Refactor schemaVersion to distinguish between 2 and 3, and drop support for 1 #531
Conversation
Signed-off-by: Marcello Seri <[email protected]>
planex/spec.py
Outdated
| spec.add_archive(idx, value["URL"], link.path, | ||
| value["patches"]) | ||
| if not value: | ||
| spec.rm_patch(idx) |
There was a problem hiding this comment.
Why does rm_patch match with add_archive here (instead of rm_archive)? If that's what is intended could you please add a comment explaining why these two go together.
There was a problem hiding this comment.
Sorry I missed this comment. I replied below. I am happy to add a comment, but I wanted first to discuss it and decide if we want it at all.
planex/spec.py
Outdated
| for name, value in link.patch_sources.items(): | ||
| idx = _parse_name(name) | ||
| if not value: | ||
| spec.rm_patch(idx) |
There was a problem hiding this comment.
Why is rm_patch the opposite of add_archive?
There was a problem hiding this comment.
Because archive is not really rewriting a patch, it's overriding where we get it from. Patch entries add the archive to override the patches, adding another patch allows us to override the archive, but the only way to delete the patch is to remove the patch entry.
Did I understand it incorrectly?
|
Removing sources could be dangerous. Sources are usually unpacked by specific macros in the Removing patches should be ok if I don't immediately see the use of removing archives and patchqueues, since those are only added by links so you could just not add them in the first place. I guess this could be useful if you were stacking a pin over a link, but at present the pin completely replaces the link. |
|
Yes, I thought the idea was allow to stack a pin over a link. I think it's good to split this PR in two. One dealing with the link refactor (this PR) and a future one, if needed, that adds the removal feature. It is such a tiny amount of effor that it's not a big deal to rentroduce it later only if we find the necessity |
Signed-off-by: Marcello Seri <[email protected]>
Signed-off-by: Marcello Seri <[email protected]>
Signed-off-by: Marcello Seri <[email protected]>
Signed-off-by: Marcello Seri <[email protected]>
This should solve #503 and #515.
This PR also introduces a way to delete a source, a patchqueue or a patch by providing an empty record in the link/pin file. I have created #529 to track that as I recall that issue being discussed at a certain point. I am happy to remove the relevant commits if we decide that this is not necessary.EDIT: Removed, it will be part of a separate PR if we decide to stack pins over links