Thank you for your comment on Friday 11/23/07. As you said, putting out.txt in a subdir does not exhibit the problem -- that's right. However, I encountered that $(wildcard ) problem in my real-world makefiles when out.txt(other name may be used though) is put in a subdir within current dir. Since my real-world working makefiles(GnumakeUniproc) are quite complicated so that it is unsuitable to paste it here as an example, I tried to dig out the simplist case that show my problem(as my post here on Wednesday 10/03/07).


However, beside from using GnumakeUniproc, I find it's hard to forge an example that can demonstrate the $(wildcard ) problem when the file being tested by $(wildcar ) is in a subdir.

Anyhow, I think it should be a problem to fix.

(Attachment is a makefile that does show this problem, for memorial purpose.)

(file #14489)

    I encountered a strange situation recently, $(wildard ) function seems failing to work. See my code below:

========================================================
all: first_target

outfile = out.txt

$(outfile):
@echo nnn > $@

.PHONY: prj_count
prj_count:
@echo "[[$(outfile)]] <<$(wildcard $(outfile))>> "

.PHONY: first_target
first_target: $(outfile) prj_count
@echo "[DONE]"
========================================================

First, delete out.txt and run make on that makefile, the output is
 
[[out.txt]] <<>>
[DONE]

then, run make again, the output is
 
[[out.txt]] <<out.txt>>
[DONE]

Result of the first run is strange. When target prj_count is being made, out.txt should have been generated on my disk(because $(outfile) had been made before), then WHY $(wildcard out.txt) gives me null result ?
 
I've tested it on make 3.80 & 3.81 on Linux and on Windows(MSVC-built), both produce the same result.

Reference mailling list item: http://lists.gnu.org/archive/html/help-make/2007-09/msg00060.html