[Omake] Another recursive omake question

[Benjamin Pierce]

Following up the discussion earlier today...  Aleksey's description  
of how phony targets are treated got me wondering how one would deal  
with a situation where there is some subdirectory that is currently  
broken -- i.e., if we do 'omake' in that directory, we will get an  
error.  But the whole system is working (this subdirectory is some  
external tool, say, that  is distributed with the rest of the package  
and uses its facilities, but that is not needed for everything else  
to build fine.)

Now, what should the OMakefiles look like?  From what I understand...

1) If the parent directory of this broken directory lists it among  
its .SUBDIRS dependencies, then anytime we do 'omake' anywhere higher  
in the directory hierarchy, we're going to attempt to build the  
broken one and fail.  But...

2) If the parent directory does NOT list the broken one as a subdir,  
then we can't do 'omake' in the broken one, so we can't work on it to  
fix it.

Of course, we can edit the parent directory's OMakefile when we start  
working on the broken directory and edit it back when we're finished  
for the day, if we haven't gotten things working yet.  But is this  
the best we can do?  (It's not very nice because, for example, it  
means we can't do 'omake clean' at the top and expect absolutely  
every subdirectory to get tidied up.)

Actually, I guess I see another way that is better than any of these:  
in the OMakefile of the broken directory, change the .DEFAULT target  
to DEFAULTBROKEN or something.  Now, when working there, we have to  
remember to do "omake DEFAULTBROKEN", but this is perhaps not so bad...

Regards,

      - Benjamin