Vladimir Prus <gho...@cs.msu.su> writes:
David Abrahams wrote:
./jam -fsl.jam -n
I get
cp b /tmp/a && cat /tmp/some-target >> /tmp/a
while I'd expect to see
cp b /tmp/a && cat some-target >> /tmp/a
Anybody knows how to fix this?
Yeah, binding doesn't work that way. When INPUT is bound, the
variables "on" the target(s) whose name it contains are used to bind
it.
Hmm.. I read your words to mean that if input contains target "some-target",
then variables on "some-target" are used for binding of "some-target"? Okay,
then why "SEARCH" variable on "some-target" is ignored?
It isn't. ./some-target is equivalent to some-target. Try setting
SEARCH to something other than "."
Another aspect is that the problem only happens when "some-target"
is not in dependency graph. In that case it's bound after variable
on $(<) are in effect, which cause the strange behaviour.
When I add
DEPENDS $(<) : some-target ;
to the "do" rule, the "some-target" is bound at a different time, and all is
fine.
I'm not sure what you mean by "all is fine". What code do you use,
what behavior do you see, and why do you expect (or not expect) it?
7 messages in com.perforce.jamming[jamming] Bug with bound vars and SEA...
.jam