summaryrefslogtreecommitdiffstats
path: root/zarb-ml/mageia-dev/attachments/20120611/2ee669f4/attachment.html
blob: 8fcfdd061667ac00dcc1ad64e304bc5b6fbc5d75 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
<p>On Jun 11, 2012 1:14 PM, &quot;Colin Guthrie&quot; &lt;<a href="mailto:mageia@colin.guthr.ie">mageia@colin.guthr.ie</a>&gt; wrote:<br>
&gt;<br>
&gt; &#39;Twas brillig, and Guillaume Rousse at 11/06/12 10:19 did gyre and gimble:<br>
&gt; &gt; Le 11/06/2012 09:56, Colin Guthrie a écrit :<br>
&gt; &gt;&gt; Well it the .so name does not change then it *should* be ABI compatible.<br>
&gt; &gt;&gt; It seems in this case it&#39;s only API compatible but that doesn&#39;t rule out<br>
&gt; &gt;&gt; certain values, macros or flags changing such that everything needs a<br>
&gt; &gt;&gt; recompile.<br>
&gt; &gt; Just rebuilding openssl with different build options already triggered<br>
&gt; &gt; such kind of issues. I remember adding SNI support broke ntp in the<br>
&gt; &gt; past, until this last package was rebuilt. The safe way should be to<br>
&gt; &gt; rebuild everything right now, but given than almost everything in the<br>
&gt; &gt; distribution uses this library, I think just letting normal update flow<br>
&gt; &gt; should be enough, unless actual problems are found.<br>
&gt;<br>
&gt; Probably true and I think we should ensure that a full rebuild is done<br>
&gt; before the beta cycle anwyay. We did moot this for mga2 but it was<br>
&gt; deemed (correctly IMO) too late to start messing with that.<br>
&gt;<br>
&gt; But until then a full rebuild is likely overkill.<br>
What about gcc 4.7? I think we should switch to that one as well. And then rebuild is needed anyway. So full rebuild at the moment is probably not the best idea.<br>
--<br>
Sander</p>