This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [Patch] Fix libstdc++/11378 (3rd take)


Nathan Myers wrote:

Does anybody know a circumstance where short writes are common enough that performance in that case matters?

For sure there are other factors I currently ignore, but it seems to me that the relationship between BUFSIZ and PIPE_BUF is important, right? I understand from http://www.opengroup.org/onlinepubs/007904975/functions/write.html (please correct me asap if I'm wrong!) that if our requests were always smaller than PIPE_BUF a Posix conforming write would never be short.

Unfortunately, a little bit of googling reveals that on many systems PIPE_BUF is quite small, smaller than the already suboptimal BUFSIZ (512 bytes on NetBSD, f.i.).

Paolo.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]