Russ Allbery wrote:
One problem with install-sh is that it does not preserve mtime. Doesn't
pinstall preserve mtime and avoid the copy if the file didn't change?
Without this feature you would rebuild the world much more often.
If it wasn't for the apparent lack of support for make's %-rules, I'd
suggest even letting make do the determination of whether the call to
whichever *install was needed.
$(DESTBIN)/% : %
$(INSTALL_PROGRAM) $< $@
INSTALLS = $(DESTBIN)/foo # ... $(DESTMAN1)/foo.man too
INSTALL_PROGRAM = $(INSTALL) -s # just an example
install : $(INSTALLS)
uninstall :
$(RM) $(INSTALLS)
However, a number of version of make, especially older ones, only have
the much more limited .x.y rules.
make should do this, not your install program.
I feel like some of install's feature just work around weaknesses in
ancient versions of make.
That being said, if it's the path of least resistance, I can hack
install-sh to support -p and make Autoconf require that the system install
program support -p in order to use it.
But I'd rather get make to solve this problem. It's better at it.
--
C. Alex. North-Keys
Catalis, Inc.
[EMAIL PROTECTED]
512.874.7666
_______________________________________________
OpenAFS-devel mailing list
[email protected]
https://lists.openafs.org/mailman/listinfo/openafs-devel
