CLIM mail archive
[Prev][Next][Index][Thread]

Re: missing OR ptype

[This is the second attempt to send this message - apologies to anyone
who has already received it.  No changes from the first attempt.]
********************************************************************************
    Date: 8 Feb 91 21:39:55 GMT
    From: Kevin Thompson <kthompso@ptolemy.arc.nasa.gov>


    In article <10371@ptolemy-ri.ptolemy.arc.nasa.gov> Dan Cerys <cerys@BBN.COM> writes:
    >Just to clarify, we are using CLIM 0.9 on a variety of platforms.  The
    >excellent presentation system done by Symbolics for CLIM 1.0 is
    >unfortunately not in 0.9.  (Sigh).  

    Aieeeeeeee.  Could someone explain why the various versions of CLIM appear
    to be diverging?  
(Actually, they're converging.)
		      I've heard rumours of this from several different parties,
    and it *does* after all seem to defeat the main purpose (in my mind at
    least) of CLIM -- platform independence.  Is there a single person (at ILA I
    assume) who can assuage fears that platform independence is far, far off?
I'll try.

    Kevin Thompson

    --
    kthompso@ptolemy.arc.nasa.gov     NASA-Ames Research Center, Moffett Field, CA

[The companies that are cooperating on CLIM are still trying to refine
this new game plan of which you've heard, and to produce a clear and
comprehensive public statement setting it forth.  I hope you will
understand my hesitation in being drawn into a dialogue prematurely.
With that said, I think some of these rumours need to be cleared up.]

As most of the readership of this list will know, there have been two
dialects of CLIM during its two-and-two-thirds-years of life, the second
coming into being roughly at the one-year mark. These dialects have
always been largely - but not 100% - compatible at the API level.  They
are different, though, in a few important ways at the architecture/
internal-superstructure level, most significantly with respect to the
"porting technology," which is to say, how the generic API is connected
to a specific underlying Window System or UI Toolkit (like X, Motif,
OPEN LOOK, MS-Windows, QuickDraw, PostScript, and so on).

Much time and energy has been expended by all of the parties involved
with the CLIM effort in balancing the various collective and company-
specific interests that are embodied in CLIM.  Most recently (and
relevantly to Kevin's message) we have been faced with an almost classic
tradeoff between quality and functionality:  The reference
implementation of the older dialect (which goes by the name of "CLIM
1.0") is now far more mature, and therefore of higher quality, than the
reference implementation of the newer dialect (which goes by the name of
"CLIM 0.9"). [Yes, we know this numbering scheme is confusing: it's a
remnant of a period where the sequencing of product releases was much
less clear than it is now.  We're going to be dropping the designation
"0.9".]  The newer dialect, though, has design features that are thought
to be important for some of the intermediate-to-long-term requirements
for anything purporting, as CLIM does, to serve as a true platform-
independent UIMS for Common Lisp.  Most importantly, it has a rational
scheme for connecting up to Toolkits (like Motif and OPEN LOOK), rather
than "bare" Window Systems (like Xlib/CLX).

(I should note that it is an over-simplification to use the designations
"older" and "newer."  There are many isolated functions and a few entire
modules in "CLIM 1.0" - of which the "excellent 1.0 presentation system"
is one - that were developed more recently than the corresponding
functions/modules in "CLIM 0.9.")

The collective plan is to try to bring to market a family of ~100%
API-compatible CLIM 1.0 products as soon as possible.  The Genera
version is now shipping, and we expect that versions for the other
interesting Lisp/OS/WS combinations will follow shortly (within a
quarter).  In parallel to the CLIM 1.0 product(s) roll-out, we will be
finalizing the CLIM 2.0 specification, which we believe fuses the best
of the "0.9" and "1.0" dialects (including the latter's presentation
subsystem, for those who were wondering).  The collective plan is to
bring (~100% API-compatible) products based on that 2.0 specification to
market within the year on the same Lisp/OS/WS platforms as will be
covered by 1.0, and one or two significant new ones.

There is one very important class of user for whom the preceding scheme
is "less than optimal."  These are the people who have already made
investments in and/or commitments to the "0.9" dialect, not a few of
whom actually got involved with CLIM long enough ago that they began
with the first dialect and subsequently migrated to the newer one.  All
of the CLIM vendor companies have committed themselves to "doing the
right thing" for these users.  In some cases, this will mean helping
them move sideways to the 1.0 dialect.  In others, it will mean
supporting them as they stay with 0.9 through the 2.0 unification
period.

It is ILA's belief that the prospects for CLIM have never been better.
The first true commercial quality product (CLIM 1.0 for Genera) is
shipping.  The loose coalition of vendors participating in the CLIM
effort is more unified and focused than it has ever been.  The resources
being devoted to CLIM are increasing, and increasingly concentrated.  A
hardening of the two-dialect condition into a permanent state of affairs
has been avoided.

I will be happy to address any further concerns that may be floating
around out there, but I would like to ask that vendor-specific questions
be directed to the individual vendors themselves.


Mark Son-Bell
ILA
114 Mt. Auburn St., 4th Floor
Cambridge, MA 02138
mas-b@ila.com

0,,
References:
Main Index | Thread Index