<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    On 04/27/2012 05:46 PM, Micah Gersten wrote:
    <blockquote cite="mid:4F9B21DA.20408@ubuntu.com" type="cite">
      <meta content="text/html; charset=ISO-8859-1"
        http-equiv="Content-Type">
      On 04/27/2012 05:20 PM, Scott Kitterman wrote:
      <blockquote cite="mid:1414140.8tMjOy7yqh@scott-latitude-e6320"
        type="cite">
        <pre wrap="">On Friday, April 27, 2012 05:13:21 PM Micah Gersten wrote:
</pre>
        <blockquote type="cite">
          <pre wrap="">Now that we have the faster i386/amd64 buildds, I'd like to suggest at
least one extra rebuild earlier in the cycle at least for i386/amd64 so
there's time to get these changes upstream.

Around alpha 1 ~ June 7 (archive raw, lots of failures, should catch
most of the Ubuntu change related gcc-4.7 issues though)
After Feature freeze ~ Aug 23 (archive should be fairly stable for most
dependency trees, check to see what new failures were introduced through
new upstream versions)
During beta 2 freeze ~ Sep 25 (shake out the last round of failures)
</pre>
        </blockquote>
        <pre wrap="">My initial reaction is that we're learning plenty from the existing rebuilds.  
There's no real need to add to the pile of work we don't have time to do.

The rebuilds do have a negative effect on distro development because, even 
though they have a low priority in Soyuz, when all the builders are full due 
to rebuild packages being built nothing can start until one of the rebuild 
packages finishes.  Since they build in alphabetical order, about the time you 
hit gcc*, this can really slow things up.

I would not add more rebuilds unless it's really clear there will be benefit.  
Until after DIF much of what turns up in a rebuild is inevitably archive skew 
that it's not hard to find other ways.

Scott K

</pre>
      </blockquote>
      Now that we can use -proposed during development, the only arch
      skew should be from autosyncs.  The current failures do tell us a
      lot, but quite a few will probably be fixed with the autosyncs as
      well.<br>
      Remaining failures from the last rebuild in precise:<br>
      <table class="grid" width="95%">
        <thead><tr>
            <th colspan="2" rowspan="2"><br>
            </th>
            <th colspan="2" class="center">Main archive</th>
            <th colspan="1" class="center">Ports archive</th>
          </tr>
          <tr>
            <th class="center">i386</th>
            <th class="center">amd64</th>
            <th class="center">armhf</th>
          </tr>
        </thead> <tbody>
          <tr class="even" style="text-align: center;">
            <td class="FAILEDTOBUILD">arch (F)</td>
            <td style="text-align: left;">Package failed to build</td>
            <td onmouseover="Tip('<table><tr><td>main:</td><td
              style="text-align:right;">6 (22
              superseded)</td></tr><tr><td>universe:</td><td
              style="text-align:right;">106 (107
              superseded)</td></tr><tr><td>multiverse:</td><td
              style="text-align:right;">4 (1
              superseded)</td></tr></table>')" onmouseout="UnTip()"
              >="" 116="" <="" td=""> <br>
            </td>
            <td onmouseover="Tip('<table><tr><td>main:</td><td
              style="text-align:right;">4 (20
              superseded)</td></tr><tr><td>universe:</td><td
              style="text-align:right;">78 (76
              superseded)</td></tr><tr><td>multiverse:</td><td
              style="text-align:right;">3 (1
              superseded)</td></tr></table>')" onmouseout="UnTip()"
              >="" 85="" <="" td=""> <br>
            </td>
            <td onmouseover="Tip('<table><tr><td>main:</td><td
              style="text-align:right;">4 (22
              superseded)</td></tr><tr><td>universe:</td><td
              style="text-align:right;">141 (80
              superseded)</td></tr><tr><td>multiverse:</td><td
              style="text-align:right;">3 (2
              superseded)</td></tr></table>')" onmouseout="UnTip()"
              >="" 148="" <="" td=""> <br>
            </td>
          </tr>
          <tr class="odd" style="text-align: center;">
            <td class="MANUALDEPWAIT">arch (M)</td>
            <td style="text-align: left;">Package is waiting on another
              package</td>
            <td onmouseover="Tip('<table><tr><td>universe:</td><td
              style="text-align:right;">5 (3
              superseded)</td></tr><tr><td>multiverse:</td><td
              style="text-align:right;">4 (1
              superseded)</td></tr></table>')" onmouseout="UnTip()"
              >="" 9="" <="" td=""> <br>
            </td>
            <td onmouseover="Tip('<table><tr><td>main:</td><td
              style="text-align:right;">0 (1
              superseded)</td></tr><tr><td>universe:</td><td
              style="text-align:right;">4 (4
              superseded)</td></tr><tr><td>multiverse:</td><td
              style="text-align:right;">3 (1
              superseded)</td></tr></table>')" onmouseout="UnTip()"
              >="" 7="" <="" td=""> <br>
            </td>
            <td onmouseover="Tip('<table><tr><td>universe:</td><td
              style="text-align:right;">105 (20
              superseded)</td></tr><tr><td>multiverse:</td><td
              style="text-align:right;">0 (1
              superseded)</td></tr></table>')" onmouseout="UnTip()"
              >="" 105="" <="" td=""> <br>
            </td>
          </tr>
        </tbody>
      </table>
      <br>
      So, most of the armhf failures are related to gnat-4.6 not being
      available.  That leaves a little more than 100 failures some of
      which will be resolved with autosyncs (also most are probably not
      easy fixes).  A new rebuild would give people a chance to pick off
      easy fixes and also let us know where we stand WRT new failures
      early on.<br>
      So, maybe we should wait until after DIF, then there would be very
      little risk of archive skew during the rebuild.<br>
      Another thing to consider, is with Debian freezing for Wheezy at
      some point, the likelihood of fixes from new upstream versions
      getting in through Debian decreases as we get further into the
      cycle.<br>
      <br>
      Micah<br>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <br>
    </blockquote>
    Sorry, the numbers were apparently lost on send:<br>
                                                        Failure type    
    Main archive     Ports archive<br>
                                                                         
           i386    amd64    armhf<br>
    arch (F)     Package failed to build                          116
        85     148<br>
    arch (M)     Package is waiting on another package     9     7    
    105 <br>
  </body>
</html>