Ingres Engineering Summit
Roy Hann
so I thought I'd report a few bits and pieces that captured my attention.
First of all, it was an open event, as befits an open source company.
Anyone was welcome to attend and a great many partners (and others) did. I
don't know the total headcount but considering it was a moderately expensive
week and a long time out of the office, it was very well attended. I'd
guess there were well over 100 people there.
The amount of openness demonstrated was impressive. It wasn't just about
exposing the code, it was also about exposing the thinking, the working
practices, the people, and--yes--the tensions. With the exception of a
one-hour slot that was employees-only, everything was out in the open.
Anyone who didn't attend has missed out on a lot. Plan to attend next year.
Just running through the agenda from top to bottom, and focussing just on
Ingres rather than OpenROAD, here are some highlights:
Andrew Ross did a one hour presentation that covered a lot of ground on the
general theme of "community development", and more precisely, the barriers
to community development and what has to change to make it easier/possible.
As we all know, there was a lot standing in our way, and there is a long way
to go still. However the need for a properly open process was as taken as
given. As well as reporting a lot of progress with things like
http://code.ingres.com (Subversion code repository), http://lxr.ingres.com
(code cross-referencer), and http://bugs.ingres.com (bug tracking), Andrew
also outlined a number of as-yet unresolved problems. I think I'll leave it
to Andrew to elaborate on those, but the big one (IMO) is the communication
channels. Mike Sale, Mike Leo and I have been discussing this too. One
thing we agree on is that the current phpBB-based Ingres forums are an
embarrassment and have to go (real soon if I have my way). Mike Leo's
suggestion of using vBulletin turns out to be top of the list at Ingres too,
so that could happen fairly quickly. There are several benefits to
vBulletin but the big one is that it will allow us to have a single
community delivered via web pages, e-mail, or NNTP, so that we can
accommodate everyone's way of working. IRC will of course remain separate.
Another big item from Andrew's presentation is that there are actually two
fairly successful virtual development systems about ready for delivery. We
should be able to get our hands on these within a week or so.
I won't dwell on the rest of the presentations in such detail. You can
infer the significance of these topics appearing on the agenda as well as I
can.
We had a number of demos of things like Ingres Café and some OpenGIS
software. There was a presentation from Gordon Thorpe on the formidable
challenge of re-architecting GCA. Hugh Darwen spoke about Project D (an
implementation of Tutorial D on top of Ingres) and earned himself the second
prize for Best Presentation. We had two presentations on column stores, one
of which went on to win the first prize Best Presentation (Marcin Zukowski
on Monet/x100). Karl spoke fluently about something or other for an hour
and many of us marvelled. Mike Touloumtzis led a discussion about how to
implement column encryption; nothing was decided but a lot of ground was
explored. Steve Ball and Alison Stillway led another discussion on how to
implement MVCC and which model of MVCC to adopt; nothing was decided except
that a design document will be drafted for public comment. (To my mind this
may be the most immediately and widely useful enhancement that popped up on
the new-features radar.) Emma McGrattan picked up where Andrew Ross left
off. The big thing in her presentation was the carefully expressed and
several-times repeated instruction that Ingres Corp requires all its
personnel to devote 10% of their time to community projects (i.e. 1/2 day
per week). That is a lot of effort folk! After Emma, Kai-Uwe Sattler
talked about some research into making Ingres more autonomous and
self-tuning (including the ability to recommend secondary indices and
statistics). There were two presentations on two different replicators from
partner companies. Roger Whitcomb told us about the work he's been doing
on Ingres Management Tools, which was really good stuff. (I was present at
the meeting where VDBA was first unveiled back circa 1995 and it was greeted
with horror and revulsion then, and nothing has changed. Roger's work is
definitely going in the right direction this time.) After this I stepped
out of server-land and saw Daryl Monge discussing Ruby on Rails. For some
reason every head in the room turned to look at me when he reminded us that
RoR requires every table to have a synthetic integer key. Evidently
everyone understands this is wicked and wrong and that they should feel
guilty about it, but equally evidently people are just going to go on doing
it anyway. That was enough for me, so I retreated back to server-land again
after that.
So there you are. I saw fewer than half the presentations, so perhaps
someone else will comment on the others.
Finally , I am hoping to get at least a couple of these repeated at the IUA
conference on June 17 in London. Please let me know if there is anything
above that particularly takes your fancy and I will see what I can do.
Roy
Michael Leo
Mike Leo
Roy Hann wrote:
_______________________________________________ Info-Ingres mailing list Info-...@kettleriverconsulting.com http://www.kettleriverconsulting.com/mailman/listinfo/info-ingres
Jonah H. Harris
<spec...@processed.almost.meat> wrote:
> As well as reporting a lot of progress with things like ...
> http://lxr.ingres.com (code cross-referencer)
Wow! I had never heard that they set up LXR for Ingres! That will
certainly make the code a bit easier to understand. Also, I'm happy
to hear that Steve is looking into MVCC. There's a few cases where
Ingres' locking was a pain, and MVCC will nearly eliminate that area
of contention. I really look forward to their design document and
seeing an MVRC architecture added to Ingres.
Similarly, it also looks like the Ingres community is growing. It's
difficult to build a community, and it takes time to evolve the
business model into something acceptable to both open-source
contributors/users and the business itself. But, it looks like
they're on the right track.
Thanks for posting your summary and I hope you're doing well.
--
Jonah H. Harris, Sr. Software Architect | phone: 732.331.1324
EnterpriseDB Corporation | fax: 732.331.1301
499 Thornall Street, 2nd Floor | jonah....@enterprisedb.com
Edison, NJ 08837 | http://www.enterprisedb.com/
Karl & Betty Schendel
On Apr 29, 2008, at 12:51 PM, Roy Hann wrote:
> ... Karl spoke fluently about something or other for an hour
> and many of us marvelled.
I was going to elaborate and expound, but after taking thought,
I'd say that Roy got it just about right. :-)
Actually it was about the work that we've done at Datallegro to
enhance Ingres. I won't go into details, because it's inaccessible
to the rest of the community at the moment. Many of the discussions
I had at the summit were aimed at correcting that situation, and
I feel like we are making progress. Datallegro has a significant
backlog of contributions built up, which makes our situation
rather more difficult than it might be otherwise.
I thought that this Summit was extremely enjoyable, useful, and
well-run. I didn't get to see and hear everything I wanted to,
but to a certain extent that's the nature of the beast.
It was great to (re-)connect with the usual notabilities
of Ingres geek-dom, and it was equally great to see some new
faces. I always learn something about GCA when Gordy talks.
The MonetDB presentation was particularly exciting, because
I've had many of the same ideas (thus validating my^H^H their
approach...!); the numbers they are getting in their
research are extremely interesting. MVCC has been talked
about for years now, so I am very happy to see some serious
design work being done on it.
As a final note, if anyone tries code.ingres.com and gets
a funky SVN error message, try
http://code.ingres.com/ingres/
and you'll feel much better.
Karl
Michael Leo
> MVCC has been talked
> about for years now, so I am very happy to see some serious
> design work being done on it.
>
>
Mike Leo
Kettle River Consulting Inc
Daryl L. Monge
"Roy Hann" <spec...@processed.almost.meat> wrote:
> After this I stepped
> out of server-land and saw Daryl Monge discussing Ruby on Rails. For some
> reason every head in the room turned to look at me when he reminded us that
> RoR requires every table to have a synthetic integer key. Evidently
> everyone understands this is wicked and wrong and that they should feel
> guilty about it, but equally evidently people are just going to go on doing
> it anyway.
Muahahahahahaha Dr. Evil will force synthetic keys ON YOU ALL. ;-)
But back to reality:
http://community.ingres.com/forums/viewforum.php?f=9
and
http://community.ingres.com/wiki/Ruby_Driver
Daryl
"This isn't reality; this is fantasy" - Uhura