> On Jul 13, 2014, at 4:37 AM, Ed Maste <[email protected]> wrote:
> 
> On 11 July 2014 19:15, Greg Clayton <[email protected]> wrote:
>> Author: gclayton
>> Date: Fri Jul 11 18:15:11 2014
>> New Revision: 212856
>> 
>> URL: http://llvm.org/viewvc/llvm-project?rev=212856&view=rev
>> Log:
>> Remove assert now that we have a 'i' character that might come through as 
>> well as the 'q' character on the interrupt pipe.
> 
> In r210592 fixed the same issue for the non-Apple select()
> implementation.  I did it by merging the two select() implementations
> into one (with a bit of #ifdef-ery) because there were a few other
> unintentional differences between the two implementations.
> 
> I didn't look into the #if 0'd poll-based implementation when I made
> that change.  Is it currently being used somewhere?

No, I tested it out while looking for a solution to my select problems when we 
were running over 1024 file descriptors in Xcode, but never ended up using it. 
This can be removed.

> 
> In any case, we still don't return eConnectionStatusInterrupted for
> 'i' in the poll version.  Is there any reason not to just fold the
> poll() version into the Apple and non-Apple select() version, with a
> bit more #define magic?

Unless there is a compelling reason to use poll() (like performance) we can 
remove this.
_______________________________________________
lldb-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/lldb-commits

Reply via email to