mirror of
https://git.savannah.gnu.org/git/gnulib.git
synced 2025-08-14 14:01:48 +03:00
32 lines
1.4 KiB
Plaintext
32 lines
1.4 KiB
Plaintext
@node configmake
|
|
@section configmake
|
|
|
|
@findex configmake @r{module}
|
|
@cindex @file{configmake.h}, module for updating
|
|
|
|
The @code{configmake} module builds a C include file named
|
|
@file{configmake.h} containing the usual installation directory
|
|
values; for example, those specified by @code{--prefix} or
|
|
@code{--libdir} to configure. Each variable is given a @code{#define}
|
|
with an all-uppercase macro name, such as @code{PREFIX} and
|
|
@code{LIBDIR}. (Automake cannot create this file directly because the
|
|
user might override directory values at @code{make} time.)
|
|
|
|
Specifically, the module retrieves values of the variables through
|
|
@code{configure} followed by @code{make}, not directly through
|
|
@code{configure}, so that a user who sets some of these variables
|
|
consistently on the @code{make} command line gets correct results.
|
|
|
|
One advantage of this approach, compared to the classical approach of
|
|
adding @code{-DLIBDIR=\"$(libdir)\"} etc.@: to @code{AM_CPPFLAGS}, is
|
|
that it protects against the use of undefined variables. That is, if,
|
|
say, @code{$(libdir)} is not set in the Makefile, @code{LIBDIR} is not
|
|
defined by this module, and code using @code{LIBDIR} gives a
|
|
compilation error.
|
|
|
|
Another advantage is that @code{make} output is shorter.
|
|
|
|
For the complete list of variables which are @code{#define}d this way,
|
|
see the file @file{gnulib/modules/configmake}, or inspect your
|
|
resulting gnulib Makefile.
|