Re: okteta

[Greg Troxel <gdt%lexort.com@localhost>]

Jonathan Perkin <jperkin%mnx.io@localhost> writes:

>>So pkgin is using a build timestamp for sorting?
>><quoterant>not sure what "newer" means</>.

So my quote rant was fair!

> The thing that separates pkgsrc from other package managers is that we
> key on PKGPATH.  pkgin's job (at least for update) is basically:
>
>   Given a set of installed packages, give me whatever the current
>   state    of the remote pkgsrc repository is for those packages.
>
> Whether the version goes backwards, forwards, or stays the same but
> the package has been rebuilt, the end result is the same.  You get the
> version from the remote repository so that your installed package set
> is fully in sync.

So we have established that pkgin simply moves to the version in the
repo, and thus for workflows using pkgin, there is no reason to use a
fake version.

The other questions are

  pbulk update builds
  pkg_chk and pkg_rr

I am pretty sure pkg_chk will report a mismatch, and thus pkg_rr will
rebuild.  I am also pretty sure pbulk will simply say "modified since"
and rebuild.

Therefore, it seems we have arrived at "there is no problem with an
occasional instance of a version number going backwards".