Re: m4 macro whose name depends on a a variable

[Vincent Torri]

On Mon, 10 Nov 2008, Eric Blake wrote:

> For that matter, even
>
> m4_default([EVAS_CHECK_LOADER_DEP_]m4_defn([UP]))(args)
>
> would work.

indeed

> > AC_CHECK_HEADER([jpeglib.h],
> >    [
> >     have_dep="yes"
> >     evas_image_loader_[]$1[]_cflags=""
> >     evas_image_loader_[]$1[]_libs="-ljpeg"
> >    ]
> > )
> >
> > AC_SUBST([evas_image_loader_$1_cflags])
> > AC_SUBST([evas_image_loader_$1_libs])
>
> This last usage is different than the first two (in the first two, you
> separate $1 from its neighbors, so that a $1 of either "LOWER" or "jpeg"
> would work.  But since you didn't separate it in the last case, you are
> stuck with:
>
> evas_image_loader_LOWER_cflags
>
> which is not a macro name, and you've lost jpeg.  You probably meant:
>
> AC_SUBST([evas_image_loader_[]$1[]_cflags])

no, I use AC_SUBST([evas_image_loader_$1_cflags])

I call that macro that way:

m4_default([EVAS_CHECK_LOADER_DEP_]m4_defn([UP]))(DOWN,
   [have_evas_image_loader_[]DOWN="yes"],
   [have_evas_image_loader_[]DOWN="no"])

I have not quoted DOWN (the 1st argument). Otherwise it does not work.

> > if test "x${have_dep}" = "xyes" ; then
> >   ifelse([$2], , :, [$2])
>
> Yes, ifelse is another raw m4 macro you should avoid.  In general, this
> can be replaced with m4_if, but in this particular use case, you can write
> the abbreviated:
>
> m4_default([$2], [:])

ok, thank you for your help. It's nice to see that we can write some kind 
of interface in m4 :)

Vincent Torri