Re: delete command weirdness

kodifik@[EMAIL PROTECTED]
 writes:

> the goal of delete is to return the result '(b c d)
> _as_quick_as_possible_.

This is not so.  The goal is to permit implementations that want to
worry about speed issues to do so. There are two material differences
between my statement and yours.

First, an implementation may desire to worry about speed issues in ways
other than returning that particular answer faster.  For example, it
may be faster to copy but the GC in some implementation may run long.
In that case, DELETE may optimize overall performance but not copying
performance.  Implementations can decide what they want to optimize.
The spec takes no position on that.

Second, some implementations may not be concerned with speed.  It has
happened that there were implementations that wanted a small binary
footprint and that made DELETE and REMOVE be synonyms.  The same for
several other destructive operations.  While this behavior may
pessimize performance, it is presumed that the implementation knows
what market it is shooting for and what is valued by that market.
Consequently, the presence of DELETE in that cir***stance preserves
the right of SOME implementations to care about speed but does not
force ALL implementations to care.

In general, I think it's not a good idea for you to keep telling
people in some authoritative tone that you know the reasons that
something is there when you have not asked around to find out all the
possible reasons.  For one thing, the language was designed by a
committee and there's really no one who is empowered to speak for the
committee; the committee did not publish a rationale, and in fact its
reasons are not required to even be the same for each person who voted
the same way on each issue.  Everyone voted for their own reason.

It's fine, of course, to say "this is A reason" but before you start
to talk about "this is the ONLY reason", especially when there are
others who voted on those matters alleging otherwise and who you are
contradicting, you need to do more research.  This may be the only
thing YOU care about (I can't say--not being an expert on your tastes
and wishes), but the language design had a number of goals in mind.

Just for example: If forcing the optimizing of speed was the key
issue, there would not be an OPTIMIZE declaration that allows you to
trade speed for other factors.