New build scripts can crash even in -no-stop

Bug #1532464 reported by Robert P. Goldman
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
ASDF
Fix Released
Critical
Faré

Bug Description

A failure in SBCL caused make-all-no-upgrade-no-stop to crash, which should never happen:

Now starting sbcl, loading the script support, loading ASDF from source, then exiting...
sbcl --noinform --no-userinit --no-sysinit --disable-debugger --eval "(quote(#.#.\`(load,(string\`|test/script-support.lisp|):verbose():print())#.#.\`(asdf-test::verbose())#.#.\`(load,(string\`|build/asdf.lisp|):verbose():print())#.#.\`(uiop/image:quit,0)))"
0a1,3
> WARNING:
> Implicitly creating new generic function
> (COMMON-LISP:SETF ASDF/ACTION:COMPONENT-OPERATION-TIME).
SUCCEEDED at starting sbcl, loading the script support, loading ASDF from source, then exiting.
Failure:
 In using sbcl: Subprocess (:PROCESS #<EXTERNAL-PROCESS (diff /Users/rpg/lisp/asdf/build/results/sbcl-nop.text /Users/rpg/lisp/asdf/build/results/sbcl-load.text)[74671] (EXITED : 1) #x30200159CDCD>)
 with command ("diff" "/Users/rpg/lisp/asdf/build/results/sbcl-nop.text" "/Users/rpg/lisp/asdf/build/results/sbcl-load.text")
 exited with error code 1
make: *** [test-all-no-upgrade-no-stop] Error 1

Revision history for this message
Robert P. Goldman (rpgoldman) wrote :

I'll try to fix the bug in ASDF/ACTION, which I think is relatively trivial.

But I don't know why this should cause the minimakefile scripts to crash. that should never happen if I ask for no-stop, and never happened before minimakefile, AFAIK.

Revision history for this message
Faré (fahree) wrote :

Maybe minimakefile should use cl-launch to dump an image while asdf is in working state, so breakage of asdf (which is likely to happen during development) doesn't stop the test system from working? The downside is having to rebuild that binary whenever there are changes in tools/ or its dependencies.

In this case of test-all-no-upgrade-no-stop, this looks like a failure of without-stopping...

Revision history for this message
Faré (fahree) wrote :

That was a bug in cl-scripting 0.1's without-stopping, that failed to prevent stopping at the first failure.

I fixed it in cl-scripting 0.2, that asdf-tools is now using. I also added a test suite for cl-scripting 0.2 to prevent regression.

I also added more information on debugging in the README.

I also made the make system create a standalone asdf-tools executable when you first use it, so that you can build asdf-tools when asdf is in good shape, and use it to test asdf even when asdf is in bad shape.

Changed in asdf:
assignee: nobody → Faré (fahree)
milestone: none → asdf3.2
status: New → Fix Committed
Changed in asdf:
milestone: asdf3.2 → 3.1.7
status: Fix Committed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.