Re: Emacs vista build failures

[Thomas Lord]

Richard M Stallman wrote:
>     It is a failure of the GNU project and of the free software movement
>     that there is so much emphasis on monolithic distributions and binary
>     package distributions.   It is a failure of the GNU project and the free
>     software movement that one so often encounters distros that offer to
>     not install source trees and even offer to not install development
>     environments.
>
> The words "it is a failure of" are ambiguous.  They could literally
> mean "it is a goal we have not achieved", but they also suggest
> placing the blame for this on us.
>   

You're a better speaker than me, I think. 

It's not blame.  It's an attempt to egg on in a direction for which I 
wish to
advocate.




> Certainly these are goals we have not achieved, but if others do not
> follow our recommendations, that's their decision, not ours.  When I
> designed the GNU specs for configuring and building source packages, I
> hoped that free software developers generally would adopt them, but
> they did not.
>   

A heck of a lot of them did.  That isn't the problem.   In my view the 
problem
is that the specs were short-sighted in some key ways -- too much pain 
and not
enough benefit.   I'd be happy to take discussion of this view off-list 
though I'm
reluctant to go into it here, given Stefan's recent feedback.



> I tried at one point to convince XFree86 to support the GNU
> configuration spec.  I even found a volunteer to implement that as a
> wrapper around their existing configuration mechanism.  But they did
> not consider such compatibility very important, and I don't think they
> installed this wrapper.
>   

The GNU standards buy a little but not a lot and there are better ways
to do it.  We should talk.



> To convince free software projects generally to adopt this spec
> would require more pressure from the community in general.
>   

No.  A lot of the reason people *do* more or less use GNU standards
is personal convenience -- autoconf / GNU make / etc. make it easier
than many alternatives.

Better standards and better (and simpler) tools could amplify that
effect and have nicer side effects.

To be fair, you folks were solving a different problem, in the moment,
way back then.   A complete GNU system was quite a ways off and, as
a good tactic, the most immediate need was producing packages that
were not hard to build and install on a diverse range of proprietary Unix
systems.   That problem got solved pretty well but at a strategic cost of
being under-prepared for the "scale up" problems of managing source for
a complete system (imo). 



> It is easy to call names (such as calling the GNU Coding Standards
> "anemic"), but given that many programs' developers won't even
> implement those, I doubt we would obtain much compliance for stricter
> ones.
>   

If you ever think I'm "calling names" except in the case of geopolitical
polemics about fascists and such then please first, instead, think: 
"This is
an area where I can help Tom improve his writing," since I almost certainly
did not mean to "call names."

I do mean "anemic" in the sense of "vulnerable" or "slightly feeble".
It took and takes a lot of code to support those standards and the pay-off
fails to address many of the needs of a complete, portable system.

Stricter standards, better thought out, with simpler tools that realize them
might (plausibly would) find stepped up adoption.

So, I guess now I have to echo Stefan to me and say, to be clear,
that I assure you I intended no insult.

-t



>