> -----Original Message-----
> From: Ivan Zhakov [mailto:i...@visualsvn.com]
> Sent: zaterdag 14 november 2015 10:19
> To: Bert Huijben <b...@qqmail.nl>
> Cc: rhuij...@apache.org; dev@serf.apache.org
> Subject: Re: svn commit: r1713489 - serf_bucket_readline()
>
> > I'm starting to think that we should switch to wrapping for the destroy
> event, but
> > perhaps for a different reason than stated before: The current code
> depends on
> > implementation details of the aggregate bucket.
> >
> > There are certain cases where it might destroy buckets (or might have
> destroyed)
> > out of order and when you wrap the bucket you know exactly. Not
> depending on
> > the knowledge how this bucket works makes this more generic... and
> better able
> > to handle other aggregation buckets.
> I thought about the same reason, but missed it in my original email. Sorry.
The moment I realized this problem I already thought you did... I just wasn't
sure.
I'm glad we reached the same conclusion before anything was released as public
api.
(I didn't make the event buckets public anyway... but perhaps the current code
is generic enough to make it public)
Now back to coding http/2 request framing...
(Which will now use this event bucket instead of a custom solution in yet
another bucket type)
Bert
