Re: [Gnu-arch-users] Features command for arch

[James Blackwell]

>
>     > From: address@hidden (James Blackwell)
>
>     > ddaa and I put our heads together not long ago, and we came up with a 
> new
>     > feature that would probably help programs like pyarch and aba. This new
>     > command, called "tla features" would list features and feature version
>     > numbers. One could also query either for a specific feature, or a 
> specific
>     > version of a feature. Please see the example output at the end of this
>     > email.

Wow. Ratta-tatta-tatta-tat

I'm not familiar with you making this style of reasoning. Does this mean
you like the idea? That you hate it? That we need to think really
carefully before we tread in this direction?

I brought this one up because David (ddaa of pyarch) is having difficulty
predicting tla behavior.  Sometimes, tla acts differently before a bug
is fixed than after. Also, he'd like to know trivially things like
whether or not tla uses pika escaping, etc. 

Eventually, either the front ends will have to carry a table of
behaviorsr, or we will. I think the right place for it to happen is us.
But if you disagree, lets discuss it.

Tom Lord wrote:
> How would user's benefit from such a feature? 

Users will benefit because the work on arch front/mid-ends will be
easier


>                                                How would it make arch
> better?  

Directly, not at all. Indirectly, yes, by easing the work the frontend
guys are doing it

>          Why is it important to work on now?  

There's more important things to work on, but it is low hanging fruit.


>                                              What is the definition
> of "a feature"? 

Any sort of behavioral change to tla that needs to be account for in
front ends.

>               How do implementors know whether or not they should
> say that such-and-such a feature is present?  

If there's any doubt, throw it in.

I suspect that I don't understand your question.

>                                                     Who shall be the
> arbitrator of "feature names"? 

I don't think there will be much need for an arbitrator, but I suppose
that if a controversial name comes up, it can be discussed.

>                                  How does this relate to the idea of
> standardizing arch with formal standards docs?

The syntax I threw up for the sake of argument. If you like, we can have
an extended conversation about the proper syntax to use.


-- 
James Blackwell          Try something fun: For the next 24 hours, give
Smile more!              each person you meet a compliment!

GnuPG (ID 06357400) AAE4 8C76 58DA 5902 761D  247A 8A55 DA73 0635 7400