Re: [Axiom-developer] Re: typo in src/boot/Makefile.pamphlet

[Ralf Hemmecke]

Hi Gaby,

> I agree. Well, almost.  Almost because I do not believe *every* patch
> should be applied to the branch.

You probably have more years of experience than me. ;-)

> I believe patches that survive public review should be
> immediately applied.

Fine.

> It is good to have unstable branch where experiments are
> conducted.

Yes, I think that is good. You will probably suggest that for those 
experimental branches there might be other people responsible not you.

GNU ARCH lets you do this easily.

> but I think we need one branch that is unstable but not too unstable.

That's a good point. It sounds like a three stage system.

A golden branch that will be the stable version. (axiom--main--1)
A silver branch that will be the almost stable branch (that is where you
    opted for maintaining)
Several experimental branches, where one person (or a group of people)
    agrees on how they do development until the experiment becomes
    resonably stable to be ready for the silver branch.

I would like that. No matter if you take my (perhaps naive) suggestion 
or anything else, but we should make clear an development model for 
Axiom that is easier to learn and includes *public* review before Tim 
gets the final decision.

> | In fact, we could have (and actually we have) several unstable
> | branches already. There are already different people responsible for
> | them and a policy of how to update things there is completely
> | open. But you are right. It would be nice to have the "hottest" branch
> | available.
>
> I'm willing to take responsability for that, but I need general
> agreement from the community (especially from Tim) for having that
> sort branch with the intent that patches applied there will move to
> the stable branch when they have survived enough testing and satisfy
> The Master (Tim!) taste.

You get my full support. And to the others: That is a generous offer we 
should not reject that and let long years of experience with gcc enter 
into the Axiom world. We need even more people like Gaby.

> | I think that flexibility of tla is a bit too much. There are simple
> | guidelines missing of how to do things efficiently without the need to
> | read several hundred pages. (I am trying to learn already since I
> | would like to contribute code.)

> Can we move to SVN? (yes, it is a taboo subject but...)

Although I by now somehow think I can live with tla, I believe that we 
should not prevent you from doing things with SVN. If you are willing to 
maintain the silver branch using SVN, I see no point why we shouldn't 
accept that.

Of course, SVN has some weaknesses (especially not being able to do a 
star-merge), but since you could live with CVS for many years, I am sure 
you can cope with that missing feature anyway.

And there is a strong point for SVN. People currently seem to believe 
that it is the successor of CVS and they are quite eager to learn SVN.
So I expect the barrier to Axiom would be lowered. And that is certainly 
more important than sticking to a very flexible tla.

Well, I am actually not saying to remove tla, but if Gaby wants to 
maintain the silver branch using svn, then fine. Anyone who still wants 
to use tla, can do it, too.

> | >    * public informative review that helps gain understanding of why
> | >      things are the way they are, possible improvements, and things
> | >      that should not be tried.  That is very essential to attain
> | >      critical mass of people understanding the system.
> | 
> | Since I have no idea about how this works, could you be a bit more specific?
>
> Well, in general a victim sends a patch for review.  Component
> maintainers and/or maintainers with global write privilege (Tim, for
> instance) review the patch publically.  If the patch is fine, they say
> "OK" (that is rare when the victim is not very familiar with the
> system).  Otherwise, they technically comment on what are wrong
> about the patch and suggest directions for improvements. If the patch
> is outrageous, let say the patch suggests to sink Axiom, then an
> answer might be "don't propose that again." :-)

But for public review, I must see the patch applied somewhere. I am not 
a machine that can read native "diff" output. Sorry, but I think you 
need to teach a bit. And in order not to hide this course inside the 
mail archive, it would be just great, if you could open up a page on 
AxiomWiki and let us know. (Thanks in advance.)

Ralf