Re: why doesn't vpath %.c subdir// work?

[David Boyce]

On Wed, Feb 17, 2010 at 10:03 AM, Mark Galeck (CW) <address@hidden> wrote:
> Hmm, this is interesting, thank you for bringing this up David.  I have two 
> comments:
>
> 1.  This (and other) open source projects, are used in corporate 
> environments, where certain version of the project is installed, and users 
> cannot just download a new version without approval.  If I use a particular 
> fixed feature of GNU make, I cannot populate it across the company without 
> everybody else downloading the latest version, which in practice makes it 
> impossible to use a fixed feature, until the IT department downloads the 
> change for everybody, which can be very delayed.

I don't know how many times in my career I've heard people complain
about a bug and then, once the bug is fixed, complain about the pain
of the upgrade. I believe the logical implications are clear. In any
case they apply to every piece of software, open and closed, cheap and
expensive. C'est la vie.

But this is orthogonal to my point anyway. Hunting down the problem in
the source might expose a workaround, if you're lucky. At the least it
will answer your question. And if you explain what you found here
someone else will probably make the fix even if you don't supply a
patch (guessing the fix would be simple once isolated). But in the end
you're the one who cares. This is not a support line; people provide
answers if they're intrigued by the question, or if they happen to
know off the top of their heads, or if it sparks a shared interest in
improving GNU make. If those things don't happen, it's time to UTSL
whether you plan to contribute the result or not.

> 2.  Pardon the naivety of this question, but I have never edited any open 
> source project:  Are you saying I could just download the sources of GNU 
> make, make and test the fix, and upload it back?  What about the approval of 
> the maintainers of the project?  I see version 3.81 as current from 2006, but 
> there must have been a gazillions of the contributions since then, why are 
> they not merged to the master download then?

This was discussed on this forum a few months ago. Paul has been
hoping to get out a new release candidate incorporating the many
bugfixes and features (including some of mine, I might add) of the
last few years. He's also hoping to achieve a more frequent release
cycle. Unfortunately (for us), he has a day job and a family.

You can easily download the latest source (if you plan to contribute a
fix) or the 3.81 source (might be better if you just want to
diagnose). If a fix is made there are two routes: (1) certain people
have earned the right to be "committers" with sufficient privilege to
check changes in directly. The rest can send in a patch in "diff -u"
format. In either case, yes, a significant waiting period tends to
follow. But you may be relatively lucky because a 3.82 RC is
pending[*] and a minor fix would be likely to get into it.

-David Boyce

[*] Not that I have any inside information.