>>>>> "Daniel" == Daniel Barlow <[EMAIL PROTECTED]> writes:
Daniel> "Christopher C. Stacy" <[EMAIL PROTECTED]> writes:
>> Is there something I should be using that's not ASDF?
>> Or are you stuck using this ASDF physical pathname solution at the moment?
Daniel> I can't find the specific mail that I wanted to repsond to right now,
Daniel> but this one will do.
Daniel> Various people have asserted that ASDF doesn't work with LPNs. For
Daniel> the record, if this is true it's not intentional. It is possible
Daniel> (and, I maintain, easiest) to use asdf without absolute pathnames at
Daniel> all (thus eliminating most of the need for LPNs) and I think most
Daniel> people are probably using it that way.
Daniel> I'll see if I can do something with your bug report this evening or
Daniel> over the weekend, depending on when I have time to install CMUCL.
FWIW, Chisheng Huang sent the following test case to me in private. I
hope he doesn't mind me posting it here.
1. In a Unix shell:
mkdir /tmp/systems
cat /dev/null > /tmp/systems/v.asd
2. In CMUCL (mine is CMU Common Lisp CVS snapshot 2003-12):
CL-USER> (setf (logical-pathname-translations "v")
'(("**;*.*" "/tmp/**/*.*")))
(("**;*.*" "/tmp/**/*.*"))
CL-USER> (MAKE-PATHNAME :DEFAULTS "v:systems;"
:VERSION :NEWEST
:NAME "v"
:TYPE "asd"
:CASE :LOCAL)
#.(logical-pathname "V:SYSTEMS;v.asd")
CL-USER> (probe-file *)
NIL
CL-USER> (probe-file #.(logical-pathname "V:SYSTEMS;v.asd"))
#p"/tmp/systems/v.asd"
CL-USER>
So, I think what is happening is the make-pathname with a logical path
and :case :local, is preserving the case of everything, including the
uppercase parts of the logical path.
Perhaps the right thing is to convert any lower case parts to upper
case when the pathname is a logical pathname.
I have not yet examined the CLHS to see what it says about this.
Ray
