Re: unison 2.9.45 - new package for review (2.9.20 too)

["Max Bowsher" <maxb at ukf dot net>]

Jari Aalto+list.cygwin-apps wrote:
> * Wed 2003-06-11 Max Bowsher <maxb@ukf.net> list.cygwin-apps
> * Message-Id: <022401c33059$93716290$78d96f83@pomello>
>> Also, I strongly question the wisdom of packaging an unreleased version
of
>> Unison, *especially* since Unison requires exact version equality on the
2
>> sides of a link.
>>
>> The 2.9.45 source isn't even available from the unison site any more, as
the
>> latest snapshot has changed.
>
> I intend to track the dev releases. I added the stable release as
> well. Now the setup.hint reads:
>
>     curr: 2.9.20
>     test: 2.9.45
>
> Since x.20. there has been lot of changes.

Maybe, but they haven't been released. You are incorrectly describing the
versions:

Unison 2.9.1 = Stable
Unison 2.9.20 = Beta
Unison 2.9.45 = Snapshot of the PRCS repository

Now, I know that the 2.9.20 beta has proven very stable, and is probably a
release in all but name, and so a good candidate for [curr].
But unless there are *strong* reasons, it is very unusual for Cygwin [test]
packages to contain bleeding-edge code - they are normally used for
release-candidate packages.

For this reason, I don't think 2.9.45 should be packaged. In any case, were
you seriously thinking of building new packages every time a Unison
developer checks in to the repository?

>>> + setup.hint
>>> @ unison
>>> sdesc: "Two way file synchronizing tool. Like rsync(1)"
>>
>> I think the rsync(1) man-section notation is probably unneccessary
clutter
>> in this scenario.
>
> Fixed. Anything more?
>
> http://www.tierra.dyndns.org:81/cygwin/unison/unison-2.9.20-1-src.tar.bz2
> http://www.tierra.dyndns.org:81/cygwin/unison/unison-2.9.20-1.tar.bz2

Yes, I object to the fact that you have patched Unison such that it depends
on an unreleased version of the OCaml compiler, for no good reason.

I also note that you have invented your own set of Cygwin-package-building
scripts. This isn't necessarily bad, but it would have been nicer if you had
discussed why you felt the existing recommended generic-build-script did not
meet your needs. Your system is moderately complex, and I don't have
sufficient time to investigate it properly right now.


Max.