Re: Automake 1.6.3 issue

[Bruce Korb]

Tom Tromey wrote:
> > Alexandre Duret-Lutz <address@hidden> writes:
> adl> but you can't do
> adl>   if INSTALL_SNPRINTFV
> adl>     pkgincludedir = something
> adl>   endif

> I've long thought that we should, eventually, support the latter use.
> It seems to have a clearly defined meaning.  And it is even useful in
> some situations.

The (my) situation turned out to be that my package incorporated
a "sub" package (snprintfv!).  In that context, snprintfv was not
separately installed.  Consequently, the handling of headers was
different.  For a separate installation, "pkgincludedir" was used
for its destination and it was specially set.  When not being
installed, "pkgincludedir" wasn't used at all, so it was not set.
It now gets set unconditionally, whether used or not.  All is well.

I did not know all this, because I don't poke my nose into
snprintfv packaging.  Anyway, I strenuously object to the way it
was handled.  viz., a hard error with an incomprehensible message.
It cost me hours of grief.

Moral of the story:

1.  Avoid hard errors unless absolutely unavoidable.
    Residual files in "makedistcheck" should not be a hard error.

2.  Make warnings when something may be amiss.
    That warning *must* be based on the inputs provided,
    not internal representations like "INSTALL_SNPRINTFV_TRUE"
    or "based on TRUE" or some such.  Refer the user to more info.

3.  Provide a way to say, "this is okay despite what
    you may think"  Generally, I know better than a tool.