moving testsuite 1 level up

Egor Duda
Fri Sep 1 00:32:00 GMT 2000


Friday, 01 September, 2000 Chris Faylor wrote:

>>DD> Perhaps it's time to update Maekfile.common to use some $(UPTOP)
>>DD> (defined to ".." in most cases) variable?
>>i  wanted  to  to  that but stumbled over code in Makefile.common that
>>sets updir* and bupdir* vars. i don't understand why not just set them
>>.. , ../.. and ../../..

CF> They are set to that in some cases.

i've   never seen the case when pwd don't return an absolute path. but
i agree that relying on it isn't good.

>>and  don't  feel  i  can  mess with the code i don't fully understand.
>>maybe someone enlighten me on this subject?

CF> It's because I don't like to see ../../../../libbfd.a on a command line.
CF> I'd rather see /foo/bar/bfd/libbfd.a and we can rely on the fact that
CF> gnumake is available for filename parsing.

i see. as for $(UPTOP), maybe instead of

bupdir1:=${patsubst %:::,%,${patsubst %/:::,%,$(dir $(bupdir)):::}}

we should define

broot:=${shell cd $(UPTOP)/..; pwd}
btargetroot:=${shell cd $(UPTOP); pwd}
srcroot:=${shell cd $(srcdir)/$(UPTOP); pwd}

and then (in the individual's) we'll use not paths with
$(bupdir2),  $(bupdir1)  but  paths  relative  to  $(btargetroot)  and
$(srcroot )or (in few occasions) to $(broot)

Egor.   ICQ 5165414 FidoNet 2:5020/496.19

More information about the Cygwin-developers mailing list