Re: Time to work on Press Release 8.0 - Mailing list pgsql-advocacy

Quoth scrappy@postgresql.org ("Marc G. Fournier"):
> On Fri, 13 Aug 2004, Jan Wieck wrote:
>
>> Every time PostgreSQL is discussed, be that slashdot or LWN or other
>> forums or even interviews with any of out competitors, the statement
>> is "doesn't have replication". If we avoid the topic in our 8.0
>> press release and leave out the word replication, instead of
>> pointing to the reasons why it is *better not to have replication
>> builtin*, we will lose. People will read the press release, don't
>> see the word replication, so "No Replication - checkmark".
>
> Except that "Have Replication" to those claiming 'No Replicatin' means
> "Integrated with the Server", we still don't *have* Replication in
> their minds, no matter how many external projects that do it we
> mention ;(

Perhaps.

Another feature that is emerging is Peter and Fabian's work to make it
easier to have a build environment that _isn't_ a PG source tree that
you can use to compile "extensions" against.

That's pretty key to the ability for people running "packaged"
distributions to be able to easily deploy extensions.  It's certainly
a prerequisite to having plenty-o-language extensions for any systems
that deploy code in binary form.  And I'm not sure that BSD Ports is
_totally_ comfortable with there needing to be packages that are
source installs; it looks as though their users kind of like to do a
"make clean" to drop out the deteriorata once one is done installing a
package.

The improved "working infrastructure for pg extensions" aka pgxs is
well worth pointing out as a way of letting there be a whole lot more
extensions that are simultaneously:

  - Decoupled so that they may be pushed _OUT_ of the source tree, and
    yet
  - Not turned into a Huge Pain In The Neck To Compile.

This has the substantial merit that new things that are, to coin a
phrase, "pgxs-compliant," can be treated as new features that, while
not included in the strict "PostgreSQL source tree," are still readily
available.

For someone that's considering what database system they should choose
to use, it would be foolish to ignore the software that sits alongside
it, readily integrable, no?

If "pgxs" makes it possible to take most of the stuff presently
sitting in "contrib" and eliminate it from the source tree of
"PostgreSQL, the Database Proper," then that _drops_ the amount of
functionality found in "The Database, Proper," which sure looks like
the wrong message to send out if what's really happening is that it
has been made _EASIER_ to have plenty-O-extensions.

You'll notice, I trust, that I never used the word 'replication' in
any of the above.  I daresay I'm biased from several perspectives,
nicely illustrated in that while I'm writing this, I'm monitoring the
installation of an ERS instance that's going to get used for a
migration to Slony-I ;-).
--
wm(X,Y):-write(X),write('@'),write(Y). wm('cbbrowne','acm.org').
http://www.ntlug.org/~cbbrowne/internet.html
I hate wet paper bags.