Re: [PATCH 00/22] Patches to avoid use of make recursion in the 'src/' subdir

[Jim Meyering]

Stefano Lattarini wrote:

> On 09/01/2012 10:20 AM, Stefano Lattarini wrote:
>> On 09/01/2012 10:08 AM, Jim Meyering wrote:
>>> Stefano Lattarini wrote:
>>>> This is a re-spin, extension and follow-up to this previous series:
>>>>
>>>>     <http://lists.gnu.org/archive/html/coreutils/2012-08/msg00257.html>
>>>>
>>>> plus other follow-up fixlets and refactoring.  Thanks to Jim Meyering
>>>> for his testing of earlier iterations of this series, and his relentless
>>>> reporting of problems and errors.
>>>
>>> Thanks for all of the work.
>>> However, even with all of that, "make distcheck" still fails, when
>>> trying to build prog.1 when the required src/prog does not yet exist.
>>> (it must be run with --help)
>>>
>>> Here's a first required patch.
>>>
>>> From cecef0d2f7ff7e4092cc991a7b8d2b0bceadf0ee Mon Sep 17 00:00:00 2001
>>> From: Jim Meyering <address@hidden>
>>> Date: Sat, 1 Sep 2012 10:00:26 +0200
>>> Subject: [PATCH] build: make each man/prog.1 depend on src/prog, not
>>>  src/prog.c
>>>
>>> * man/local.mk: Creating a prog.1 man page requires running
>>> src/prog --help.
>>> List the exceptions, e.g., install.1 depends on src/ginstall
>>> and arch.1 depends on src/uname.
>>>
>> But this will force the users to have perl installed, since help2man
>> will now have to be run unconditionally after the compilation step.
>>
> Oops, sorry, I take that back.  I see that the '.x.1' suffix rule will
> be a no-op when perl is missing, so the users lacking perl will just
> see a bunch of extra warning messages, but will have no real issue with
> "make install" since the manpages will still distributed.

Right.

> This is not so obvious at a first glance, and deserves an explanatory
> comment in the code IMHO.

Patch welcome.

> The catch is that, after you patch, all the users with perl will have
> two copies of each manapge: one in $(srcdir), distributed, and one in
> $(builddir), rebuilt by "make all".  Which is a sort of confusion for
> humans, and might maybe confuse some make implementation too ...

I like your changes, and for now am mainly concerned with ensuring
that "make distcheck" continues to work with them.
If you find an actual problem or a way to eliminate the
potential confusion, patches are always welcome.