[288] in athena10
Re: autodebathenify
daemon@ATHENA.MIT.EDU (Timothy G Abbott)
Mon Jun 30 16:18:25 2008
Date: Mon, 30 Jun 2008 16:17:39 -0400 (EDT)
From: Timothy G Abbott <tabbott@MIT.EDU>
To: ghudson@mit.edu
cc: athena10@mit.edu
In-Reply-To: <200806280524.m5S5OtfY027387@outgoing.mit.edu>
Message-ID: <alpine.DEB.1.10.0806280141500.970@vinegar-pot.mit.edu>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; format=flowed; charset=US-ASCII
On Sat, 28 Jun 2008, ghudson@MIT.EDU wrote:
> Tim and Anders expressed a preference for starting with automatic
> builds and notifications, but manual uploads. Obviously SIPB can
> choose this approach even if Athena 10 doesn't, but:
>
> 1. From my perspective that would only be minimally helpful in
> keeping the repository up to date.
>
> 2. I haven't yet seen a case of a debathenify script succeeding
> while producing a regression.
>
> 3. debathenify scripts currently do not make it easy to detect
> whether or not they concluded there was something new to build;
> that would have to change to make notification of new builds
> possible.
>
> 4. debathenify scripts currently repeat work if you tell them to
> "binary" multiple times without doing an "upload". That would
> have to change to prevent the cron job from performing large
> amounts of work on repeated invocations before a human does the
> upload.
One strategy one could use is to upload to a staging repository rather
than the main repository, which is in the sources.list of the build server
but not production systems. There could then be notification on the
condition of anything being in the staging repository, and a script to
move newly built things to the main repository and delete them from the
staging one.
-Tim Abbott
