[Pianod] /etc/resolv.conf issue

[Peter Li]

Indeed; I glanced over the Mac OS resolv.h docs and it looks like 
basically the same setup, but it also says that most Mac system calls 
use an updated dns.h framework instead.  I still think the patch should 
go over okay on Mac OS, but also could just preproc it for Linux builds.

It does kind of amaze me that this is the Linux method.  Seems much too 
easy for a boot time process to get trapped with bad DNS info.  But I 
wonder why WSGW doesn't have the same problem, unless it is due to 
fork() calls or is just luck of the draw in how the boot process develops.

The things you listed sound good for next release.  I'll look into this 
resolv business more and get back to you.  Talking to 
pianobar/libwaitress upstream about it makes sense though?
P



On 09/06/2013 05:26 AM, Perette Barella wrote:
> Good debugging, Peter.
>
> I haven't run an experiment, but I hazard that Mac OS X is fixed. 
>  I've never had problems with any kind of application, Cocoa or Unix; 
> I suspect Apple patched the code to reinitialize and update when 
> networks are adjusted.  Avoids glitchy troubles for millions of Mac 
> laptop users on the go.
>
> We probably should do a fix, but we may also want to notify the 
> appropriate group on the Linux side of things and get them to look 
> into a broader fix for the millions of Linux laptop users out there.
>
> It'll be a bit before r150, which should have this patch, ... actually 
> it's a good time to update libpiano and libwaitress too... and the 
> setgroups fixes.  Anything else we should get in there?
>
> Perette
>
> --
> Perette Barella . perette at barella.org <mailto:perette at barella.org> . 
> 585-286-1312
> 176 Middlesex Road, Rochester NY 14610
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.deviousfish.com/pipermail/pianod-deviousfish.com/attachments/20130906/edc51e7a/attachment-0002.htm>