Subject: Re: [ANNOUNCEMENT] The pkgsrc-2005Q4 branch
To: None <tech-pkg@pkgsrc.org>
From: George Georgalis <george@galis.org>
List: tech-pkg
Date: 12/30/2005 14:24:36
On Fri, Dec 30, 2005 at 05:08:33PM +0100, joerg@britannica.bec.de wrote:
>On Fri, Dec 30, 2005 at 10:31:41AM -0500, George Georgalis wrote:
>> 
>> Recursive Make Considered Harmful
>

I didn't say that, I said that paper was interesting.

>Please, stop quoting this paper. It is in large parts bogus, because it
>starts from the average stupid GNU makefile, completely ignoring the
>descriptive power either GNU or BSD make have. It also doesn't really
>apply to the problem at hand at all, since being recurisve is a good
>thing here. This is independent from the lack of a *self-recursive* make
>implementation.

of course it would be stupid not to use recursion in the pkgsrc tree.

the paper does point out make problems that can occur, and it was
educational for me.  I only recall posting the link one other
time, about a year ago, to the dfly-kernel list; about which there
was a healthy discussion. the only "problem at hand" I see here is
the the package build time.

I appreciate your allusion to recursion vs self-recursion, which
is a subtle distinction I'd not considered, that makes the paper's
applicable focus more clear -- and my understanding of "recursion
but not self-recursion" less clear.

When I look at the abstract I focus on this:

        The resolution of these problems can be
        found by looking at what make does, from
        first principles, and then analyzing the
        effects of introducing recursive make to
        this activity.

While you seem to focus on the following part:

        This conclusion runs counter to much
        accumulated folk wisdom in building
        large projects on UNIX. Some of the main
        objections raised by this folk wisdom are
        examined and shown to be unfounded.

but per my reading, the pkgsrc tree does exactly what the paper
recommends...

I've never done a "cd /usr/pkgsrc && make bulk-package" but
because the GNU packages get "packaged" I presume it works okay,
so maybe it is not too applicable to the package build time
problem, nevertheless, I still find the paper interesting.

Ciao,
// George


-- 
George Georgalis, systems architect, administrator <IXOYE><
http://galis.org/ cell:646-331-2027 mailto:george@galis.org