Re: [Gcl-devel] proposed find-restart bugfix

[Peter Wood]

On Sun, Oct 20, 2002 at 07:16:58PM -0400, Camm Maguire wrote:
> Hi Peter!  Thanks for looking into this!
> 
> 1) I added an optional condition argument to compute-restarts a little
>    while ago to address some test failures Paul was seeing.  It was
>    just a placeholder argument, i.e. it is currently ignored, as there
>    is no 'slot' in the restart structure yet to support it.  This
>    would be straightforward to add, but I'd let this mini-project wait
>    for a while due to 2).  This is why the item is still active on the
>    web site.
> 
> 2) There appears to be a possible lexical scope issue with the
>    *restart-clusters* handling.  I put in some work around to
>    clcs/handler.lisp so that Paul's tests could detect a restartable
>    ('correctable') package error, but was a bit confused as to why this
>    should be necessary.  Paul had written that I should ignore the
>    restart code for now as he was unsure if the tests were accessing this
>    information correctly.
> 
> 3) The background discussions should still be accessible via the recent
>    mailing list archives.
> 
> 4) This having been said, your patch looks fine to me, but will
>    obviously need to be supplemented as in 1) and 2).  You might want
>    to slightly modify to something like this for readability:
> 
>  (defun find-restart (name &optional condition)
>        (let ((restarts (if condition (compute-restarts condition) 
> (kcl-top-restarts))))
>          (dolist (restart restarts)
>            (when (or (eq restart name)
>                      (eq (restart-name restart) name))
>              (return-from find-restart restart)))))
> 
> If you would like to commit either one, that would be fine with me.
> Please note that should some analogous patch be committed to a lsp
> file in the lsp or cmpnew directories, the corresponding .c, .h, and
> .data files would have to be remade and committed as well.  In this
> case, the lisp patch alone will suffice given the way the build is
> currently structured.
> 

Hi,

I think I'd better have a closer look at the conditions stuff, and reread
the background discussion before I do anything.

Regards,
Peter