Re: [j-devel] recompiling anonymous functions

[Andras S. <as...@ma...>]

On Fri, 27 Aug 2004, Peter Graves wrote:

> On Fri, 27 Aug 2004 at 20:30:01 +0200, Andras Simon wrote:
>> I tried to compile a file with the offending DEFMACRO form with
>> :OUTPUT-FILE in a different directory, but it doesn't seem to make
>> difference (ie., it compiles just fine). Also, this wouldn't explain
>> the nondeterminism.
>
> I just checked in a fix for an obvious error in compile-file.lisp.

I'll try it.

>
> Doing (load "swank-loader") in the slime directory seems to work
> correctly now.
>
> I'm not too worried about the nondeterminism. I think it probably had
> to do with what directory you were in when you started abcl and what
> the value of *DEFAULT-PATHNAME-DEFAULTS* was when you did (load "swank-
> loader"). Running under emacs might have muddied the waters a bit too.

In all cases, I started emacs (and then slime) from the same directory.

>
> I'll take another look at all of this next week.

Thanks!

>
> On the subject of slime, it looks like there might be some weird
> interactions between Gray streams and pretty-printing streams that will
> need to be ironed out as part of the ongoing pretty printer cleanup
> work.

That'd be great!

I've noticed one (new) problem: when there is an error in the
listener, instead of the *sldb* buffer popping up, abcl gets into an
infinite recursion. Replacing PPRINT with PRINT in PRINT-FRAME (in
swank-abcl.lisp) is a workaround for now.

Andras