httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dean Gaudet <>
Subject Re: [STATUS] Apache 1.2.2 (Sun 3-Aug-1997 12:22 MET DST)
Date Wed, 06 Aug 1997 08:48:40 GMT
Yeah you're right as long as all the error conditions are properly dealt
with there is no need for the extra start_chunk.  I just wasn't thinking
straight enough. 


On Tue, 5 Aug 1997, Roy T. Fielding wrote:

> >> There is no need to start a chunk until there is data to be written,
> >
> >Did you notice that bflush() never calls start_chunk?  A call needs to be
> >added there or here.  I figured it was better here, because it defended
> >against any other lameness.
> I don't get it.  I looked at the code again and still don't understand
> why you would want bflush() to call start_chunk().  After the buffer
> is flushed, fb->outcnt == 0, so the next write will have to go through
> bcwrite() if it is bigger than fb->bufsiz, which is exactly what we want
> it to do.  Calling start_chunk() in bflush() just defeats that optimization,
> and isn't needed in any case because it is called by bwrite() before it
> buffers any data into an empty buffer.
> ....Roy

View raw message