Re: [RFC] Added editor driver restrictions

My opinion:

  * Based on what I have seen, I doubt this aspect of keyword
substitution is particularly slowing down checkouts (although it is
probably slowing down exports).

  * The editor driver restriction is probably worth it just to speed up
exports, since it has no particular cost.

  * The svn_stream_tee() thing is probably not worth the complexity.
Having to reread the text base is probably not slow. (But feel free to
do some tests to prove me wrong. I may also be overestimating the
grossness factor.)

  * We may get a bigger boost on checkouts by rewriting the translation
functions to do reads in larger chunks than one byte. (The input and
output files are buffered, but doing several function calls per byte
checked out may be straining most people's CPUs.) I suspect it will be
no more than a 5-10% improvement in overall checkout speed, but that's
probably the most improvement we can get from a single small change at
this point.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Oct 22 20:03:07 2003