Re: libc6_2.0.7 release notes...

[Philip Hands <phil@hands.com>]

> When properly used epochs do not hang around forever. Consider the
> situation where epochs are supposed to be used:
> 
> Upstream           Debian
> 
> 1.0                  1.0
> 2.0                  2.0
> 3.0                  3.0
> 2.0                1:2.0
> 3.0                1:3.0
> 4.0                  4.0

Dong!  You loose ;-)  [as has already been pointed out]

 1:3.0 > 4.0

> If we apply epochs to the problem of pre-release version numbering (with
> my proposal along side) you should be able to see why I don't like it.
> 
> Upstream          Your Proposal              My Proposal
> 
> 2.0.8pre1            2.0.8pre1               2.0.7.99.1
> 2.0.8              1:2.0.8                   2.0.8
> 2.0.9pre1          1:2.0.9pre1               2.0.8.99.1
> 2.0.9              2:2.0.9                   2.0.9
> 
> As you can see, for every point release, the epoch number must increase.
> This presents this problem as an infinitely folded list of repeating
> version numbers, which is not actually the case.

I don't think anyone was suggesting this.  What was being suggested, was that 
where a mistake is made (i.e. the use of a ``pre'' version in the first 
place), the right way to recover from it (in the absence of time-travel.deb) 
was to use an epoch, so the sequence goes:

  Upstream          Debian

  2.0.7pre1           2.0.7pre1     (can you see the silent 0: ?)
  2.0.7             1:2.0.7
  2.0.8pre1         1:2.0.7.99.1
  2.0.8             1:2.0.8
  2.0.9pre1         1:2.0.9.99.1
  2.0.9             1:2.0.9

or whatever other solution the maintainer comes up with to avoid having to 
use epochs again, until the next SNAFU.

> Just a retorical question: Would you insist on epochs if the upstream
> author accepted my numbering scheme? Would there be any reason to use
> them?

To answer your retorical question: Yes there is.  If the maintainer typos the 
changelog, and releases 2.0.9.99.1 as 2.0.99.9.1 (easy mistake to make, easy 
to miss on the upload), then we'd use an epoch to fix it, although I expect 
some genius would suggest that we use:

   2.0.x9

until 2.1.0 comes out, so that we wouldn't need to use a ``dirty, evil epoch''.

> I am also certain that I have
> not misrepresented the technical consequences of the use of epochs)

Apart from the fact that they never go away, even when used ``properly'' :-)

Cheers, Phil.



--  
To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org