Re: Orphaned packages with very low popcon numbers
Michael Stone <mstone@debian.org> writes:
> No, what is unsubstantiated is your characterization of what is
> happening. One of things that happens *every time* this sort of
> discussion comes up is that someone starts creating straw men along
> the lines of "what if some important package gets removed because of
> some arbitrary criteria; it would be horrible, so we should complain a
> lot about the criteria and then do nothing".
Yes. Deleting information is a big step. We *should* be extremely
conservative about deleting information. If we make a mistake and
fail to delete a package that should have been deleted, it entails
little cost, and we can rectify the mistake at any time very easily,
even years later.
By contrast, if a package is deleted that shouldn't be deleted, it
costs a lot of work to reverse, and becomes harder and harder to
reverse the more time has gone by.
I was myself bit by this, when the gnome maintainers decided that
gnome-1 was obsolete, and started deleting things, despite there being
packages (such as gnucash) which depended on the libraries in question.
> And yet, a list of
> packages was generated and comments about the list were solicitied. If
> the arbitrary criteria were used to silently and automatically remove
> packages then the "every package is sacred" crowd would have a leg to
> stand on with that line of reasoning--as it is, it's silly.
Um, I didn't object to that. All I did was say, "let's be sure to
give individualized attention."
The process did not, in fact, say that there would be such. All it
said was that "if there is no reason to keep them" that packages
meeting a criterion would be dropped. This suggests no more process
to me than posting a list and seeing if anybody squeaks, rather than
the actual individualized examination that is called for.
Thomas
Reply to: