Hi Mark,
Dunno if I'll be making it to our monday progress meeting tomorrow (travelling to be done). But this is my status:
* My work on DEFINE-METHOD-COMBINATION can be considered Done; I left some todo's behind which should be relatively small in comparison to the original problem. It's now just a matter of changing the symbols in the backquote patterns. * My work on the autoloader has been mostly done as far as our current infrastructure allows; next steps: - Clean autoload.lisp for any symbols which automatically end up in autoload-gen.lisp - Create infrastructure for autoloading of SETF functions (ie function names not designated by a symbol) * Crashes evaluating a form referencing an undefined function in a clean image --> solved by requiring PRINT-OBJECT in the right place With respect to this point: it's possibly a good idea to do a search of the code base and find out if we're doing this everywhere PRINT-OBJECT methods are being defined * COMPILER-UNSUPPORTED-FEATURE-ERROR derived from CONDITION --> solved * Autoloader fails when *READ-xxx* variables bound to unexpected values --> solved * Implement USE-FAST-CALLS properly --> resolved by deleting the USE-FAST-CALLS infrastructure from the interpreter; the compiler already does the right thing * Compiling #n# causes stack overflow --> solved; however, Stas Boukarev reports that the case I've fixed only fixes in-memory compilation and the issue still exists in file compilation. Ticket #191 should be re-opened to and should say so.
All in all a pretty good result for one week. However, from here on $WORK will start again which means I'll only be able to address the bigger issues in small steps. Or just the smaller issues, of course. Which is why I'm glad to have worked out the D-M-C stuff. That was a week more than full time work to completely comprehend the spec and macrology going on in our implementation.
That's it for now. If I'm available tomorrow at 21:00 CET, I'll be on #abcl.
Bye,
Erik.