This is the mail archive of the cygwin-developers@sources.redhat.com mailing list for the Cygwin project.


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

Re: confirmation of 1.1.5(0.29/3/2) 2000-10-16 build


On Mon, Oct 16, 2000 at 10:52:27AM -0700, Earnie Boyd wrote:
>Much improved processing.  I was able to build without incident.  CPU usage
>still high but not constantly so that other processes can respond
>interactively.  I was unable to cause the bash exiting problem with this build.
> I have *not* updated my bash and ash with Corinna's new fixes as I didn't want
>that variable to enter into the results.  Timing results:
>
>        Command being timed: "ls --color -l /bin"
>        User time (seconds): 0.09
>        System time (seconds): 0.22
>        Percent of CPU this job got: 11%
>        Elapsed (wall clock) time (h:mm:ss or m:ss): 0:02.62
>
>/bin is mounted
>  d:\Cygwin-1.1\bin   /usr/bin            user         binmode,cygexec
>  d:\Cygwin-1.1\bin   /bin                user         binmode,cygexec

Strange.  I just duplicated the problem ten minutes ago.  I wasn't
really expecting it, of course.  I'm feverishly trying to cause it to
fail again.  One thing that I noticed is that the bash command prompt
was returning prior to command execution.  That's usually an indication
of something being screwed up in the 'exec()' code.  Coincidentally
enough, that's what I've been working on for the last several months
"in my spare time".

I'm obviously doing something to improve things, though.  I wish I knew
what...

AFAICT, 1.1.5 is still as much as 18% slower than 1.1.4.  I don't know
when this happened.  I'm tracking this down, too.

Have you timed the build with 1.1.4 vs.  1.1.5, Earnie?  I suspect that
you'll see slower execution?

cgf

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