Re: Add ZSON extension to /contrib/ - Mailing list pgsql-hackers

From Aleksander Alekseev
Subject Re: Add ZSON extension to /contrib/
Date
Msg-id CAJ7c6TMSat6qjPrrrK0tRTgZsdXwFAbkDn5gjeDtFnUFrjZX-g@mail.gmail.com
Whole thread Raw
In response to Re: Add ZSON extension to /contrib/  ("Joel Jacobson" <joel@compiler.org>)
Responses Re: Add ZSON extension to /contrib/
List pgsql-hackers
Hi hackers,

Many thanks for the feedback and all the great suggestions!

I decided to add the patch to the nearest commitfest. You will find it in the attachment.

Differences from the GitHub version:

- Code formatting changed;
- More comments added to the code;
- SGML documentation added;
- Plus several minor changes;

I very much like the ideas:

- To use varint, as Tomas suggested
- Make dictionaries variable in size
- Somehow avoid calling gettimeofday()
- Improvements by 2ndQuadrant that Andrew named

However, I would like to decompose the task into 1) deciding if the extension is worth adding to /contrib/ and 2) improving it. Since there are people who already use ZSON, the extension should be backward-compatible with the current ZSON format anyway. Also, every improvement deserves its own discussion, testing, and benchmarking. Thus I believe the suggested approach will simplify the job for reviewers, and also save us time if the patch will be declined. If the patch will be accepted, I will be delighted to submit follow-up patches!

If you have any other ideas on how the extension can be improved in the future, please don't hesitate to name them in this thread. Also, I would appreciate some code review.

--
Best regards,
Aleksander Alekseev
Attachment

pgsql-hackers by date:

Previous
From: Pavel Borisov
Date:
Subject: Re: Parallel scan with SubTransGetTopmostTransaction assert coredump
Next
From: Greg Sabino Mullane
Date:
Subject: Re: Update maintenance_work_mem/autovacuum_work_mem to reflect the 1GB limitation with VACUUM