Re: (no subject)

Aron Rubin wrote:
> Steve Arons wrote:
> 
>> It's quite a sad day when a better technology--ngpt--is dropped.  One 
>> thing the free software community can't afford is to put warm fuzzy 
>> feelings ahead of performance.  Or rather--dress up the community's 
>> reliance on glibc as an endorsement of its quality.  Linux would 
>> benefit more from a robust, efficient, standards-compliant libc than 
>> from a better pthreads at this point.  But in the meantime, if you 
>> happen to have a winning pthreads package, why abandon it?
> 
> 
> Because the M:N paradigm has some critical flaws in it.

e.g. "greater complexity for little benefit".
The simpler 1:1 paradigm is probably easier to implement
correctly (and maintain).

BTW does anyone have benchmark results for the benchmark
NGPT was using, http://www.uwsg.iu.edu/hypermail/linux/kernel/0301.1/1050.html,
comparing current versions of the two packages?
It'd be interesting to hear whether NPTL has edged ahead
of NGPT again in the three months since that announcement.

- Dan


-- 
Dan Kegel
http://www.kegel.com
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=78045