>>>>> "Akim" == akim <[EMAIL PROTECTED]> writes:
Akim> I will look at this tomorrow and fix it. Without looking at the
Akim> code, I think I must have been somewhat too systematic Iin
Akim> inserting ?EXEC?, I don't think info page should depend on exec
Akim> vs. data. Tom? I'm not sure of myself, there are some rules
Akim> Tom is trying to teach me, but something escapes from my
Akim> understanding.
Here is how directory installation works:
* There are a bunch of directories defined by the GNU Standards.
We separate these into two categories: system-specific, which are
installed by `install-exec', and system-independent, which are
installed by `install-data'.
* Users can define their own directories using variables whose names
end in `dir'. User variables which contain `exec' as a substring
are installed by install-exec; others by install-data.
The above stuff should alreay be documented. If not a patch would be
great, but a bug report would be ok too.
For a given primary we try to detect which directories are valid.
* A user-defined directory is valid for any primary
* The valid standard directories are chosen on a per-primary basis.
Look for calls to &am_install_var to see what is valid for a given
primary.
Eg. for headers:
my @r = &am_install_var ('-defaultdist', 'header', 'HEADERS', 'include',
'oldinclude', 'pkginclude',
'noinst', 'check');
This means `includedir', `oldincludedir', and `pkgincludedir' are
valid. Also `noinst' and `check' (which aren't directories) are
valid.
Akim> I would really love to use %FOO% instead in Automake. Tom, what
Akim> do you think? This would make it possible for automake to check
Akim> what it outputs.
That would be fine.
Tom
