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

Re: The cost of CLIM and its future

  Date: Wed, 16 Sep 1992 18:11-0400
  From: Scott McKay <SWM@stony-brook.scrc.symbolics.com>

  Could somebody suggest a way that us Lisp companies get to stay in
  business?  Please don't make any suggestions that you and your
  management aren't willing to pay cash-money for.  I'm serious. 

Well, we are desperately in need of a portable graphics substrate for
Common Lisp, and we are willing to pay for it.  We are not
at all interested in grabbing unsupported graphics toolkits of dubious
quality for the sake of saving a few hundred or even a few thousand
dollars.  Free software is not free because you must maintain it
yourself and because inevitably it doesn't do exactly what you want.
So you spend money paying programmers anyway and you end up with a
subset of what clim offers.  And if you want to go to another
hardware platform you may have to start all over again.

On the other hand, with a portable program we can respond to
the market rapidly.  Many customers require specific platforms
that we may not have previously tried.  Portability means we
can win those contracts merely by buying the new hardware/software
and recompiling.

Software maintenance can account for as much as half of the lifetime
cost of a software project.  Portability issues can weigh
heavily in this area.  

   Date: Wed, 16 Sep 92 19:27 MDT
   From: Simon Leinen <simon%lia.di.epfl.ch@lucid.com>

   For me, CLIM is exactly like OSF/Motif except
   that it is technically more mature and lacks critical mass.

Yes but if next year Motif falls out of favor and OpenLook is
the sexy thing to have, you can port your program with one line of code:
   (setf *default-server-path* '(:openlook))
This is more than a minor difference.  CLIM operates at
a much higher level than Motif because it does not presume
a look and feel, it lets the programmer choose the look and
feel with a switch.  CLIM provides a wide variety of high-level
tools that we would otherwise reinvent if all we had was Motif,
so here again we reap a benefit.

Agreed that CLIM is quite immature.  But look at the specification:
it is huge!  There are many more things to go wrong than in Motif.
On the other hand, Motif is not perfect either; the next release
this fall is supposed to have over 400 bug fixes.  Nothing is
perfect.  

  From: Marty Hall <hall@aplcenmp.apl.jhu.edu>
  Date: Thu, 17 Sep 1992 11:39:06 -0400
  
  At LUV-92, during the "CLOS vs. C++" Panel, one of the points that 
  struck me was that the "bundledness" of LISP, although better for
  us users, was much worse for the market.

Yes, everyone agrees that clim has too much in it, but no one agrees
about what to remove.  For us, this is the most important long
term issue.

A good example is DRAW-LINE.  Several of us
felt that DRAW-LINE should take just five arguments:
(stream x1 y1 x2 y2 ink).  This would be in everybody's lisp image
because its a simple common denominator
that others could use to implement libraries of higher-level
graphics operations, such as a version that uses the dreaded "point" arguments.
But just because somebody uses "point" arguments doesn't mean that every
lisp image should include the capability.  Similarly, I don't see why
clim bothers with dashed lines.

There have been a lot of requests for new features on this mailing list
lately, and I am against accretion of features for this reason.  But
I am FOR accretion of libraries, which is how this whole discussion
began.  A "lean and mean" clim could probably be ratified into ANSI
Common Lisp, but as it stands now clim is a beast.

In the short term, I just want the current design fielded.  But
in the long term, this issue must be addressed.

jeff morrill
jmorrill@bbn.com


0,,
Main Index | Thread Index