Skip to contentSkip to conversation

license archiving is failed to tmp/deploy/licenses

scarthgap Livius 3 participants 7 messages Jul 17, 2024 – Jul 30, 2024 1 message not in the archive

lore
Share & tools
lore

Livius

author
#0 lore raw

License file archiving is broken, package/recipe folders are fully empty in tmp/deploy/licenses directory after the finished image build, manifest files also missing in it.

It seems to me this is broken for me since the following patch:
https://github.com/openembedded/openembedded-core/commit/1a4ab9fc26659507e678e87312b514e8ea515673

This is my settings for it in my local.conf. What goes wrong? INHERIT += "archiver"
ARCHIVER_MODE[src] = "patched" ARCHIVER_MODE[compression] = "bz2" ARCHIVER_MODE[mirror] = "combined" COPYLEFT_RECIPE_TYPES = "target" COPYLEFT_AVAILABLE_RECIPE_TYPES = "target"

# Don't filter the license
COPYLEFT_LICENSE_INCLUDE = ""
COPYLEFT_LICENSE_EXCLUDE = ""

# Deploy licenses
COPY_LIC_MANIFEST = "1"
COPY_LIC_DIRS = "1"
LICENSE_CREATE_PACKAGE = "1"

#1 lore raw
Quoted 13 lines · Livius via lists.openembedded.org

License file archiving is broken, package/recipe folders are fully empty in tmp/deploy/licenses directory after the finished image build, manifest files also missing in it.

It seems to me this is broken for me since the following patch:
https://github.com/openembedded/openembedded-core/commit/1a4ab9fc26659507e678e87312b514e8ea515673

This is my settings for it in my local.conf. What goes wrong? INHERIT += "archiver"
ARCHIVER_MODE[src] = "patched" ARCHIVER_MODE[compression] = "bz2" ARCHIVER_MODE[mirror] = "combined" COPYLEFT_RECIPE_TYPES = "target" COPYLEFT_AVAILABLE_RECIPE_TYPES = "target"

Quoted 8 lines

# Don't filter the license
COPYLEFT_LICENSE_INCLUDE = ""
COPYLEFT_LICENSE_EXCLUDE = ""

# Deploy licenses
COPY_LIC_MANIFEST = "1"
COPY_LIC_DIRS = "1"
LICENSE_CREATE_PACKAGE = "1"

I wanted to say yes, I have seen this report and yes, I'm aware I wrote the commit seeming to cause it. No, I'm afraid I don't know what is wrong.

The mailing list is the right place to discuss things (or the bugzilla), not github issues.

My patch added in a directory sublevel to the structure and it would seem that some piece of code wasn't translated properly. The change likely needs to be debugged with the settings you're using to find the issue with the change.

I did make best efforts to ensure everything worked but our test coverage isn't always 100% and it would seem we're missing some test case.

Cheers,

Richard

author
#2 lore raw

https://github.com/openembedded/openembedded-core/blob/master/meta/classes-global/license.bbclass#L418
In line 418 if i use ${SSTATE_PKGARCH} in path of do_populate_lic[sstate-inputdirs], deploying of licenses can start to work but there will be many warnings and at end of it a fatal error will stop the bitbake building. do_populate_lic[sstate-inputdirs] = "${LICSSTATEDIR}/${SSTATE_PKGARCH}"

Warnings:
...
WARNING: prambo-customer-image-1.0-r0 do_rootfs: QA Issue: The license listed BSD-2-Clause was not in the licenses collected for recipe util-linux [license-file-missing]
WARNING: prambo-customer-image-1.0-r0 do_rootfs: QA Issue: The license listed BSD-3-Clause was not in the licenses collected for recipe util-linux [license-file-missing]
WARNING: prambo-customer-image-1.0-r0 do_rootfs: QA Issue: The license listed BSD-4-Clause was not in the licenses collected for recipe util-linux [license-file-missing]
WARNING: prambo-customer-image-1.0-r0 do_rootfs: QA Issue: The license listed MIT was not in the licenses collected for recipe util-linux [license-file-missing]
WARNING: prambo-customer-image-1.0-r0 do_rootfs: QA Issue: The license listed MIT was not in the licenses collected for recipe volatile-binds [license-file-missing]
WARNING: prambo-customer-image-1.0-r0 do_rootfs: QA Issue: The license listed BSD-3-Clause was not in the licenses collected for recipe zip [license-file-missing]
WARNING: prambo-customer-image-1.0-r0 do_rootfs: QA Issue: The license listed Zlib was not in the licenses collected for recipe zlib [license-file-missing]
...

Error:
do_rootfs: Couldn't find license information for dependency alsa-conf

As i see, alsa-conf is not exist in x86_64 licenses this is why it is failed all the time.

#4 lore raw
Quoted 6 lines · Livius via lists.openembedded.org

it's been half a year since it is reported in bugzilla issue ticket:
https://bugzilla.yoctoproject.org/show_bug.cgi?id=15394

When it will be fixed? At the moment Scarthgap LTS is not able to use for any "production" because it has this bug and we can not generate the full manifest and licenses in our SD/eMMC card image build process of end-user products FOSS docs in our company.

Can you propose a patch?

Alex

author
#5 lore raw

No, it is not in my scope. I am not the responsible or mainntainer for it, and i do not understand the latest patch what it like to do and what should it resolved. I can not solve it, sorry. It was reported a half year ago.

Please if it is not working as before, do not commit it, rather revert this patch which cause a totally buggy license deploying at the moment in the latest Scarthgap LTS. LTS branch shall not contains this kind of quality, in normally.

A message that is not in the archive (sent off-list or not yet received): [email protected]

#7 lore raw
Quoted 4 lines · Livius via lists.openembedded.org

No, it is not in my scope. I am not the responsible or mainntainer for it, and i do not understand the latest patch what it like to do and what should it resolved. I can not solve it, sorry. It was reported a half year ago, it should be resolved by a real maintainer/developer.

Please, if it is not working as before, do not commit it. Rather revert this patch which cause a totally buggy license deploying at the moment in the latest Scarthgap LTS. LTS branch shall not contains this kind of quality, in normally.

Calibrating expectations will serve you well.

Yocto is an open source project, and not your commercial supplier. You can take what there is without any fees whatsoever, put it into a product, sell the product, and keep all of the income the product generates. How maintainers and contributors actually make a living and find time to fulfill their commitments is not your problem.

Here's the other side of the deal: nobody owes you a bugfix for a problem you're having. Not only that, no one owes you an investigation into your problem either. We're all volunteers; if your problem isn't particularly interesting, relevant or common, there are always far more important issues to us. Especially if you act entitled and demanding.

If no one reacts to your issues, the correct thing to do is one of these:

- own the problem. Study the code, study the history of changes, ask questions about those things if you have them. Then work your way towards a fix and a test for it. This can take significant time, which is not going towards adding value to your product.

- hire someone to do this for you. You can hire me, for example.

Alex