look at this "bug"...

[Dr. Jörn von Holten]

Jup, its me again, and I found another one. Maybe I'm a bit too
busy with this and maybe I have problems to scan the whole bug
archive to check if my problems already have been noted. Thus...

I currently try to use make with many many many nested defines - wouldnt
it be nice to be able to write e.g. Python scripts inside of
Makefiles instead of this own define-endef thing; but thats another
side of the coin that shouldn't be that difficult (as an optional plug-in).
I currently try to code a make-framework that is more easy to use and
where the users do not have to think alot (especially if they want to
do complex things like CORBA-idl-generations and alike).
Thats why I found this one (shortened as usually):

define D1
  $(1)$(2)
endef

define D2
  $(call D1,$(1))
endef

all: ; @echo $(call D2,1,a)

Suprisingly to me the result is "1a" and not "1". My idea is that D1
gets 2 parameters and all that are not passed (as in D2) should be
just empty and not "borrowed" from somewhere.
I personally can live with that (currently) but one actually has to
be very carefully and write

define D2
  $(call D1,$(1),)
endef

to supress the "borrowing".

cheers, Jörn

P.S.: A public developent-source-distribution would be nice. I have
problems to reach the CVS rep of gmake - actually it does not work due
whatever reason.


--
Dr. Jörn von Holten
Senior Consultant

Barco Orthogon AG
Hastedter Osterdeich 222
D-28207 Bremen
Tel +49-421-20122-426
Fax +49-421-20122-999
www.barco-orthogon.com
address@hidden