> It's unsurprising that a POSIX program ported straight to an 
> independently developed, properly done POSIX layer performs 
> better than a version of that program hacked to directly 
> target the platform.

On the contrary, I am somewhat surprised. Cygwin is wonderful, but there
is a definite overhead to the POSIX emulation layer when it comes to
being careful about the semantics.

For instance, a "cp -rp" by Cygwin is noticeably slower than XCOPY or
even MKS' pure-native port of "cp".

I wonder if there are other factors at work here (e.g. line-ending
conversion, etc. - was the Cygwin version just writing out Unix line
endings (no conversion) whereas the native CVS was, or something?).
Unfortunately I'm not near my repository where I can do a reasonable
comparison (access over the WAN will swamp the readings), but I can try
this next week and see which is faster given a level playing field.

Still, either way, either Windows port of CVS will be slower than a
Linux CVS.
--
Shankar.



_______________________________________________
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs

Reply via email to