Closed vilhelmgray closed 5 years ago
I appreciate the verbosity, but I feel like this is too complicated for an obscure corner-case (see comment on the PR #65). I'm ok with --enable-man
basically being a no-op.
One of the important reasons this is useful is that it follows the canonical convention of Autotools configure scripts -- that a build should not fail after a successful configuration.
On my distribution, it is common for users to batch install several packages via scripts that test the results of configure scripts to verify support for requested features and options; when a configure script reports an error, it's obvious to the user that a configuration setting is incorrect and should be changed.
If I package Deutex 5.2.1 as it is now, I'll end up with several users reporting build errors for Deutex under the belief that the Deutex Makefile is broken when it's that their system does not support manpage generation.
Fixing the behavior of --enable-man
should resolve confusion for users of my distribution before I release the deutex 5.2.1 package, so that I avoid these sort of error reports.
Merged the PR, but I'll hold on making a new point release for now. Feel free to include the patch with your ebuild if you wish.
If a user explicitly requests manpage generation, the configure script should report an error if the system is unable to perform the request -- otherwise the build will fail unexpectedly on machines that do not have a2x.
There are 3 possible scenarios:
./configure
Here the configure script should check for a2x and build the manpage if able. If not able, the configure script will automatically disable the manpage build without error../configure --disable-man
The user explicitly requests to bypass the manpage build so the configure script should disable it../configure --enable-man
The user explicitly requests to enable the manpage build so the configure script should report an error if the system is unable to perform the build.