Re: *_DATA containing blanks

[Alexandre Duret-Lutz]

[...]

 Ralf> Just a thought: Don't automake parse the contents of make-variables? 
 Ralf> Why doesn't it expand these variables and use expanded loops instead of
 Ralf> trying to use these vars directly? This would also give automake the
 Ralf> opportunity to quote sick vars (like those using $ or ' ') correctly?
 >> 
 >> Not for AC_SUBST variables :(
 Ralf> ??? 

 Ralf> I had been referring to automake variables (EXTRA_DIST and *_DATA in
 Ralf> Makefile.ams).

 Ralf> Yes, they can contain autoconf substitions (@foo@), refer to
 Ralf> make-variables ($(foo)) and can be implicitly generated by automake (eg.
 Ralf> DISTFILES). But I don't see why automake can't parse them and expand all
 Ralf> fixed values (*_DATA = foo).

AFAICT because Automake can't expand @foo@, so it can't quote it properly.

Consider

  if ...; then
     TO_INSTALL='this\ one';
  else
     TO_INSTALL='that$$one';
  fi
  AC_SUBST([TO_INSTALL])

and

  EXTRA_DIST = this\ one that$$one
  data_DATA = $(TO_INSTALL)

What should the install loop for data_DATA look like?

[...]

 Ralf> distdir-hook doesn't seem to exist, dist-hook seems to be
 Ralf> applicable to work-around my problem.

Yes, sorry.

 Ralf> However, this forces me to rely on automake internals ($distdir)

$(distdir) ought to be documented if it's not already: it's hard
to write a dist-hook without it.
-- 
Alexandre Duret-Lutz