Re: [Gcl-devel] pathname-name behavior change

[Camm Maguire]

Greetings, and thanks Tim for isolating the issue.

Rest assured that we will support the old behavior in our CLtL1
build/mode, so that axiom can change to ansi if and when it wants.
This is the raison d'etre of this build, which I consider a valuable
GCL feature.  Much of the heavy lifting in the lisp world has been
done under CLtL1 or the like, and we need to be ready to catch these
gems as they might propagate into the open source world.

This said, what we really mean by our CLtL1 build is a small build as
close to the original GCL as practicable which doesn't break anyone's
existing code.  There may be some ansi features we implement in this
build which don't happen to break such code.  The development process
in general is move to ansi in as straight a line as possible, see what
breaks when done, and revert the mods as necessary in CLtL1.  

This is much like gcc's support for K&R as well as ansi C, selected by
command line options.

I'll let you know when I pinpoint the location of modification and
commit a revert in the case of CLtL1.

BTW, axiom is already making use of this duality in its in-package
usage, which under ansi must trigger an error if the package does not
already exist.  We support the older behavior in CLtL1.

Take care,

root <address@hidden> writes:

> Camm,
> 
> I traced the problem in the GCL CVS HEAD that I tried to use.
> This is a check of a broken axiom system vs a working axiom using 2.6.5
> I know that CVS HEAD is not ready yet but this change will surely 
> break axiom.
> 
> 
> ====================================================================
> first we trace the failing axiom function
> ====================================================================
> (1) -> )lisp (trace localdatabase)
> 
> Value = (LOCALDATABASE)
> (1) -> )library ZMOD
> 
>   1> (LOCALDATABASE (ZMOD) NIL)
>    )library cannot find the file zmod.
>   <1 (LOCALDATABASE #<hash-table 08508d58>)
> 
> ====================================================================
> then we narrow it down to the failing lisp function
> ====================================================================
> (1) -> )lisp (trace pathname-name)
> 
> Warning: PATHNAME-NAME is being redefined.
> Value = (PATHNAME-NAME)
> 
> ====================================================================
> we load the interpreted localdatabase code
> ====================================================================
> (1) -> )lisp (load "/tmp/axiom/int/interp/daase.lisp")
> 
> Value = T
> 
> ====================================================================
> and try the failing function again. Notice that given a symbol the
> pathname-name function returns a lowercase string
> ====================================================================
> (1) -> )library ZMOD
> 
>   1> (LOCALDATABASE (ZMOD) NIL)
>     2> (PATHNAME-NAME ZMOD)
>     <2 (PATHNAME-NAME "zmod")
>    )library cannot find the file zmod.
>   <1 (LOCALDATABASE #<hash-table 08508d58>)
> 
> ====================================================================
> in a working axiom we do the same thing but given a symbol the
> pathname-name function returns an uppercase string
> ====================================================================
> (1) -> )library ZMOD
> 
>   1> (LOCALDATABASE (ZMOD) NIL)
>     2> (PATHNAME-NAME ZMOD)
>     <2 (PATHNAME-NAME "ZMOD")
>    IntegerMod is now explicitly exposed in frame initial 
>    IntegerMod will be automatically loaded when needed from 
>       /tmp/axiom/int/algebra/ZMOD.NRLIB/code
>   <1 (LOCALDATABASE #<hash-table 0837e118>)
> 
> (1) -> 
> 
> Tim
> 
> 
> _______________________________________________
> Gcl-devel mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/gcl-devel
> 
> 
> 

-- 
Camm Maguire                                            address@hidden
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah