On Tuesday 17 September 2002 11:22 pm, Bruce Momjian wrote:
> This is a better description tha[n] I could make. If you look at the
> script it is very well commented so you should be able to see it works.
> Also, read the manual page first.
I don't know how, but this time looking at the script, I just grokked it.
Maybe that's because it finally clicked in my mind what was happening;
regardless, thanks for the compliment; feel free to use that, edited as
necessary, in any documentation you might desire.
But you are certainly correct about the comments...some of which are more than
a little tongue in cheek...
# Strip off the trailing directory name and store our data there
# in the hope we are in the same filesystem so 'mv 'works.
:-)
> However, we do make releases
> more frequently than commercial folks, so the pain is more consistent.
Well, for me and Oliver it comes in waves -- every major release has its
paroxysm. Then things cool off a little until next cycle. These one year
cycles have, in that way, been a good thing..... :-P
You know, if the featureset of the new releases wasn't so _seductive_ it
wouldn't be nearly as big of a problem...
> You could pretend you are
> using MySQL and just not upgrade for 5 years. ;-)
Don't say that too loudly, or my production 6.5.3 database that backends the
larger portion of my intranet will hear you....I'm just now moving the whole
shooting match over to 7.2.2 as part of our delayed website redesign to use
OpenACS. That dataset started with 6.1.2 over five years ago, and it was the
6.2.1->6.3.2 fiasco Red Hat created (by giving no warning that 5.1 had 6.3.2
(5.0 had 6.2.1)) that got my dander up the first time. I lost a few thousand
records in that mess, which are now moot but then was a bad problem. Since
there wasn't an official Red Hat RPM for 6.1.2, that installation was from
source and didn't get obliterated when I moved from Red Hat 4.2 to 5.0. I
was able to run both 6.1.2 and 6.2.1 concurrently, and the migration went
smoothly -- but there were less than ten thousand records at that point.
So I _do_ have a three-year old database sitting there. Rock solid except for
one or two times of wierd vacuum/pg_dump interactions, solved by making them
sequential.
--
Lamar Owen
WGCR Internet Radio
1 Peter 4:11