Re: source code reorg & makefiles

From: Alex Rousskov <rousskov_at_measurement-factory.com>
Date: Wed, 27 May 2009 22:40:20 -0600

On 05/27/2009 05:45 PM, Robert Collins wrote:
> On Wed, 2009-05-27 at 16:31 -0600, Alex Rousskov wrote:
>> Do all modern environments have good support for large non-recursive
>> projects? Can somebody provide an example of a large project using
>> non-recursive Makefiles? Or is that just a nice idea that never really
>> took flight and is not being optimized/designed for?
>
> Without going out and doing an audit of other projects, I can't state
> specifics about where nonrecursive make is/isn't used.
>
> The autotools toolchain, which we are using, has tolerably good support
> for non-recursive make, and it is improving over time.
>
> I suspect that most large projects don't use non-recursive make, simply
> because they have generally been created some time ago before toolchain
> support existed at all.
>
> I do know that CMake, SCons, WAFfle, bake, cook, and other such
> toolchains all use a global-dependency-graph approach (which is the same
> as non-recursive make).

This does not sound very encouraging. It is obviously not your fault
that there are no *well-known* examples of large projects using
non-recursive makes. However, your remarks seem to imply that as long as
we are using autotools, migrating to non-recursive makes will force us
to use less popular "parts" of the toolchain (to put it mildly). Given
Squid size, we may end up on the cutting edge of autotools use in that
context.

This sounds like a recipe for serious autotools pains. Do you think I am
being too pessimistic? If not, are those autotools pains worth the
advantages offered by non-recursive Makefiles?

Thank you,

Alex.
Received on Thu May 28 2009 - 04:40:39 MDT

This archive was generated by hypermail 2.2.0 : Sat May 30 2009 - 12:00:04 MDT