--with-MODULE vs. --with-experimental-modules
AFAICT, --with-module doesn't do anything; you have to supply --with-experimental-modules to let non-stable modules build, and that pulls all of them in. Am I missing something? How might I, for example, compile the server with all stable modules plus rlm_redis?
On 14 Nov 2012, at 13:51, Phil Mayers <p.mayers@IMPERIAL.AC.UK> wrote:
AFAICT, --with-module doesn't do anything; you have to supply --with-experimental-modules to let non-stable modules build, and that pulls all of them in.
Am I missing something? How might I, for example, compile the server with all stable modules plus rlm_redis?
Nope. That's weird. It's not obvious from git blame how this was ever meant to work. There's no reason for it to rediscover the module names, it should just use the value of $MODULES and test each of the specified module directories for a configure script. There's a good reason for having both $MODULES and $mysubdirs but AFAICT $mysubdirs should always be a subset of $MODULES. I'm going to push a fix, and if Alan has any insight he's welcome to revert it/correct it. -Arran
On 15 Nov 2012, at 10:31, Brian Candler <B.Candler@pobox.com> wrote:
On Wed, Nov 14, 2012 at 01:51:01PM +0000, Phil Mayers wrote:
Am I missing something? How might I, for example, compile the server with all stable modules plus rlm_redis?
I just edit src/modules/stable and add
rlm_redis rlm_rediswho
Uh well now you can just use a working configure script :) I guess I should add another fix not to re-initialise the module list to an empty string so non-stable modules can be specified in config.site. -Arran
participants (3)
-
Arran Cudbard-Bell -
Brian Candler -
Phil Mayers