<br><br><div class="gmail_quote">On Sat, Dec 11, 2010 at 4:57 AM, Kevin Kofler <span dir="ltr"><<a href="mailto:kevin.kofler@chello.at">kevin.kofler@chello.at</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<div class="im">Bill Nottingham wrote:<br>
> Right. To do this in systemd implies that you're patching openssh to<br>
> do socket-based activation... hence why I asked about upstream's opinion<br>
> on it.<br>
<br>
</div>Why would we care?<br>
<br>
It's our goal to have ALL network daemons be socket-activated eventually.<br>
This is a distribution-wide feature and as such software MUST be patched to<br>
support it.<br></blockquote><div><br>Socket activation is not mandatory or even a benefit in all cases. Just because we have a patch doesn't mean it is the right one. Upstream might have the foresight and the knowledge to see problems with patches we might not. There might be security implications. It specific cases, we can consider overriding upstream decisions but there should be very strong reasons to do so. Just because a license allows it doesn't make it best practise. It is always useful to get more peer reviews from patches with the expertise in the codebase. Upstream is a good place for that. <br>
<br>Rahul</div></div>