From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Thomas Munro <thomas(dot)munro(at)enterprisedb(dot)com> |
Cc: | Robert Haas <robertmhaas(at)gmail(dot)com>, Pg Hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Parallel safety of binary_upgrade_create_empty_extension |
Date: | 2018-03-30 21:26:21 |
Message-ID: | 12685.1522445181@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
I wrote:
> Anyway, that looks like 11 functions that there's no question we
> need to relabel. I'll go do that.
Oh, while I'm looking at this, I notice that cursor_to_xml[schema]
is not merely mismarked as to its parallel safety, but its volatility:
it's marked stable, which is frickin' insane considering it has side
effects on the state of the cursor.
query_to_xml* are likewise marked stable, which seems also foolish
since we have no idea whether the source query has side-effects.
All of this breakage is pretty old. I guess the best we can do is
fix it without a catversion bump in the back branches.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Jeremy Finzel | 2018-03-30 21:36:59 | Passing current_database to BackgroundWorkerInitializeConnection |
Previous Message | Tom Lane | 2018-03-30 21:09:15 | Re: Parallel safety of binary_upgrade_create_empty_extension |