From: | Andrew Gierth <andrew(at)tao11(dot)riddles(dot)org(dot)uk> |
---|---|
To: | david(at)kineticode(dot)com ("David E(dot) Wheeler"), tgl(at)sss(dot)pgh(dot)pa(dot)us (Tom Lane), pgsql-hackers(at)postgresql(dot)org |
Subject: | Re: revised hstore patch |
Date: | 2009-07-22 18:17:49 |
Message-ID: | 87prbsk2eq.fsf@news-spur.riddles.org.uk |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
>>>>> "David" == "David E Wheeler" <david(at)kineticode(dot)com> writes:
>> The other option would be to fix the wasted-space bug in the
>> existing hstore, and document that stored data must be updated at
>> least once subsequent to that fix before doing a binary migrate to
>> 8.5.
[...]
David> Could it be that only folks with such a manifestation of the
David> bug would need to do a binary upgrade? If so, I certainly
David> think it'd be worth it to fix the bug.
Let's go through the options.
A)
- don't fix the wasted-space bug (or don't rely on it, anyway)
- change the new format to be more distinguishable
Result:
- seamless binary upgrade for contrib/hstore users
- users of unreleased CVS hstore-new will have to ensure all
values are updated after installing a release version and
before doing a binary upgrade to 8.5
B)
- fix the wasted-space bug
- leave the new format as-is
Result:
- seamless binary upgrade for hstore-new users
- contrib/ users will have to remove wasted space from at least
any hstore with a zero-length key before doing a binary upgrade
To me (A) is looking like the obvious choice (the people smart enough
to be using hstore-new from CVS already can handle the minor pain of
updating the on-disk format).
Unless I hear any objections I will proceed accordingly...
--
Andrew (irc:RhodiumToad)
From | Date | Subject | |
---|---|---|---|
Next Message | Dimitri Fontaine | 2009-07-22 18:46:45 | Extension Facility |
Previous Message | Robert Haas | 2009-07-22 17:45:50 | Re: revised hstore patch |