Add the xmin horizon to running query display

[blogh]

The xmin horizon was added in 9.4 in commit [1], the addition of the column was asked in #425.

[1] postgres/postgres@dd1a3bc

[dlax]

Looks good; just a couple of minor comments.

Also, I'm wondering if you put the new column close to left side intentionally? Is there a business logic reason for this?

[blogh]

The tests crash on this although the read / write parts where not changed in this patch (tests/test_views.txt).

    -6239   1        pgbench                   pgbench         postgres            local 0.1    1.0  7B       12B       0.000000                N N        idle in trans UPDATE pgbench_accounts SET abalance = abalance + 141 WHERE aid = 1932841;
    +6239   1        pgbench                   pgbench         postgres            local 0.1    1.0  7B       12.3B     0.000000                N N        idle in trans UPDATE pgbench_accounts SET abalance = abalance + 141 WHERE aid = 1932841;
    -6228   2        pgbench                   pgbench         postgres            local 0.2    1.0  0B       1.08M     0.000413                  Y               active \_ UPDATE pgbench_accounts SET abalance = abalance + 141 WHERE aid = 1932841;
    +6228   2        pgbench                   pgbench         postgres            local 0.2    1.0  0.2B     1.08M     0.000413                  Y               active \_ UPDATE pgbench_accounts SET abalance = abalance + 141 WHERE aid = 1932841;

Notes:

• What to look for 12B -> 12.3B and 0 -> 0.2B
• The unit seems to be B/s instead of B and it's a delta from last mesurement

I also don't get this crash in my local env...
Do you have an idea of why the display of these floats would change ?

[dlax]

The tests crash on this although the read / write parts where not changed in this patch (tests/test_views.txt).

No idea why; I can take a closer look today or at least before the end of the week.

In the meantime, I have pushed your three other fixes (manpage, mypy, isort) to master (you might want to rebase the remaining commit here). The test failure reproduces there so it's indeed unrelated to this PR.

[dlax]

It's due to humanize 4.12.0, I can reproduce locally after pip install -U humanize.

[dlax]

Fixed in #432.

[blogh]

Also, I'm wondering if you put the new column close to left side intentionally? Is there a business logic reason for this?

No, it was easier while I was testing and it seemed as good a place as any.