[Openmcl-devel] Re: Status of Cocoa under Tiger?

[Gary Byers]

On Wed, 11 May 2005, Raffael Cavallaro wrote:

>
> On May 11, 2005, at 8:17 PM, Gary Byers wrote:
>
>> As far as I know, the CVS server works.
>> 
>> The bleeding-edge tree is pretty volatile these days (keep telling 
>> yourself:
>> "that's a GOOD thing.")  I try to put up bootstrapping images every few 
>> days.
>

It's also a little buggy ... that's not necessarily a good thing, but it's
unfortunately hard to avoid.

> Yes, CVS is working now, and I've gotten as far as compiling hemlock after 
> building from sources using the latest boot image.
>
> However, in the middle of executing (ccl::compile-hemlock) I get:
>
> ;Loading #P"ccl:hemlock;bin;openmcl;echocoms.dfsl.newest"...
>> Error in process listener(1): value #<SIMPLE-BIT-VECTOR 256> is not of the 
> expected type SIMPLE-VECTOR.
>> While executing: HEMLOCK-INTERNALS::%SET-CHARACTER-ATTRIBUTE
>> Type :POP to abort.
> Type :? for other options.
>
> Any ideas?

Well, the compiler was sometimes misinterpreting array type
declarations that didn't specify an element type as if they specified
an element type of T, and there may have been other related
problems/secondary effects of this.  (The code that prints arrays was
miscompiled under this assumption, so printing certain arrays with
specialized element types failed.)

Apparently, there aren't too many other things in the lisp itself
that call AREF on specialized arrays without declaring the specialized
array's element type (or at least those other things were failing
quietly.)

I'll try to check in the fix (and a bunch of other potentially
destabilizing changes ...) and upload new bootstrapping images soon.

It's good if people who're interested in doing so can test these
images, but it can be a frustrating experience.  I want to try to make
several other potentially destabilizing changes before the next
release (to make things a bit more portable, mostly).  In practice,
most of the (nearly daily) bootstrapping images in the testing
directory have had very little testing (at least from me: if the lisp
compiles itself without errors or crashes, that's about it ...) and,
in practice, some are more stable/useful than others.  It might be
possible to identify those in some way, e.g, "the Cocoa demo is
believed to work in 050511".

I did check things in a minute ago, and the Cocoa demo is believed
to work in 050511.  (At least, if everything's freshly compiled in
050511.)