core-image.bbclass...
<ak77>
mcfrisk: thank you for answers on "dobule lives". when is the next LTS expected?
<ak77>
"nice" product i am working on should be release in april 25 :)
<RP>
Jones42: unconditionally adding is insane. I wasn't sure what "messes with" meant
<Jones42>
RP: ah, sorry for not being more precise. I thought that IMAGE_* should not be modified by layer.conf in general. But I guess there's always exceptions to any rule
<rburton>
layer.conf shouldn't really touch anything apart from the layer setup and _maybe_ HOSTTOOLS
<RP>
Jones42: I'm a supported of letting people innovate and do interesting things but that kind of change which can be done better elsewhere doens't make sense
<RP>
rburton: please not HOSTTOOLS :(
<RP>
we need recipe specific host tools
<guest20>
Jones42 rburton thank you very much for your inputs
guest20 has quit [Quit: Client closed]
Daanct12 has quit [Quit: WeeChat 4.4.3]
jmiehe has joined #yocto
jmiehe has quit [Client Quit]
<LetoThe2nd>
long shot, but: is there a way, possibly through some variable, to reference a file in a layer in local.conf, without knowing the absolute position of the layer directory? Like pseudocode MAGIC_FILE = "${TOPDIF_meta-demo}/files/magic-file"
<LetoThe2nd>
kanavin: demoing a pipeline where such an asset is required. if the asset can be bundled into a "layer" like this, and then be set up using the usual layer setup tools, then I just think it is more convenient and reproducible than saying "place the absolute path on your machine into this variable"
<kanavin>
LetoThe2nd, but where and how is it used? Typically if something is used in a recipe, then you add it to SRC_URI, and place it next to the recipe file
sotaoverride has quit [Ping timeout: 260 seconds]
<LetoThe2nd>
kanavin: its used in a recipe, but an upstream one. so yeah, the next best way would be to use bbappends.
sotaoverride has joined #yocto
<kanavin>
RP: I'm rather enjoying sane rpm times again, this glibc-locale thing been bothering me for months
<rburton>
LetoThe2nd: so yesterday i had a colleague rant at me about how mender only supports LTS releases
<LetoThe2nd>
rburton: ya I can also tell you which colleague :-)
<LetoThe2nd>
rburton: updating master-next for him is on my to-do list
<rburton>
:)
<RP>
kanavin: I had noticed rpm was taking longer, nice find in fixing it!
rob_w has quit [Remote host closed the connection]
Saur_Home35 has quit [Quit: Client closed]
Saur_Home35 has joined #yocto
<Scorpi>
Hi, I want to add nodejs to the SDK. Am I right that I need to add nodejs-nativesdk the package list of the SDK?
<mcfrisk>
that enables running nodejs natively in SDK environment. "nodejs" would enable using target nodejs in SDK.
<RP>
Scorpi: nativesdk-nodejs too btw, nativesdk is a prefix
<Scorpi>
mcfrisk: I wand to be able to build nodejs related projects using npm in the SDK
<Scorpi>
RP: oops, thanks
<mcfrisk>
Scorpi: for target or for the SDK environment?
<Scorpi>
mcfrisk: for target
* RP
doesn't remember how node works for cross builds
<mcfrisk>
for building, it could be that the nativesdk-nodejs works. But there can be issues if something is architecture dependent
<Scorpi>
RP: this was my error (prefix, not suffix). Now I need to adjust the nodejs dependencies to provide the nativesdk part, too
swirl has quit [Quit: Client closed]
cyxae has joined #yocto
yannd has quit [Remote host closed the connection]
goliath has joined #yocto
ddee has joined #yocto
florian_kc has joined #yocto
jclsn has quit [Quit: WeeChat 4.4.3]
sotaoverride has quit [Ping timeout: 264 seconds]
sotaoverride has joined #yocto
jclsn has joined #yocto
<Scorpi>
mcfrisk: actually, the build produces binaries that use /usr/local/oe-sdk-hardcoded-buildpath/sysroots/x86_64-pokysdk-linux/lib/ld-linux-x86-64.so.2 as the loader
florian_kc has quit [Ping timeout: 252 seconds]
rob_w has joined #yocto
Xagen has joined #yocto
ddee has quit [Ping timeout: 256 seconds]
<RP>
Scorpi: if that isn't what you want, you're after something "cross-canadian" in our recipe namespace
<LetoThe2nd>
not sure if I'm misunderstanding things, but to me it sounds like overlayfs essentially is a dance around four directories, and not something one could lay over a full rootfs filesystem, right?
CrazyGecko has quit [Quit: Konversation terminated!]
<rburton>
zeddii RP: patch sent
florian_kc has joined #yocto
Saur_Home99 has joined #yocto
florian_kc has quit [Ping timeout: 252 seconds]
Saur_Home35 has quit [Ping timeout: 256 seconds]
<RP>
rburton: thanks
Saur_Home17 has joined #yocto
Saur_Home28 has joined #yocto
Saur_Home99 has quit [Ping timeout: 256 seconds]
Saur_Home17 has quit [Ping timeout: 256 seconds]
Saur_Home16 has joined #yocto
rob_w has quit [Ping timeout: 272 seconds]
Saur_Home28 has quit [Ping timeout: 256 seconds]
<khem>
mckoan:sysbench does decent job too especially on smaller systems, it has less deps
<mckoan>
khem: thank you again
rfuentess has quit [Remote host closed the connection]
<RP>
JPEW: I meant ask - the SPDX ID issue keeps corrupting builds when we change it. Are there manual dependencies we can add with SPDX functions to avoid this?
<JPEW>
SPDX 2 or 3?
<RP>
JPEW: 3
<JPEW>
RP: I thought we had fixed those, sorry. Is there a bugzilla or logs I can look at