[Openmcl-devel] Hunchentoot and CCL on Windows XP

[Gary Byers]

On Thu, 19 Mar 2009, Sean Ross wrote:

>
> Is there a way to detect a lack of support for timeouts, I know it's a long 
> shot but a patch which
> would gracefully degrade when CCL was patched would be a great thing.
>
>
> Cheers,
> Sean.
>

CCL's mostly intended to be patched via svn e.g., someone will patch
the windows sockets timeout stuff and commit that change. That change
to the source will have an svn revision number associated with it. If
someone subsequently does an "svn update" and "(rebuild-ccl)", then
the svn revision that was current should show up in
LISP-IMPLEMENTATION-VERSION (and elsewhere.)

In theory, it'd be easy to "check for the patch" by doing something
like

(if (< *svn-revision-of-current-lisp* *svn-revision-of-commit-with-fix*)
   (work-around-bug))

but some things are cruftier than they need to be to make that easy
(the lisp's notion of *svn-revision-of-current-lisp* is currently
a string that can contain other info besides the svn revision, it's
necessary to know what branch the lisp was built on and what branch
the fix was committed on, etc.)

I'd be all for making that work in practice and making it easier
to test just what the revision level/branch/etc of the running
lisp is, so that it'd be practical to test to see if fixes have been
applied.  I'm less enthused about the idea of having a separate
patch mechanism.