Re: 2022-02-10 release announcement draft

On 2/6/22 10:20 PM, Justin Pryzby wrote:
> On Sun, Feb 06, 2022 at 08:01:02PM -0500, Jonathan S. Katz wrote:
>> Hi,
>>
>> Attached is a draft for the release announcement for the 2022-02-10
>> cumulative update release.
>>
>> Please review for technical accuracy or if you believe any items should be
>> added/removed. Please provide feedback no later than 2020-02-10 0:00 AoE[1].
>
> I guess you mean 2022 ;)

Maybe ;)

>> * The [`psql`](https://www.postgresql.org/docs/current/app-psql.html)
>> `\password` command now defaults to setting the password for to the role defined
>> by `CURRENT_USER`. Additionally, the role name is now included in the password
>> prompt.
>
> s/for to/for/

Done.

>> * Build extended statistics for partitioned tables. If you previously added
>> extended statistics to a partitioned table, you can fix this by running
>> [`ANALYZE`](https://www.postgresql.org/docs/current/sql-analyze.html).
>> [`autovacuum`](https://www.postgresql.org/docs/current/routine-vacuuming.html#AUTOVACUUM)
>> currently does not process these statistics, so you must periodically run
>> `ANALYZE` on any partitioned tables that are using extended statistics.
>
> Instead of "you can fix this" , I suggest to say "you should run ANALYZE on
> those tables to collect updated statistics."

I opted for most of this suggestion.

> It should say "not process these TABLES" (not stats), and should not say "that
> are using extended statistics" since partitioned tables should be manually
> analyzed whether or not they have extended stats.

>
> It's good to say "..you should periodically run ANALYZE on partitioned tables
> to collect updated statistics." (even though this patch doesn't change that).

I reread the release note an agree, I misinterpreted this. I updated the
language.

>
>> * Fix crash with [multiranges](https://www.postgresql.org/docs/current/rangetypes.html)
>> when extracting a variable-length data types.
>
> remove "a" or s/types/types/

Fixed.

>> * Disallow altering data type of a partitioned table's columns when the
>> partitioned table's row type is used as a composite type elsewhere.
>
> the data types ?

This is correct, per 7ead9925f

I've attached the updated copy. Thanks!

Jonathan