Re: [Gnu-arch-users] [BUG] feature plan -- version variables

[Aaron Bentley]

Tom Lord wrote:
> The only thing I've said with the restriction that string datums 
> have to be ISO-8859-1 is that all of the `t_uchar' values in a string
> datum are, in fact, Unicode code points.

Yeah, got it.

> Ah.  Ok, you must have been thinking of version variable values as
> strings-that-can-be-interpreted-otherwise.   No, this is actually
> values as (a subset of) scheme datums.

Yes.  I think I've heard that Scheme is derived from Lisp.  That's where 
my knowledge of it begins and ends.

> Suppose a variable is locally set to #f.
>
> Then there's no reason to record in the patch log at all when a commit
> happens.   By omitting the variable from the patch log, you
> effectively set its patch-log value to #f.

Actually, I think it's worthwhile to preserve the distinction between 
missing and false.  In my work, I frequently encounter variables that 
should be treated as true if missing, and I think the problem could only 
be worse with a metadata system.  It's a distinction embraced by 
hackerlab's hashtrees and the C++'s std::map too.

Because you have to be able to answer the question: did they 
deliberately disable that?  Or were they unaware of the possibility of 
enabling that?

Aaron
--
Aaron Bentley
Director of Technology
Panometrics, Inc.