Re: PATCH: logical_work_mem and logical streaming of largein-progress transactions

Here's a rebase of this patch series.  I didn't change anything except

1. disregard what was 0005, since I already pushed it.
2. roll 0003 into 0002.
3. rebase 0007 (now 0005) to account for the reorderbuffer changes.

(I did notice that 0005 adds a new boolean any_data_sent, which is
silly -- it should be another txn_flags bit.)

However, tests don't pass for me; notably, test_decoding crashes.
OTOH I noticed that the streamed transaction support in test_decoding
writes the XID to the output, which is going to make it useless for
regression testing.  It probably should not emit the numerical values.

-- 
Álvaro Herrera                https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services