Re: Counting distinct names

From: Peter Eisentraut <peter_e(at)gmx(dot)net>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: Christopher Sawtell <csawtell(at)xtra(dot)co(dot)nz>, "pgsql-sql(at)postgreSQL(dot)org" <pgsql-sql(at)postgresql(dot)org>
Subject: Re: Counting distinct names
Date: 2000-04-21 22:06:26
Message-ID: Pine.LNX.4.21.0004211909460.351-100000@localhost.localdomain
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-sql

Tom Lane writes:

> *Should* there be a default location established by configure?

Not IMO. Firstly (broken record time), configure should not be used to set
run-time options period. (And there's a single good reason for that: when
users build a package they won't know what options they need later on. I
get really annoyed at packages that do that.) Secondly, this will lead to
all kinds of funny problem reports because everyone will have a different
default. Worse yet, people might not even know what their default is
because they use a binary package. I think the current setup is the
cleanest by far because it emphasizes the independence of the data storage
location and the fact that multiple locations are equally valid.

> I think I'd prefer that to having the dependence on an environment var

Environment variables are a common way of pointing programs to their data.
Personally, I rarely bother with it and start the postmaster with -D
instead, then I know what I'm getting. I suppose there's no reason that
users couldn't do the same. The environment variable is only wildly useful
if people start and stop the server all the time.

--
Peter Eisentraut Sernanders väg 10:115
peter_e(at)gmx(dot)net 75262 Uppsala
http://yi.org/peter-e/ Sweden

In response to

Browse pgsql-sql by date

  From Date Subject
Next Message Tom Lane 2000-04-21 22:54:22 Re: TOAST (was: BLOB)
Previous Message Peter Eisentraut 2000-04-21 22:05:47 Re: Counting distinct names