#37 defsystem add-registry-location failing

closed-accepted
defsystem (13)
5
2003-12-29
2003-03-19
Eric Blood
No

It appears that src/defsystem-3.x/defsystem.lisp was
changed recently to use pushnew, and that change is
using an incorrect keyword (tets instead of test).

Attached is the patch to correct it--but, it should be
apparent.

Discussion

  • Eric Blood
    Eric Blood
    2003-03-19

    patch to defsystem.lisp

     
    Attachments
  • Logged In: YES
    user_id=8016

    adding makefiles and/or *.system files for
    single-file systems seems to be a waste.
    you should be able to compile such systems using
    a simple command like
    ${TOP}/bin/run-lisp -c *.lisp

    if you wish, you may discuss this on clocc-devel

     
    • status: open --> pending-accepted
     
  • Logged In: YES
    user_id=8016

    Ok. Thanks.

    This was entered in a recent set of patches.

    However, I am not so sure that the use of PUSHNEW is correct altogether, as you can conceivably add a PATHNAME with ADD-REGISTRY-LOCATION (hence opening the can of worms of pathname equality).

    Cheers

    Marco

     
  • Logged In: YES
    user_id=8016

    adding makefiles and/or *.system files for
    single-file systems seems to be a waste.
    you should be able to compile such systems using
    a simple command like
    ${TOP}/bin/run-lisp -c *.lisp

    if you wish, you may discuss this on clocc-devel

     
  • Logged In: YES
    user_id=8016

    Sorry, I did not actually send the previous message. Lokks like a SF glitch.

    Anyway. It is true that the Makefile setup in the CLOCC is redundant for single file systems.

    However, note that MK:DEFSYSTEM is os agnostic. It is supposed to be portable everywhere. Nowadays, Windows and Unix filesystems are pretty much standardized, but, just two years ago they were not.

    Cheers

    marco antoniotti

     
    • status: pending-accepted --> open-accepted
     
  • Sam Steingold
    Sam Steingold
    2003-03-21

    Logged In: YES
    user_id=5735

    pushnew is correct (IMO) because equality for patchnames is
    well defined
    I fixed this typo.

     
    • status: open-accepted --> closed-accepted