Re: PostgreSQL 13 Beta 1 Release Announcement Draft

On 5/20/20 7:30 PM, Alvaro Herrera wrote:
> Hello, sorry for the timing on this, but here's some feedback.
>
> On 2020-May-20, Jonathan S. Katz wrote:
>
>> PostgreSQL 13 introduces many features and improvements to the world's most
>> advanced open source database. The following is just a small sample of new
>> features introduced into this latest version of PostgreSQL.
>
> "A small sample" suggests to me that there are other equally important
> features that you did not list, in a somewhat random way. I doubt
> that's true; as a user, I would expect you to list the most important
> changes. I would use the phrase "The following are the most visible
> improvements" or some such.

After trying to reword it, I ended up striking the whole section. A
bunch of it was redundant to the above section. The title of the section
can have it stand on its own: "PostgreSQL 13 Highlights"

>> PostgreSQL 13 adds incremental sorting, which accelerates sorting data when
>> existing data in a query is already sorted.
>
> I'm not sure this is very accurate. I would say "incremental sorting,
> which can take advantage of partially sorted data". (Could add more to
> it, "when only a subset of the result set is needed", but that is
> perhaps too many words).

I made modifications based on your downthread comments.

>> There are more improvements added to PostgreSQL's partitioning functionality in
>> this release, including an increased number of cases where a
>> "partitionwise join" (a join between matching partitions) can be used,
>
> I'd say "A join between partitioned tables". (The new feature precisely
> is that the partition strategies don't have to match exactly anymore).

Cool, modified.

>> One of the most anticipated features of PostgreSQL 13 is the ability for the
>> `VACUUM` command to process indexes in parallel. This functionality can be
>> accessed using the new `PARALLEL` option on the `VACUUM` command (or
>> `--parallel` on `vacuumdb`), which allows you to specify the number of parallel
>> workers to use for vacuuming indexes. Note that this does not work with the
>> `FULL` option.
>
> I feel this "Note" is unnecessary; VACUUM FULL is of fringe use anyway.
> The phrase on incremental sorting above is better use of these extra
> words, if you have a surplus.

While I'm not running in production as much as I used to, there will
still times I would need to use a VACUUM FULL ;)

>
>> This release includes more ways to monitor actibity within a PostgreSQL
>> database: PostgreSQL 13 can now track WAL usage statistics and the progress of
>> streaming `pg_basebackup`s, and the progress of an `ANALYZE` command.
>> `pg_basebackup` can also generate a manifest that can be used to verify the
>> integrity of a backup using a new tool called `pg_verifybackup`.
>
> In this monitoring paragraph I would add that it's possible to limit WAL
> reserved by replication slots. "It is now possible to limit the amount
> of WAL space reserved by replication slots."

That is a good one - I added that mostly word-for-word.

Thanks!

Jonathan