FS#13414 - [repo-add] restrict readding the same package to the repo db

Attached to Project: Pacman
Opened by Allan McRae (Allan) - Sunday, 22 February 2009, 10:28 GMT
Last edited by Xavier (shining) - Monday, 19 October 2009, 11:03 GMT
Task Type Feature Request
Category Scripts & Tools
Status Closed
Assigned To Xavier (shining)
Architecture All
Severity Low
Priority Normal
Reported Version 3.2.1
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

It might be a good idea to restrict the re-addition of a package to a repo db without a flag of some variety. Or at least give a warning.

This would prevent issues like this:
http://bbs.archlinux.org/viewtopic.php?id=65098
This task depends upon

Closed by  Xavier (shining)
Monday, 19 October 2009, 11:03 GMT
Reason for closing:  Implemented
Additional comments about closing:  added a warning with commit f09f82ee59531
Comment by Dan McGee (toofishes) - Sunday, 22 February 2009, 16:26 GMT
I'm not sure what you mean here. When we add a package with the same name, the old one should always be removed first, whether the version differs or not.
Comment by Xavier (shining) - Sunday, 22 February 2009, 17:07 GMT
The weird thing is to replace a package with the same version but a different md5sum. This is what happened here and this should clearly be avoided.

However, there might be cases where we want to re-add exactly the same version of a package (same md5sum), to exploit a fixed or improved repo-add.
This is not too common though, so printing a warning when the same version is added sounds alright.
Comment by Allan McRae (Allan) - Sunday, 22 February 2009, 20:37 GMT
Basically, the idea was I add foo-1.0-1 to my db and then someone downloads that db. Then I re-add a different foo-1.0-1 to the db. Now, if someone had downloaded the db between these two additions and now does a "pacman -S foo", they get a corrupt package.

I think repo-add should not allow the readdition of the same package without a flag to indicate that this is a known action, or at least a warning.
Comment by Dan McGee (toofishes) - Tuesday, 24 February 2009, 02:46 GMT
I'd be more for a warning rather than a flag, only because a flag would break any existing scripts that don't expect required flags to repo-add.
Comment by xduugu (xduugu) - Monday, 19 October 2009, 10:21 GMT
The warning was added in pacman 3.3, so can this be closed?
Comment by Xavier (shining) - Monday, 19 October 2009, 11:03 GMT
Yes, thanks for the reminder :)

Loading...