A number of projects fail to build with the latest ASDF from git. A few of them seem to be related to defining a custom file component class that has a custom file type. For example, see jwacs and portable aserve:
http://report.quicklisp.org/jwacs/2012-12-19/failtail.txt
http://report.quicklisp.org/portableaserve/2012-12-19/failtail.txt
For jwacs, the file component has a type of "js", and for portableaserve, it has a type of "cl", but it seems like ASDF is looking for different file names.
FSet also fails with this error:
Error while trying to load definition for system fset from pathname /[...elided...]/fset-20121219-svn/fset.asd:
Invalid initialization argument: :PARENT in call for class #<STANDARD-CLASS SYSTEM>.
Zach
Zach Beane xach@xach.com writes:
A number of projects fail to build with the latest ASDF from git. A few of them seem to be related to defining a custom file component class that has a custom file type. For example, see jwacs and portable aserve:
[snip]
CommonQT's failure:
qt.asd: The slot TYPE is unbound in the object #<MAKEFILE "qt" "commonqt.pro">.
Zach
My bad. I tried to remove the default type on source-file; obviously, it's too late and many systems use it already. Restored it to nil in 2.26.35.
For next time I'm ill-inspired like that, is there a way to load everything *that's suppose to work* in quicklisp? I found I could try to load everything with: (time (dolist (s (quicklisp-client::provided-systems t)) (quicklisp:quickload (ql-dist:name s)))) but it fails with things like "abort-ks-execution-example".
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org When all lawful citizens are disarmed, will we have an omnipresent police state to protect us from armed criminals?
On Wed, Dec 19, 2012 at 9:30 AM, Zach Beane xach@xach.com wrote:
Zach Beane xach@xach.com writes:
A number of projects fail to build with the latest ASDF from git. A few of them seem to be related to defining a custom file component class that has a custom file type. For example, see jwacs and portable aserve:
[snip]
CommonQT's failure:
qt.asd: The slot TYPE is unbound in the object #<MAKEFILE "qt" "commonqt.pro">.
Zach
Faré fahree@gmail.com writes:
My bad. I tried to remove the default type on source-file; obviously, it's too late and many systems use it already. Restored it to nil in 2.26.35.
For next time I'm ill-inspired like that, is there a way to load everything *that's suppose to work* in quicklisp? I found I could try to load everything with: (time (dolist (s (quicklisp-client::provided-systems t)) (quicklisp:quickload (ql-dist:name s)))) but it fails with things like "abort-ks-execution-example".
Loading everything into the same image isn't practical. I wrote qlmapper to do it instead:
https://github.com/xach/qlmapper
You still must have some extra stuff installed, e.g. foreign libraries and tools, but it should allow you to test many systems.
Zach
Faré fahree@gmail.com writes:
My bad. I tried to remove the default type on source-file; obviously, it's too late and many systems use it already. Restored it to nil in 2.26.35.
Thanks. What about the fset :parent issue?
Zach
Faré fahree@gmail.com writes:
My bad. I tried to remove the default type on source-file; obviously, it's too late and many systems use it already. Restored it to nil in 2.26.35.
FYI, the projects I mentioned still fail with 2.26.35, with the same symptoms.
Zach
The FSet failure is due to its using (:system "misc-extensions") amongst its components. That's always been invalid, I wonder how it could ever possibly work, and I'm not willing to kluge ASDF into keeping supporting it. misc-extensions is already present in the depends-on, which is the correct place for it to be.
Scott, can you fix that? Just remove that bogus component.
PortableAserve and Jwacs and probably many others were failing because they use an ASDF1-only recipe for defining source-file type, unhappily propagated by the asdf.texinfo itself (mea culpa), which broke when I made system not inherit from module anymore. Sigh. I reluctantly restored system as a subclass of module for now (2.26.36), but I'll be contacting the authors so they fix their system definitions.
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org For pretty much every writer, the big problem isn't piracy, it's obscurity. — Tim O'Reilly, as cited by Cory Doctorow
Scott, here's a trivial patch that makes fset compile to me.
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org Suppose 2/3 Congress were incinerated, what would we lose to offset our gain of their salaries and the salaries of their parasites? — H. L. Mencken
-
Faré -
Zach Beane