Re: [Gcl-devel] Build failing with --enable-ansi

[Paul F. Dietz]

Camm Maguire wrote:
> Hi Paul!  Yes, I saw this, and it should now be fixed.  You must have
> caught the CVS update in between two rapid revisions last night.  My
> apologies.  

Thanks.  I noticed early on that when the build failed it left some
files that were not removed by 'make clean'.  clcs/precom.{h|c|data}
was among them.  If anyone is having problems building they should
do a make clean followed by cvs -q update, and look for files that
shouldn't be there.


> BTW, could either you or Peter elaborate on what is needed for typep,
> subtypep, and coerce regarding the recent structure test failures?  I
> tried Peter's experimental fix, but it led to a value stack overflow.

I am not familiar with the implementation, but the effect should
be that classes are acceptable as type specifiers in any of these
functions (or, indeed, anywhere a type specifier is called for.)


I added many new tests this morning that fail.  Many more will
be coming -- I'm going to check that all the builtin functions
that take keyword arguments also accept :allow-other-keys <true>,
which I don't think many of them do.

        Paul