home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.os2.advocacy
- Path: sparky!uunet!csfb1!jbrock
- From: jbrock@csfb1.fir.fbc.com (John Brock)
- Subject: single-threaded message queue
- Message-ID: <BxuJCH.9w1@csfb1.fir.fbc.com>
- Sender: news@csfb1.fir.fbc.com (Usenet News Account)
- Reply-To: uunet!csfb1!jbrock
- Organization: First Boston Corporation
- References: <1992Nov13.164155.14309@msc.cornell.edu> <BUHR.92Nov16184637@ccu.umanitoba.ca>
- Date: Tue, 17 Nov 1992 06:00:17 GMT
- Lines: 21
-
- In article <BUHR.92Nov16184637@ccu.umanitoba.ca>, buhr@umanitoba.ca (Kevin Andrew Buhr) writes:
- |>
- |> > It seems odd that in a multitasking environment one can write any
- |> > application that takes down the whole OS. Blocked semaphores may
- |> > trap your process, but the scheduler should allow others to run.
- |> > Is it really so hard to preempt a running process?
- |>
- |> The single-threaded message queue problem has been beaten to death.
- |> This, and operating system bugs, can hang a system so it is unable to
- |> preempt processes, as you have seen. For the most part, OS/2 does a
- |> respectable job of keeping things running under adverse conditions, at
- |> least in my experience.
-
- Just for the record (I missed the beating :-) ), has IBM stated any
- intention to eventually fix this single-threaded message queue thing,
- or are they satisfied with this "feature"?
-
- --
- John Brock
- uunet!csfb1!jbrock
- jbrock@csfb1.fir.fbc.com
-