Re: Need for PostgreSQL demand? - Mailing list pgsql-advocacy

At 4:21a -0500 on 05 Nov 2007, ioguix wrote:
>> This is why I tried hard to promote Once:Radix when it open-sourced.
>> However, the reception from our community has been lukewarm at best.
>> Personally, I see a Filemaker replacement which runs on PG to be a
>> killer app, but nobody wants to listen to me, apparently.
>
> I am 100% agree with you about this kind of "killer app" as I was
> looking for such a project some time ago...

I think it's safe to say that a fair number of us on this particular
list are looking for such a project ... :-P

> Actually, there's one project which is on this way : Glom[1]. It's
> getting better and better, but probably needs some more hands to grow
> faster and stable enought for a professionnal use...

This is perhaps my lack-of-awareness of the general playing field, but
neither of these projects seem to me to be of the "killer app" variety.
They certainly do have their merits, though.  Indeed, Once:radix /does/
look cool, and I will take a more in-depth look at it after the SC
conference, but my initial impression is that it is geared toward a
small niche of the (web) developer segment.  Similarly, Glom also looks
cool, but seems at second glance to be a glorified schema interface.  (I
mean no disrespect towards either project, but them's my impressions.)
In terms of massively gaining Postgres popularity, I think a killer app
will take away the developer-esque feel of working with the database.

The problem, as I see it, is that the term "killer app" means different
things to different people and different functions for different
environments.  For example, people often tout Photoshop as being their
killer app for why they'd never use *nix.  For the way I work and what I
do, however, my killer app set is ssh, user-inet tools (web browser,
mail client, rss reader, voip), (multi)text editor, and lots of seamless
virtual desktops.  I don't have that combination in Windows by a long shot.

That said, *my* first thought on a killer app for this era (~2008) lies
more along the lines of relying heavily on the database but hiding that
fact from the end-user.  I know that it has so far proven infeasible,
but what about something like a DB-backed filesystem?  Having the
ability to ask the filesystem for what I want (in terms of tags,
identifying characteristics, etc) rather than having to keep everything
organized myself ... well, I won't go into the merits of the concept as
I'm sure y'all know way more than I.  The point is that a project of
this nature would use a DB but hide that fact from the end-user.  I
don't know if Postgres would fit in this role, but all the qualities of
the DB would shine through (ACID, speed, parallel usage, fine-grained
security, ...)

A killer app for Postgres would be something that absolutely uses the
strengths of Postgres.  Complexity of it aside, the FS idea would
brilliantly highlight the robustness of Postgres, a point that would
instantly drop some other DB's from the running.

Short of a killer app, though, I don't agree with the original thought:

At 9:37p -0400 on 02 Nov 2007, Ron Peterson wrote:
>>>> I think the best possible PostgreSQL marketing would be for someone to
>>>> develop a knock-your-socks-off killer app.  It wouldn't have to be

The "best possible marketing" is difficult to ascertain, but I think a
broader approach would work just as well.  Rather than making a killer
app (which is admittedly less work, but only *after* you find the idea),
create apps that utilize Postgres /and/ its strengths.  Along these
lines, here are a couple not-so-killer apps, but apps and areas that are
important nonetheless and that I'm fairly certain would benefit from
Postgres and OpenSource TLC:

- University-style ERP (Banner, PeopleSoft, ...)
  * AFAIK no Postgres based version exists

- Web/distributed Financial applications
  * LedgerSMB is great, and we use it, but it needs a fair amount work
    as well

- A genealogy project (mentioned by Leif earlier)
  * Don't know of any official projects
  * I can't count how many times genealogy and genealogy records have
    come up in the past year.  I think this would make an /excellent/
    history/CS project for an undergrad student to start.

- Blogging for high-traffic websites
  * I'm certain these exist, but I don't know of any that rely on
    Postgres other than as a plugin to Drupal

I'll stop there, but you get the idea.  The last two are admittedly not
of the large-style projects at which Postgres excels, but would go a
long way towards getting hobbyist folks involved and aware.

I'm hesitant to say much more of what's on my mind because I just don't
know, and I fear I may have already said something false.  (/me waits
for TL/GS/JD/JB/DF/RT/MM to correct some fallacy).  :-)

Cheers,

Kevin