Path: news.daimi.aau.dk!news.uni-c.dk!nntp-oslo.UNINETT.no!nntp-trd.UNINETT.no!online.no!oslonett.no!sn.no!Norway.EU.net!EU.net!newsfeed.internetmci.com!in3.uu.net!manawatu.planet.co.nz!manawatu.gen.nz!news.express.co.nz!waikato!canterbury.ac.nz!usenet
From: kasper@pukeko.cosc.canterbury.ac.nz (Kasper Osterbye)
Newsgroups: comp.lang.beta
Subject: Re: THIS and COMPONENTs
Date: 08 Aug 1996 20:43:46 GMT
Organization: University of Canterbury
Lines: 51
Distribution: world
Message-ID: <KASPER.96Aug9084346@pukeko.cosc.canterbury.ac.nz>
References: <4tq4s1$muk@gjallar.daimi.aau.dk>
NNTP-Posting-Host: 132.181.12.13
In-reply-to: "S\xren Brandt"'s message of 1 Aug 1996 11:36:33 GMT

Hi again,

I guess it was I who started this, and I can see that the discussion
went on, but I have not followed it on a day by day basis, and now
some parts are missing.

The low-level gismo that Soeren gave me/us is not the way it SHOULD be
done, but a way it CAN be done until we find out what can be done
about it, and until the compiler can handle it (those two things were
never the same with BETA - does :: work btw?).

My problem was that I wanted to do simulation, and needed to implement
processes that are scheduled according to simulation time.
But it should also be possible to put such an entity to sleep in a
queue, and I found it would be natural for the process to be able to
say "myQueue.wait", and I wanted to implement "wait" like

wait :
(# do "the presently executing coroutine"[] -> theQueue.insert;
   SUSPEND
#)

Also I wanted to be able to start new processes like
"myProcess.start", and wanted to implement start like:

start :
(# do this(Process)[] -> scheduler.insert
#)

Soerens gizmo helped me do the last one, and I do not find it very
problematic to do
start:
(# do this(Process)[] -> objectToComponent -> scheduler.insert
#)

However, the "wait" I could not do without introducing an explicit
"presently executing process" variable in the scheduler. This is also
done in the system libraries, so I assume that the problem is well
known, and that a "presently executing process" is part of the
standard workaround in building schedulers. Indeed, Soeren thought
that this was my original problem, and suggested that I used the
that idea.

Notice that in the low-level library, there is a pattern that will
test if an object-reference does indeed refer to a component, that
will allow us to use objectToComponent in a safe manner.
I assume that Soerens gizmo will be part of the lowlevel libraries
sometime in the future.

-- Kasper