Re: FYI: libiberty's maintainer-clean breaks if SUBDIRS is empty

[Alexandre Oliva <aoliva at cygnus dot com>]

On Apr 29, 2000, msokolov@ivan.Harhan.ORG (Michael Sokolov) wrote:

> Alexandre Oliva <aoliva@cygnus.com> wrote:

>> -	@if test "x$(SUBDIRS)" = x; then exit 0; fi; \
>> +	@subdirs='$(SUBDIRS)'; \
>> target=`echo $@ | sed -e 's/-subdir//'`; \
>> -	for dir in $(SUBDIRS) ; do \
>> +	for dir in $$subdirs ; do \

> It was my earlier patch that made it work the way it did. It was
> necessary as what was there before my patch failed on 4.3BSD. It
> still works on 4.3BSD after your patch, so I'm not complaining, but
> why was your patch necessary?

Because some shells will parse the whole sequence of commands before
evaluating them.  Bash, for example, will complain when if sees:

if test "x" = x; then exit 0; fi; ...; for dir in  ; do ...
                                                  ^
this can't be parsed, even if it is never going to be executed.

subdirs=''; ...; for dir in $subdirs; do ...
                            ^^^^^^^^

Now this can be parsed correctly, and will do The Right Thing (TM)
even if $subdirs expands to nothing.

> Why did my way break for maintainer-clean-subdir?  And how is
> maintainer-clean-subdir different from any other *-subdir?

It is not different, I had just failed to notice that all *-subdir
commands shared the same rule.  It's just that some of the existing
clean rules call $(MAKE) some-other-clean, setting SUBDIRS="", and so
they fail.

-- 
Alexandre Oliva    Enjoy Guaranį, see http://www.ic.unicamp.br/~oliva/
Cygnus Solutions, a Red Hat company        aoliva@{redhat, cygnus}.com
Free Software Developer and Evangelist    CS PhD student at IC-Unicamp
oliva@{lsd.ic.unicamp.br, gnu.org}   Write to mailing lists, not to me