Re: Crash report for some ICU-52 (debian8) COLLATE and work_mem values

On Mon, Aug 14, 2017 at 9:23 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Peter Geoghegan <pg(at)bowt(dot)ie> writes:
>> The comments that pg_import_system_collations() adds to each locale
>> have an annoying tendency to exclude anything that isn't ascii-safe,
>
> Right, but that's *necessary* for anything that goes into template0.

Of course. I phrased this badly.

>> Another thing that I dislike about get_icu_locale_comment() is that
>> the descriptions we show are always English display names, regardless
>> of pg_database collation or anything else. We shouldn't hard-code the
>> use of English for display name, on general principle.
>
> Again, that was mostly driven by the desire to get ASCII results
> (or at least so I imagine; Peter E. might have a different take).

Probably.

> I think there's value in having a function that produces such a
> description, but we could have a different one that produces a
> description in the language indicated by the database's locale.

I suppose it's useful to have it match the database locale/language,
too. The point IMV is that display name is strictly for, uh, display.
I'm not attached to any particular scheme for determining which
language to display in which context. That's for v11.

I take it that you're in agreement with me on
pg_import_system_collations() adding a display name as a comment -- we
should prevent that, even for v10?

--
Peter Geoghegan