[asdf-devel] Unable to upgrade asdf using clisp 2.49 on Mac OS X
On Mac OS X, when I try to upgrade my ASDF using the logical pathnames "home:lisp;asdf;" and "home:lisp;asdf;uiop;" I get this error: *** - While searching for system "asdf": "home:lisp;asdf;uiop;" evaluated to "home:lisp;asdf;uiop;" which is not an absolute directory. Note that I am using ASDF:*CENTRAL-REGISTRY*, not the configuration DSL. But: Break 1 [2]> (translate-logical-pathname "home:lisp;asdf;uiop;") #P"/Users/rpg/lisp/asdf/uiop/" Is this expected behavior from clisp + ASDF? This same configuration works on ACL and SBCL, which suggests its a clisp issue.
On Sat, Feb 1, 2014 at 11:02 PM, Robert P. Goldman <rpgoldman@sift.info> wrote:
On Mac OS X, when I try to upgrade my ASDF using the logical pathnames "home:lisp;asdf;" and "home:lisp;asdf;uiop;" I get this error:
*** - While searching for system "asdf": "home:lisp;asdf;uiop;" evaluated to "home:lisp;asdf;uiop;" which is not an absolute directory.
Weird. It looks like an absolute directory to me indeed. Was it parsed properly? Did you forget to use #p"...", or somehow got the pathname parsed before the logical pathname was configured?
Note that I am using ASDF:*CENTRAL-REGISTRY*, not the configuration DSL.
But:
Break 1 [2]> (translate-logical-pathname "home:lisp;asdf;uiop;") #P"/Users/rpg/lisp/asdf/uiop/"
Is this expected behavior from clisp + ASDF?
This same configuration works on ACL and SBCL, which suggests its a clisp issue.
Can you publish steps to reproduce? This works for me on CLISP 2.49+ (2010-07-17) (built 3590935887) (memory 3590936564), which I believe is actually a recently compiled CLISP from hg. (setf (logical-pathname-translations "home") '(("**;*.*" #p"/home/tunes/**/*.*"))) (translate-logical-pathname "home:cl;asdf;uiop;") (describe #p"home:cl;asdf;uiop;") (push #p"home:cl;asdf;uiop;" asdf:*central-registry*) (push #p"home:cl;asdf;" asdf:*central-registry*) (asdf:locate-system "asdf") (setf asdf:*resolve-symlinks* nil) (asdf:locate-system "asdf") (asdf:load-system "asdf") —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org Nothing is particularly hard if you divide it into small jobs. — Henry Ford
On Feb 1, 2014, at 22:25, Faré <fare@tunes.org> wrote:
On Sat, Feb 1, 2014 at 11:02 PM, Robert P. Goldman <rpgoldman@sift.info> wrote: On Mac OS X, when I try to upgrade my ASDF using the logical pathnames "home:lisp;asdf;" and "home:lisp;asdf;uiop;" I get this error:
*** - While searching for system "asdf": "home:lisp;asdf;uiop;" evaluated to "home:lisp;asdf;uiop;" which is not an absolute directory. Weird. It looks like an absolute directory to me indeed. Was it parsed properly? Did you forget to use #p"...",
I didn't use #p when setting ASDF:*CENTRAL-REGISTRY*, but I don't use it on SBCL or ACL, either. The central registry has always accepted namestrings for me, with this one exception... Cheers, R
On Mac OS X, when I try to upgrade my ASDF using the logical pathnames "home:lisp;asdf;" and "home:lisp;asdf;uiop;" I get this error:
*** - While searching for system "asdf": "home:lisp;asdf;uiop;" evaluated to "home:lisp;asdf;uiop;" which is not an absolute directory. Weird. It looks like an absolute directory to me indeed. Was it parsed properly? Did you forget to use #p"...",
I didn't use #p when setting ASDF:*CENTRAL-REGISTRY*, but I don't use it on SBCL or ACL, either. The central registry has always accepted namestrings for me, with this one exception...
I suppose sysdef-central-registry-search and/or probe-asd could ensure that defaults is parsed as a pathname, using (pathname ...) or some invocation of (ensure-pathname ...). I leave that to you. —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org If you were plowing a field, which would you rather use? Two strong oxen or 1024 chickens? — Seymour Cray
Thanks. I'll have a look at fixing this. Sent from my iPad On Feb 1, 2014, at 23:30, Faré <fare@tunes.org> wrote:
On Mac OS X, when I try to upgrade my ASDF using the logical pathnames "home:lisp;asdf;" and "home:lisp;asdf;uiop;" I get this error:
*** - While searching for system "asdf": "home:lisp;asdf;uiop;" evaluated to "home:lisp;asdf;uiop;" which is not an absolute directory. Weird. It looks like an absolute directory to me indeed. Was it parsed properly? Did you forget to use #p"...",
I didn't use #p when setting ASDF:*CENTRAL-REGISTRY*, but I don't use it on SBCL or ACL, either. The central registry has always accepted namestrings for me, with this one exception... I suppose sysdef-central-registry-search and/or probe-asd could ensure that defaults is parsed as a pathname, using (pathname ...) or some invocation of (ensure-pathname ...). I leave that to you.
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org If you were plowing a field, which would you rather use? Two strong oxen or 1024 chickens? — Seymour Cray
Faré wrote:
On Mac OS X, when I try to upgrade my ASDF using the logical pathnames "home:lisp;asdf;" and "home:lisp;asdf;uiop;" I get this error:
*** - While searching for system "asdf": "home:lisp;asdf;uiop;" evaluated to "home:lisp;asdf;uiop;" which is not an absolute directory. Weird. It looks like an absolute directory to me indeed. Was it parsed properly? Did you forget to use #p"...", I didn't use #p when setting ASDF:*CENTRAL-REGISTRY*, but I don't use it on SBCL or ACL, either. The central registry has always accepted namestrings for me, with this one exception...
I suppose sysdef-central-registry-search and/or probe-asd could ensure that defaults is parsed as a pathname, using (pathname ...) or some invocation of (ensure-pathname ...). I leave that to you.
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org If you were plowing a field, which would you rather use? Two strong oxen or 1024 chickens? — Seymour Cray
This is only fixable going forward, right? Because it's a bug that affects the upgrade.... I guess this means that clisp is more fussy about accepting a pathname, rather than a pathname-designator in some circumstances. cheers, r
This is only fixable going forward, right? Because it's a bug that affects the upgrade....
Old versions of ASDF may or may not like logical pathnames so much, anyway.
I guess this means that clisp is more fussy about accepting a pathname, rather than a pathname-designator in some circumstances.
Debugging (logical) pathnames is something I really don't want to do anymore, but with the help of trace, you can probably figure out where the bug is happening. —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org Science is like sex: sometimes something useful comes out, but that is not the reason we are doing it — Richard Feynman
participants (2)
-
Faré -
Robert P. Goldman