I appreciate your mail, Collin. For the rest of the readership I want to make a 
clarification that the subject line doesn't do justice on: this change has been 
effected through the newly-introduced coreutils-from *source* package. The 
coreutils source package has not made any changes made to it whatsoever. Your 
statement here seems accurate:
> The maintainer of the "coreutils" [source] Debian package said that he would 
> not make this change without discussion on debian-devel, which I feel is 
> reasonable.

To put it another way, the new src:coreutils-from source package which was 
accepted into Debian experimental includes an actual binary transitional 
package named 'coreutils' exactly. If this were to be uploaded to the unstable 
suite, then whichever binary package has the greater version number will "win", 
with the other becoming cruft.

I am concerned that this *looks* like hijacking of a binary package, by using a 
new source package as a vessel for a coreutils=9.7-999+0.0.0 binary package 
that supersedes the preexisting one coming from a different maintainer. I hope 
to find reassurance that this was an orderly handover that the src:coreutils 
maintainer consents to.

Regardless of that, the current src:coreutils-from and coreutils=9.7-999+0.0.0 
package revision in experimental seems to infringe on this Debian Policy 
requirement: https://www.debian.org/doc/debian-policy/ch-binary.html#base-system
> You must not tag any packages essential before this has been discussed on the 
> debian-devel mailing list and a consensus about doing that has been reached.
With the understanding that coreutils from src:coreutils-from is effectively a 
new package, and one which indeed has Essential: yes marked in its control 
file, this looks like a discrepancy. If Michael Stone for src:coreutils 
consented, though, then this would merely be part of the coreutils maintenance 
effort and not have any consequence. If the long-term plan is to detach 
coreutils from the src:coreutils package, then using an epoch for the binary 
package would be a good idea, so that the transitional package always 
supercedes the traditional binary package.

Attachment: signature.asc
Description: This is a digitally signed message part

Reply via email to