Re: Have `make distclean` also remove .i and .s files

[Alexandre Julliard]

Gerald Pfeifer  writes:
> After using GCC's -save-temps option to debug something, I noticed
> that `make distclean` fails to actually remove .i and .s files.
>
> This patch fixes it.  I hope it's the right place to make this change?

It would be, but I don't think we want make clean to remove all
possible output files from all possible tools. The current list is
probably too much already, it should really be limited to files that
are generated as part of the normal build process.

Have `make distclean` also remove .i and .s files

[Gerald Pfeifer]

On Mon, 15 Oct 2007, Alexandre Julliard wrote:
> > After using GCC's -save-temps option to debug something, I noticed
> > that `make distclean` fails to actually remove .i and .s files.
>
> It would be, but I don't think we want make clean to remove all
> possible output files from all possible tools.

I would argue that these aren't output files from some weirdo tool,
but from GCC, at which point you could argue that these files are
not generated with a default ./configure ; make. ;-)
> The current list is probably too much already, it should really be 
> limited to files that are generated as part of the normal build process.

Ugh, in fact that argument is already in your first mail. ;-)  Well,
I figured it's worth a try.

The GNU Coding Conventions (make distclean is a GNU invention as far
as I know) says the following:

distclean: Delete all files in the current directory (or created by
  this makefile) that are created by configuring or building the program.
  If you have unpacked the source and built the program without creating
  any other files, `make distclean` should leave only the files that were 
  in the distribution.

Any chance of you reconsidering this?

Gerald

Have `make distclean` also remove .i and .s files

[Alexandre Julliard]

Gerald Pfeifer  writes:
> The GNU Coding Conventions (make distclean is a GNU invention as far
> as I know) says the following:
>
> distclean: Delete all files in the current directory (or created by
>   this makefile) that are created by configuring or building the program.
>   If you have unpacked the source and built the program without creating
>   any other files, `make distclean` should leave only the files that were 
>   in the distribution.
>
> Any chance of you reconsidering this?

Well, following the GNU conventions would mean removing a lot of
things from the existing make distclean, and I could be persuaded to
do that.

Adding *.i and *.s would be moving in the wrong direction, they are
not generated by building Wine, except if you add special options; but
if you allow this then there is no limit to the amount of things that
we could potentially add.

If you really want to remove everything except tracked files you can
always use git-clean -x (of course you'd need to switch to git first ;-)

Have `make distclean` also remove .i and .s files

[Michael Stefaniuc]

Alexandre Julliard wrote:
> Gerald Pfeifer  writes:
> > After using GCC's -save-temps option to debug something, I noticed
> > that `make distclean` fails to actually remove .i and .s files.
> >
> > This patch fixes it.  I hope it's the right place to make this change?
>
> It would be, but I don't think we want make clean to remove all
> possible output files from all possible tools. The current list is
> probably too much already, it should really be limited to files that
> are generated as part of the normal build process.

That's why the right way to keep the source tree clean is to use a
separate build dir. That way a "make reallyclean" is a simple "rm -r
$wine_build_dir".

bye
    michael