Re: mcl and acl for ibm pc

["Don Mitchell" <dhm%proact@uunet.UU.NET>]

Subject:   RE>mcl and acl for ibm pc
ACL for ibm pc and ibm risc are two completely different products.  
I'll only speak to general differences between acl pc and mcl.  If you 
have specific questions, I'll answer them.

They both support most of CLtL2 or dpANS.  There are a few annoying 
ommissions in both such as map-into.  ACLPC is missing the 
dynamic-extent declaration while mcl uses it to the utmost advantage.  
More important to me,  ACLPC is missing conditions.

MCL's compiler does wonders.  Someone had an example of a function 
last week in which mcl generates faster code than C!   ACLPC's compiler 
is horrid.  It does not even catch missing arguments let alone do 
anything in the way of optimization.  It does do some type optimization 
if you use type declarations.  Note, to do much with any compiler, you 
must use declarations.

MCL's editor is very powerful.  It's a version of emacs.  It does not offer 
m-x commands, however.  ACLPC offers four editor modes: Windows, Mac, 
Emacs, and Brief.  The emacs is somewhat powerful, but not as much as I 
would like.  It is extendable, however.  It's major downfall is that it only 
allows files that are shorter than 32K.  If you hit the magic limit, it just 
ignores input!  (The cause is the use of a Windows edit widget as the basis 
of the editor.)

Both seem to offer good support for native window look-and-feel 
generation.  MCL's use of views makes the organization much easier to 
deal with.  ACLPC uses rectangles which unlike views cannot be subclassed 
or the basis for specialized methods.

MCL supports logical pathnames and there are a number of defsystems 
that work with it.   ACLPC does not support either although I've converted 
Kantrowitz's logical pathnames to ACLPC and am in the process of 
converting his defsystem as well.

MCL is fast.  ACLPC even on a 486 66MHz is not fast as a development 
environment.  The code seems acceptably fast if you're doing GUI 
routines primarily because Windows limits the speed for all applications.  
I don't know about compute intensive code yet; however, the poor quality 
of the compiler doesn't bode well.

MCL's debugger is very good.  It allows full investigation of stack frames, 
restart from arbitrary frame, integration with the inspector, etc.  
ACLPC's is practically non-existant.  It lists the frame buffers but has 
no inspect or restart capability.  You can disassemble code in MCL and look 
for the current debugger position in the disassembled code to aid debugging. 
You cannot disassemble code in ACLPC.

Official product support for MCL from Lucid stinks; however, the MCL 
developers are great.  They provide vigorous, thorough, and patient email 
based support via info-mcl and bugs-mcl.  info-mcl is wonderful as a 
general community conversation and mutual support society.  ACLPC's 
official support is wonderful.  They've been very responsive.  They seem 
genuinely interested in making the most of their product.  They do run a 
bbs; however, I've seen zero traffic on it; so, there's nothing comparable 
to info-mcl for ACLPC.

Well, that's all I can think of without actually looking at anything :-)  If 
you have any specific questions, don't hesitate to ask. I want to learn too.

--------------------------------------
Date: 9/13/93 4:03 AM
To: Don Mitchell
From: Youngcook Jun
Date: 12 Sep 1993 20:14:14 GMT
Organization: University of Illinois at Urbana

Can anybody tell me what are differences between mcl 2.0 and
acl for ibm pc or IBM RISC/6000?


---------------------
Donald H. Mitchell             Domain:  dhm@pro-solution.com
Proactive Solutions Inc.         UUCP:  uunet!proact!dhm   
5314 S. Yale Ave., Suite 402    Voice:  918.492.5192
Tulsa, OK  74135                  FAX:  918.492.5193