Using Processes in CLIM

• To: Kim Barrett <kab@chestnut.com>, DCPL@fuji.ila.com
• Subject: Using Processes in CLIM
• From: "David C. P. Linden" <DCPL@fuji.ila.com>
• Date: Tue, 2 Apr 91 11:35 EST
• Cc: jonl%kuwait@lucid.com, clim@BBN.COM
• In-Reply-To: <9104021536.AA00892@chestnut.com>

    Date: Tue, 2 Apr 91 10:36:08 EST
    From: kab@chestnut.com (Kim Barrett)

    I don't have any problem with the internal-to-CLIM substrate making use of
    implementation specific features to do its job.  I am concerned that some of
    that internal substrate might get shifted into the external interface
    inappropriately.

That's a valid concern, which is why we are discussing putting it in a
package other than CLIM:, and I also think we need the documentation to
be very clear what the story is.

    For example, I'd really hate to see CLIM applications
    explicitly dispatching on the *multiprocessing-p* variable you mentioned.

Are there really more than three choices?
 -- Do nothing.  [Pressure from users is making this harder.]
 -- Abstract multiprocessing so that is is globally and universally
    usable in all tasking models?  [If possible, I'm not sure this will
    result in anything usable.]
 -- Define ways to get to the various models, thus requiring
    applications to know the various models and interact with them.  [I
    agree this isn't aesthetic, but I'm really at a loss of what other
    alternatives exist.  I'm wide open to suggestions.  Hint, Hint.]



0,,

• Using Processes in CLIM
• From: Kim Barrett <kab@chestnut.com>

• Prev: Using Processes in CLIM
• Next: Re: Using Processes in CLIM
• Index: CLIM Mailing list (5-Oct-90 to 31-Dec-91)
• Thread: CLIM Mailing list Index (5-Oct-90 to 31-Dec-91)