Bill Barker wrote:
Peter's test was stressing my poor little box a lot (which it really
shouldn't, since it is a high-concurrency/low-traffic test). It was one
of the things I wanted to look into when I had time to run more tests.
Also, I've been running the test with sendfile disabled. It seems that
Solaris7 doesn't have this particular feature :(It was added to
Solaris8). With sendfile enabled, I get a nice core-dump (since there is
no apr_socket_sendfile in my libapr.so :). I'm guessing that a for
production release we need something like this in AprEndpoint.init:
if(useSendfile && !Apr.hasFeature(Apr.SENDFILE)) {
log.warn("disabling sendfile, since your apr version doesn't
support it");
useSendfile = false;
}
I didn't know that. Yes, the check is needed then.
He still needs to remove some syncs in the native code, apparently (=
the hacks he added on top of APR, but which didn't quite work).
It seems I need to look for a few more patches before I can usefully test.
Which changes do you suggest besides fixing sendfile ?
Rémy
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
