Re: [PATCH] proposal for regexp_count, regexp_instr, regexp_substr and regexp_replace

On 03/21/21 09:19, Gilles Darold wrote:
>>> On 2021.03.20. 19:48 Gilles Darold <gilles(at)darold(dot)net> wrote:
>>>
>>> This is a new version of the patch that now implements all the XQUERY
>>> regexp functions as described in the standard, minus the differences of
>>> PostgerSQL regular expression explain in [1].
>>>
>>> The standard SQL describe functions like_regex(), occurrences_regex(),
>>> position_regex(), substring_regex() and translate_regex() which
>>> correspond to the commonly named functions regexp_like(),
>>> regexp_count(), regexp_instr(), regexp_substr() and regexp_replace() as
>>> reported by Chapman Flack in [2]. All these function are implemented in
>>> [v2-0001-xquery-regexp-functions.patch]

I quickly looked over this patch preparing to object if it actually
purported to implement the ISO foo_regex() named functions without
the ISO semantics, but a quick grep reassured me that it doesn't
implement any of those functions. It only supplies functions in
the alternative, apparently common de facto naming scheme regexp_foo().

To be clear, I think that's the right call. I do not think it would be
a good idea to supply functions that have the ISO names but not the
specified regex dialect.

A set of functions analogous to the ISO ones but differently named and
with a different regex dialect seems fine to me, especially if these
different names are de facto common, and as far as I can tell, that is
what this patch provides. So I have no objection to that. :)

It might then be fair to say that the /description/ of the patch as
implementing the XQuery-based foo_regex functions isn't quite right,
or at least carries a risk of jarring some readers into hasty
double-takes on Sunday mornings before coffee.

It might be clearer to just mention the close correspondence between
the functions in this differently-named set and the corresponding ISO ones.

If this turns out to be a case of "attached the wrong patch, here's
the one that does implement foo_regex functions!" then I reserve an
objection to that. :)

Regards,
-Chap