Re: SELECT .. FOR UPDATE: find out who locked a row

Tom,
this whole discussion started because Enrico did not originally specify the
PostgreSQL version he was working with. So after he did advise it was for
9.6, I felt it necessary to explain to him why a certain section of my
query was commented out and that it would also work for 10. I have
previously made it a policy to request ops include the PostgreSQL version
and O/S when submitting to this list, but I was berated by others for
always requesting ops to provide that extremely difficult information to
obtain.
I also felt it important that I express my opinion that the changes needed
were caused by what I felt was cosmetic and unnecessary changes to the
catalog. There is an old saying "If it ain't broke, don't fix it" and that
certainly applies here.

Now, as to your request for me to read the thread in the url's you
suggested, I did read most of the content. I note that the entire
discussion was amongst
PostgreSQL-development <pgsql-hackers(at)postgreSQL(dot)org>, So at no time
was the generic population of PostgreSQL users and DBA's involved.
Therefore, said population, myself included, had no foreknowledge of the
intended changes which is the cause of the problem. Therefore your
statement "you might want to consider speaking up in some reasonable time
frame, not six years later" is abrasive at best, since I, and others, only
found out about it after the fact. Not to mention, even if I did complain
earlier, I seriously doubt the changes could or would be reversed.

At this point I have said all I have to say and will discuss it no further.
I can only strongly recommend that in the future, proposed changes to
system catalogs that could adversely affect existing scripts and
applications be sent to the generic PostgreSQL population (IE:
pgsql-general(at)lists(dot)postgresql(dot)org) for comment BEFORE said changes are
implemented.

On Thu, Mar 15, 2018 at 11:23 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:

> Melvin Davidson <melvin6925(at)gmail(dot)com> writes:
> > Yes, Stephen, I certainly understand making changes to system catalogs
> > _when necessary_. That being said, the first change was the renaming of
> > pid to procpid in pg_stat_activity. However, I contend that was more
> > because someone felt that it was more to make the column names
> > consistent across catalogs, rather than necessity.
>
> Please read all of
> https://www.postgresql.org/message-id/flat/201106091554.
> p59Fso314146%40momjian.us
> where this was discussed to death (and rejected), and then read all of
> https://www.postgresql.org/message-id/flat/CAKq0gvK8PzMWPv19_o7CGg8ZQ0G%
> 2BUuAWor5RrAg0SOmWTqqLwg%40mail.gmail.com
> which is the thread in which the change was agreed to after all
> (on the grounds that we were breaking backwards compatibility of
> the view anyway with respect to other, more important, columns).
>
> If you still feel that we make incompatible changes without adequate
> consideration, that's your right, but you might want to consider
> speaking up in some reasonable time frame, not six years later.
> This could have been objected to as late as 9.2 beta, so it's not
> like you need to be drinking from the pgsql-hackers firehose continually
> in order to weigh in. But 9.2 is not just released, it's EOL, so it's
> really kinda late to be objecting.
>
> regards, tom lane
>

--
*Melvin Davidson*
*Maj. Database & Exploration Specialist*
*Universe Exploration Command – UXC*
Employment by invitation only!