On Fri, Jun 6, 2008 at 1:15 AM, Sebastien Bacher <<a href="mailto:seb128@ubuntu.com">seb128@ubuntu.com</a>> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Le jeudi 05 juin 2008 à 11:03 -0700, Jordan Mantha a écrit :<br>
<div class="Ih2E3d">> I seriously question the usefulness of -proposed for regression<br>
> testing.<br>
<br>
</div>Hi,<br>
<br>
Taking a recent example, the evolution 2.22.1 to <a href="http://2.22.1.1" target="_blank">2.22.1.1</a> stable updates<br>
had several regressions, in the days following the hardy-proposed update<br>
we got several bugs from people who updated to the new version, we<br>
worked with upstream on those issue, it took some evolution-data-server<br>
and evolution updates iterations until having the people confirming that<br>
those issues were fixed before we moved the new version to<br>
hardy-updates, using hardy-proposed has been really valuable and worked<br>
as expect in this case, and for several other updates since hardy</blockquote><div><br>Well, perhaps I should have qualified my statement a bit :-) For high-use packages I can imagine that we would get enough people to have a decent amount of regression testing. For a great many packages (especially in Universe) however, 1 week in -proposed just isn't going to be enough time to get adequate usage for regression testing. My point wasn't so much that -proposed is *bad*, just that I think we could maybe do *better* than to just having people turn it on and see what happens.<br>
</div></div><br>-Jordan<br>