Re: [PATCH] Re: Cygwin select() issues and improvements

[Corinna Vinschen <corinna-cygwin at cygwin dot com>]

Hi John,

On May  8 16:43, john hood wrote:
> On 3/29/16 8:49 AM, Corinna Vinschen wrote:
> > John, ping?
> 
> Sorry it took so long to reply, but I finally got around to cleaning up
> the patchset, I've mailed it separately.

I don't see the patchset anywhere.  Did I miss your mail or did it
fail to make it to this list?!?

> I was pretty frustrated at my
> slow Windows machine and the friction in dealing with the project,

What friction?  Was there anything I or others did to alienate you?
If there's some problem, please also feel free to discuss on the
#cygwin-developers IRC channel @Freenode.  You're apparently lurking
anyway.

> also sick for a while.

My sympathy.

And no worries if anything takes longer than anticipated.  We're not
on some tight schedule here :)

> > On Mar 20 16:00, Corinna Vinschen wrote:
> >> Does it really make sense to build up and break down a timer per each
> >> call to select_stuff::wait?  This function is called in a loop.  What
> >> about creating the timer in the caller and only arm and disarm it in the
> >> wait call?
> 
> Not fixed.  Managing the resource gets more complicated if you hoist it
> up to the caller with its various exit conditions.  Plus, doing that
> doesn't make the minimum or typical cost of select any less, it only
> makes the handling of events that select ignores slightly more
> efficient.  As I see it, ignored mouse events on the console are the
> only case where we're likely to see many ignored events, correct me if
> I'm wrong.

I don't know ATM.  I'll have another look when I review the patch but
actually we can also just keep it in mind for some later optimization,
*iff* it makes sense.

> >> Why?  It's a lot of additional debug output.  Was that only required for
> >> developing or does it serve a real purpose for debugging user bug reports
> >> in future?  If so, I wouldn't mind to have a bit of additional description
> >> in the git log to explain the debug statements...
> 
> Split out into a separate patch, apply or discard as you please.  I
> think it's useful for user bugs (when select is breaking, it really
> helps to see what's going on inside) but I don't feel strongly about it.

Ok.


Thanks,
Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

Attachment: signature.asc
Description: PGP signature