[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Tue, 28 Aug 2001 21:55:41 +0200
Aaahm, it was just a simple example - I do sometimes like
to play around in automake and add another rule-set function
which in turn I like to turn on/off somehow (easily). The
option-hash is a good place - I did use generic option-settings
successfully in many projects that were able to load secondary
modules. Yes, I know that the automake script is currently just
monolithic and fears off users not familiar with the whole of
those 230k of perl code.
One can however start adding experimental rules in that
generate_makefile - and sometimes I like to just disable
feature-handlers. Quite some of these can be achieved with
a generic option - if it just would not hurt older makesets.
The actual kind of implemenation is uninteresting - it might
be a good idea to have automake just warn (instead of error)
about a specified set of options - no need to accept all
kinds, possibly just /^--with-.*/ ;-)
Among the experimental things that I like to patch in once in
a while, is that automatic LICENSE detection for distributin.
No, I don't mean the current scheme - instead glob(COPYING.*)
and check the first line to contain LICENSE ... ;-) .. makes
some things easier... and another thing are changes to some
regex-patterns and simple added handling that ensures the
Makefile.am is a valid gmake-input. The current am_conditional
is one such place that would have had no need to be incompatible.
anyway, I am not familiar with all the side-effects that could
come around in those 230k of code, so allow me to hesitate to
bring around those pieces.
have fun, Guido /stillnotsubscribed/
> >>>>> "Guido" == Guido Draheim <address@hidden> writes:
> Guido> proposal:
> Guido> accept colon settings of subarea-specific options. enables a
> Guido> scheme for modified and extended automake
> Guido> (alpha-)functionality to receive options that does not baffle
> Guido> released automake versions.
> I don't understand this. Options generally change the semantics of
> something. So having a way to make it so that an option isn't
> recognized doesn't seem useful -- if you use the wrong automake you'll
> end up with the wrong semantics for your Makefile.
> Guido> COPYING:COPYING.BSD and alpha-functionality accepting
> Guido> license files for automatic common_dist when *this* file
> Guido> with its alternative name is found. Guido>
> Why not just use EXTRA_DIST?