Subject: Re: [linux-audio-dev] Article about multithreading
From: Paul Davis (pbd_AT_Op.Net)
Date: Mon Jun 18 2001 - 23:28:59 EEST
>> mutex might conceivably lead to a deadlock eventually). Deadlock is
>> really programmer error, so trying to avoid it is a really misplaced
>> effort.
>
>The conclusion "trying to avoid deadlock is misplaced effort" does not
>follow from "deadlock is programmer error".
Joe didn't say that. You've reversed the order of his logic.
If a program deadlocks, the programmer made a mistake writing
it. Thats all there is to it. A correctly written programmer never
deadlocks. A correctly written program, however, can be hard to write.
Thus, how much effort should be put into:
* run-time code to help resolve or avoid deadlocks?
* tools to help the programmer analyse deadlocks?
* programming metaphors to help avoid deadlocks?
I would say none on the first, and then split any effort you want to
expend between the 2nd and 3rd.
--p
This archive was generated by hypermail 2b28 : Mon Jun 18 2001 - 23:27:12 EEST