Re: deferred evaulation of function calls

[Noel Yap]

Technically, I'm not sure what the output should be since the order in which the 
dependencies will be made is undefined.  This is clearer to see if you consider that 
"-j" can be used.

Here's a better-defined makefile:

$ cat GNUmakefile
def = $(wildcard *.aoeu)
imm := $(wildcard *.aoeu)

.PHONY: main
main: t2

.PHONY: t2
t2: clean
        ls *.aoeu
        @echo "wildcard = $(wildcard *.aoeu)"
        @echo "imm = $(imm)"
        @echo "def = $(def)"

.PHONY: clean
clean: t1
        rm *.aoeu

.PHONY: t1
t1: touch
        ls *.aoeu
        @echo "wildcard = $(wildcard *.aoeu)"
        @echo "imm = $(imm)"
        @echo "def = $(def)"

.PHONY: touch
touch:
        touch aoeu.aoeu


$ gmake
touch aoeu.aoeu
ls *.aoeu
aoeu.aoeu
wildcard =
imm =
def =
rm *.aoeu
ls *.aoeu
ls: *.aoeu: No such file or directory
gmake: *** [t2] Error 1


Judging from the above, what Paul said about the wildcard caching is the most 
likely suspect.

Noel
Robert P. J. Day wrote:

> On Thu, 27 May 2004, Noel Yap wrote:
>
>
> > ':=' evaluates the RHS at the time of the assignment.  '=' evaluates the
> > RHS at the time of actual use.
>
>
> but in my original post, i did mention that my tests seemed to show no
> difference.  here's a sample makefile i used, in a directory with at least 
> a couple of .o files:
>
> --------------------
>
> .PHONY: main t1 t2 clean
>
> imm := $(wildcard *.o)
> def = $(wildcard *.o)
>
> main:   t1 clean t2
>
> t1:
>         @echo imm = ${imm}
>         @echo def = ${def}
>
> clean:
>         rm *.o
>
> t2:
>         @echo imm = ${imm}
>         @echo def = ${def}
>
> ---------------------
>
>   this is invoked in a directory with some object (*.o) files.  based on 
> what you just said, what do you expect as the output?  i got the 
> following:
>
> imm = a.o b.o
> def = a.o b.o
> rm *.o
> imm = a.o b.o
> def = a.o b.o  ????? why?
>
>   given that "def" was a deferred assignment, i would have thought that 
> its second value would be the empty string since, at the "time of actual 
> use", as you put it, there are no .o files left.
>
>   or am i misreading something?
>
> rday