This is the mail archive of the
mailing list for the glibc project.
Re: [Freedce-devel] pthread_clear_exit_np()
- From: Mark Brown <bmark at us dot ibm dot com>
- To: Luke Kenneth Casson Leighton <lkcl at lkcl dot net>
- Cc: freedce-devel at lists dot sourceforge dot net, libc-alpha at sources dot redhat dot com, opendce at opengroup dot org, ros-dev at reactos dot org, thomas dot schuetzkowski at web dot de
- Date: Mon, 24 Apr 2006 10:42:43 -0500
- Subject: Re: [Freedce-devel] pthread_clear_exit_np()
> Luke Kenneth Casson Leighton <firstname.lastname@example.org> wrote:
> On Fri, Apr 21, 2006 at 01:52:58PM -0700, Roland McGrath wrote:
> > Have you considered instead using standard features as specified in
> > since around 1996?
> dce predates the posix standards - that's why all dce projects, some of
> them multi-hundred-million-dollar projects by ibm, fujitsu, EDS etc -
> have had to use posix draft 4 threading.
To clarify a bit:
DCE development did predate the 1996 POSIX specs. The developers wanted
to use POSIX threads, though, so they went with Draft 4. Unfortunately
for them, the programming model changed before POSIX threads were
finalized. (As I recall this wasn't Open Group but OSF timeframe,
Open Group picked things up later.)
There are risks involved in going with an uncompleted standard; when
the standard was finalized DCE was too far along to recode - it was
actually shipping in some commercial products.
IBM (and other commercial UNIX vendors) did indeed put in special code
to allow DCE to work with the POSIX threads provided by the OS. They were
looking to save development costs. Speaking from personal experience,
keeping DCE (and DFS) working in a POSIX environment was occasionally
a real challenge. Again, this was in the mid-90s timeframe.
DCE has since had TEN YEARS to revamp and meet the POSIX standard, which
is the de jure (IEEE and ISO) and de facto standard for threading. That it
has not done so is not glibc's fault, nor is it glibc's problem (IMHO).
This is not to say that DCE (or DFS, if its scaling problems were solved)
are not good or useful ideas. Just that if they are truly important
and "strategic", then they are worth updating properly to work with
the standard (and "strategic") threading technology.
BTW: IBM stopped selling DCE/DFS years ago. The end of support
for the product is this month.