On 2008-09-27 at 21:48 +0100, Peter Stephenson wrote:
> I appreciate there's a problem here, but I'm a little bit lost as to
> what exactly you're after.  Do you mean you'd like the "|" in the (j,|,)
> to be treated as a pattern character, but the characters substituted
> from the paramater named by b not to be expanded?

What I was after wasn't anywhere near as good as what you provided.

I was thinking that just as we have (q) to apply shell quoting, we
should have a flag to escape pattern characters, which would have
laboriously allowed me to do what you've just done much more easily by
avoiding the whole problem.  :)

Escape+join+parse_that vs only-parse-the-join

> If I'm not following, you may want white chococlate mousse followed by a
> cappuccino.

No, but my wife would love that.

> Anyway, I'm not sure why I didn't do this ages ago, it makes testing
> lists of alternatives vastly easier.

Oh good, that's two different uses for it which means it's definitely
more broadly helpful and it's not just me.  *phew*  Thanks.

Okay, the two attachments provide the resulting clean set operations,
with a stupid bug fixed and the intersection operations cleaned up, and
the Test file for these.

I'm not attached to any names, but I put the first file in
Functions/Misc/load_dataset and the test file depends upon that.  I just
named load_dataset "load_dataset-post-20080927.zsh" to remind anyone
using this that it uses your new (~) parameter expansion option.

I'm also not attached to this code, since messing with 'let' to have zsh
understand sets in arithmetic context is more appealing to me, every
time I think about it.  But the two approaches should be mutually
compatible and this is the one I opted for.

And I *just* realised that I needed set_equal, which sets $?, to provide
a clean way to test for equality, since not only are the sets not
maintained in sorted order, but the need to deal with arbitrary data
would lead to bugginess.  *sigh*  Arithmetic with no equality tests.
What was I thinking?  Done.

Do people think superset/subset/strict_superset/strict_subset would be
useful too?

-Phil