Re: help building boost with V2

From: David Abrahams (gclbb-jamboost_at_[hidden])
Date: 2003-09-29 10:39:55

Vladimir Prus <ghost_at_[hidden]> writes:

> David Abrahams wrote:
>
>> > Some time ago I wrote the following on the subject of
>> > property-adjuster
>> > (in http://article.gmane.org/gmane.comp.lib.boost.build/3589/)
>> >
>> > Please feel free to remove this feature for now. It would be best if
>> > the change is made with a single commit that touches nothing else, so
>> > that we can use the diff if/when this behaviour should be implemented
>> > again.
>> >
>> > So, the ball is on your side.
>>
>> Of course it is, but this is "just a drop in the bucket", as they say.
>> I should probably do that, but right now I'm not using v2 for anything
>> and I think it would make only a small difference in the overall
>> picture.
>
> I'm sorry, I don't understand your attitude. You say you're
> interested in contributing to V2. You're +1 on removing
> property-adjuster, I'm -/+0. And you don't seem to care about
> removing it yourself, even though it's small and easy change.

I just did it.

The thing is, when I was working hard to understand the architecture
so I could make a contribution, removing property-adjuster would've
made a difference to my life because it would've simplified a very
complicated part of the system. Now, I've forgotten so much that the
only difference it made is that I'll have a slightly easier time
re-learning it.

> Maybe, you could clarify what kind of contributions you're
> interested in, and how we can make architecture good enough to allow
> those contributions

Ali and I made some suggestions for reworking things several months
ago, but I guess you were unconvinced that they would actually result
in improvements. I suggested what I considered to be an architectural
improvement to the generators system and even coded a prototype of the
search mechanism in Python. I'm not saying that you were unreceptive;
I admit that it would be a major undertaking to retrofit something
like that into the system. The fact remains that generators have
complicated and somewhat special-cased logic, and it's a difficult
project (or it was last time I tried) to build a new one.

I added a rule parameter type declaration system a while back because
Ali and I decided it would improve readability and error-checking.
Has it seen any use?

> except by making a number of comitts, each of them makes a small
> difference. As you put it, it sounds like you won't contribute
> anything unless it makes a big difference

No, it's not that; I was just explaining why I hadn't bothered to
make the change yet. I have other things going on and as you said,
you're +/-0 on it so I didn't think it mattered all that much.

> and a good architecture is definitely required to make a big
> change. The problem if that as long as architecture is really good,
> all which remains is writing various toolsets, and there's little
> place for big changes.

I'm not so sure. I think we need a python.jam (is that a toolset?)
and testing support at the very least. I think the ability to extend
generators and file types easily is important, too.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com