Subject: Re: [linux-audio-dev] Article about multithreading
From: John Regehr (regehr_AT_cs.utah.edu)
Date: Tue Jun 19 2001 - 02:45:15 EEST
On Mon, 18 Jun 2001, Richard C. Burnett wrote:
> I don't agree myself. Type checking is something you do to prevent
> programming error BEFORE the program is used. Trying to 'back-out' of a
> deadlock is something you do AFTER the program is compiled and run. In my
> mind, if you get deadlocks then it is programming error and instead of
> trying to back-out of it, you want something to help narrow down the
> reasons and conditions that caused it, not fix itself taking alot of time
> and not being made aware of it.
Since I agree completely, either my mail was unclear or you
misconstrued it.
The original mail said that "Deadlock is really programmer error, so
trying to avoid it is a really misplaced effort." My position in this
conversation is that I completely disagree with that statement. In
other words, you clearly want to spend effort avoiding deadlock. As
Paul says, this might be through tool support such as a lock analyzer or
it might be through language constructs that do not admit deadlock. Of
course it can also be through disciplined and/or careful programming,
and you have to be careful and disciplined anyway, if you're writing
real-time systems...
> Additionally, garbage collection is something totally different and in my
> mind designed for people that don't really program that well. I have yet
> to use a language in which the garbage collector worked well in many
> situations.
Perl.
John Regehr
This archive was generated by hypermail 2b28 : Tue Jun 19 2001 - 02:43:56 EEST