Re: Information on savepoint requirement within transctions

On Fri, Jan 26, 2018 at 9:47 AM, Melvin Davidson <melvin6925(at)gmail(dot)com>
wrote:

> > ...the presence of a version is bad, only the badgering of people asking
> questions to provide it when it has no bearing on the answer...
>
> Really? Is it that hard for someone to provide version and O/S?
>

​Its difficult to remember to include it when it should make no difference
in receiving an answer the question being asked today.

> >The timestamp on the email is likely more than sufficient
> Do you really think op's look for the timestamp of answers as opposed to
> content?
>

Maybe not, but in 5 years if someone pulls up this thread and sees that the
OP was talking about 9.6 what are they supposed to do with that
information? They would have no way to know whether things have changed
between 9.6 and 15. Its more in their face and they might be more
skeptical but they are still going to try using the information if the
topic seems similar.

> As someone that spent 8 years in high level tech support for Ingres (the
> predecessor to Postgres), I assure you
> the inclusion of version and O/S is critical for historical purpose.
>

​The -general mailing list is not tech support, its a social forum. -bugs
is tech support and for that list the community does indeed post a request
that version and O/S information be provided, and even has a form-field to
be filled in.

If you feel so strongly then by all means add version and O/S information
to all of your responses. I'm doubtful it will generally be of benefit and
even less certain that having that information appear on the 3rd email (and
maybe only the third if it gets trimmed away during responses) in the
thread would be an improvement. One cannot control the initial email and
by the time you ask and give an answer anyway the second one is gone and
the thread may be done with (if one provides sufficient and correct
advice). Having been given an answer I doubt most people would reply:
"thanks, and for the record my version and O/S is such-and-such". Besides,
they may still be on 9.3 while the response covers 9.3-10; it is the
version of the response that would seem to matter more than the version the
questioner happens to be using and fails to provide up-front anyway.

David J.