On 14 November 2015 at 17:19, Bert Huijben <b...@qqmail.nl> wrote: >> -----Original Message----- >> From: Bert Huijben [mailto:b...@qqmail.nl] >> Sent: zaterdag 14 november 2015 15:15 >> To: 'Ivan Zhakov' <i...@visualsvn.com>; rhuij...@apache.org >> Cc: dev@serf.apache.org >> Subject: RE: svn commit: r1714297 - in /serf/trunk: buckets/event_buckets.c >> outgoing.c serf_private.h > > >> Writing that function as a wrapper would require coding that function, >> reviewing it and testing it, while I don't see any usecase where somebody >> uses this function. >> And in all current implementations/usages just using the default readline >> would 100% do the same thing without measurable performance >> degradation. >> >> Same reason as why we never implement read_for_sendfile().... There is not >> a single user in serf that would use it. > > Note that 1 week ago (before r1713788) the typical implementation of > serf_bucket_readline() on most advanced buckets was a NULL pointer with a ### > TODO comment. > Understand. I was just checking whether this intentional or oversight. As far I understand serf_bucket_readline() is not that efficient as native implementation could be. I also think that event bucket should be transparent and shoud not change behavior of wrapped bucket if possible. I'll implement native readline for event bucket if you would not mind.
-- Ivan Zhakov