serf-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Bert Huijben" <b...@qqmail.nl>
Subject RE: svn commit: r1713788 - in /serf/trunk: buckets/buckets.c buckets/dechunk_buckets.c buckets/deflate_buckets.c buckets/hpack_buckets.c buckets/http2_frame_buckets.c buckets/prefix_buckets.c serf_bucket_util.h
Date Sat, 14 Nov 2015 14:25:15 GMT


> -----Original Message-----
> From: Ivan Zhakov [mailto:ivan@visualsvn.com]
> Sent: woensdag 11 november 2015 10:45
> To: rhuijben@apache.org
> Cc: dev@serf.apache.org
> Subject: Re: svn commit: r1713788 - in /serf/trunk: buckets/buckets.c
> buckets/dechunk_buckets.c buckets/deflate_buckets.c
> buckets/hpack_buckets.c buckets/http2_frame_buckets.c
> buckets/prefix_buckets.c serf_bucket_util.h
> 
> On 11 November 2015 at 12:39,  <rhuijben@apache.org> wrote:
> > Author: rhuijben
> > Date: Wed Nov 11 09:39:09 2015
> > New Revision: 1713788
> >
> > URL: http://svn.apache.org/viewvc?rev=1713788&view=rev
> > Log:
> > Implement a best effort serf_default_readline() via read() and peek() to
> > be used on buckets that don't have a specialized implementation and
> currently
> > use a NULL function.
> >
> Nice! Just curious did you fixed this for completeness or we have some
> real case where we need readline for dechunk buckets and others?

Having a segfault implementation with a ### TODO comment is not really a nice solution.
And it is not unlikely users want to use this function on http2 frame based buckets in the
future, as some http response will be backed by those. It is nice that all functions just
work via some default implementation.

Assuming that the bucket implements peek and read following the conventions this now 'just
works'. Buckets that can peek in future reads and are then able to copy data to a continuous
region could do better, but there are not that many that can, and don't already have a specialized
implementation.
), both functions would +- be as efficient.

	Bert


Mime
View raw message