Re: [PATCH] Allow generating a non-default set of documentation

From: Junio C Hamano <gitster@pobox.com>
Date: Oct 8, 2012, 16:33 UTC
Message-ID: <7vpq4sncte.fsf@alter.siamese.dyndns.org>
In-Reply-To: <7vhaq5onig.fsf@alter.siamese.dyndns.org>

Junio C Hamano <gitster@pobox.com> writes:

Show 30 quoted lines

> Jeff King <peff@peff.net> writes:
>
>> On Sun, Oct 07, 2012 at 04:25:58PM -0700, Junio C Hamano wrote:
>>
>>> Yeah, modulo that the "defaults" is tracked and does not have to
>>> have the dash before "include" (it is an error if it is missing,
>>> no?).  It may want to be named with s/defaults/autodetect/, though.
>>> 
>>> > diff --git a/Makefile b/Makefile
>>> > index e3e3cd5..c00fd32 100644
>>> > --- a/Makefile
>>> > +++ b/Makefile
>>> > ...
>>> > -endif
>>> > -
>>> > +-include config.mak.defaults
>>> >  -include config.mak.autogen
>>> >  -include config.mak
>>
>> Yeah, sorry, mindless copying on my part from the lines below. It
>> clearly should not have the "-".
>>
>> I'm fine with something besides "defaults". I meant it to be "these are
>> the config defaults before you tweak them", but yeah, it is more like
>> "these are the config options we picked up from uname".
>
> Let's put this on hold in the "possibly a good direction to go in"
> pile, and defer it to post 1.8.0; I haven't even looked at (and do
> not plan to before the release) these "auto-detect" bits are all
> safe to be included in Makefiles in subdirectories.

Oh by the way, by "this (on hold)" I meant not just your "move the default-by-platform to a separate include", but the whole "a platform may not want manpages" topic. The patch posted was merely me wanting to format git-push.txt and nothing else when sanity checking a patch to that file, and it is insufficient for the purpose of "a platform may not want manpages" without a matching change to the install target.

Even though your "include" thing was primarily done to make it easier to include it from Makefiles in subdirectories, I very much like it even without that benefit. The change makes it clear that there is another platform specific bit after '-include config.mak' (yes, I am looking at you, Darwin), which is a huge eyesore we would want to do something about.