Re: Code for cond*

[Stefan Monnier]

>    > FWIW, I'm opposed to adding it without first unifying the Pcase and
>    > `match*` pattern languages (including making sure that third party
>    > packages which define their own patterns don't need to define them
>    > separately for `pcase` and for `match*`).
>
>    > [ It's trivial to do by making `match*` reuse the Pcase machinery, so
>    >   there is no serious obstacle for that.  ]
>
>    I hope that cond* won't be written on top of pcase.

[ Side note: my comment above specifically refers to `match*`,
  not to `cond*`.  Let's not confuse the two: `cond*` is an actual
  addition, whereas `match*` mostly reinvents Pcase patterns, with a few
  tweaks to some of its syntax.  ]

I don't have a strong opinion on that, actually.

I hope that we don't end up having to maintain two implementations of
the pattern matching code, but there can be good arguments for such
a duplication (there are several different implementation strategies,
each with their own advantages), so if that's what it takes, I'm OK
with it.

> and maybe even that that pcase could be rewritten using cond*.

As already mentioned in this long thread, the current `match*`
implementation does not generate good enough code for existing `pcase`
uses (and fixing that probably requires non-trivial changes).


        Stefan