[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Help-smalltalk] [PATCH] Process
From: |
Holger Hans Peter Freyther |
Subject: |
Re: [Help-smalltalk] [PATCH] Process |
Date: |
Mon, 24 Mar 2014 15:07:51 +0100 |
User-agent: |
Mutt/1.5.21 (2010-09-15) |
On Mon, Mar 24, 2014 at 12:01:16PM +0100, Gwenaël Casaccio wrote:
>
> ProcessBeingTerminated is in SystemExceptions namespace otherwhise it's nil
oops. so this has never worked so far?
> Yes also for another reason with my change the process is always on
> the right priority queue,
> the current implementation would only put it when it's yielding or
> scheduled. I strongly
> believe that the invariant : ((processor processesAt: priority)
> select: self) should be correct.
yes. Could you make a separate patch to enforce this?
> >hmm. I don't think test will be quite flaky in terms of scheduling and other
> >parts. Do you think we can mock/test this in a more reliable way?
> >
> I agree it only works when you've a clean image and this is what I
> want otherwise
> tests needs to rely on multiple process/scheduling parts that I want
> to tests separately.
"clean" it means that the call-in this test is executed from needs
to have lower priority than the forked call. E.g. what happens if we
ever execute it from VisualGST or if we ever have multiple native
dispatchers?