[ZCommerce] RE: Philip leaves Arsdigita (was: Re: [Zope] kerberos ? + LDAP + ecommerce + ZEO replication etc)

Albert Langer Albert.Langer@Directory-Designs.org
Tue, 3 Apr 2001 09:32:44 +1000


I agree with your suggestion that ecommerce based on OpenACS4
should be "fishbowled" and will even count that as
expressing "interest".

However after nearly a year, and at this particular moment, it
is nowhere near *enough* interest because of events that have
just happened and are happening right now.

I also agree that in asking DC to put some resources into it:

'...the question that needs to be answered is "How does it immediately
help DC to reimplement the ACS ecommerce module inside Zope?"  The answer so
far hasn't been compelling enough for a project to be created.'

and that:

'DC has limited resources, and making a committment to port the
ACS ecommerce module can't possibly be done without being able to see a
tangible monetary benefit to DC as a company.'

How *could* a compelling enough answer to that question for a
project to be created be available if nobody from DC has
actually studied the recent events to seek an answer?

So who is to answer that question? A grassroots community effort?
Why should we, if it so happens that we merely wish DC well and are
very appreciative of DC having made possible the benefits of Zope
for the things we *are* interested in, but do *not* have a focus
on tangible monetary benefit to DC as a company?

The only way *that* question can be answered is by DC staffers
being assigned to take a look and answer it. It can't be answered
by me having said that I've taken a very close look and concluded
that DC would have the most to gain from getting it done quickly.
Nor can it be answered by you off the top of your head.

It can only be answered by someone from DC taking the time (I
believe your estimate of a week, made nearly a year ago is
reasonable), to checkout what's been happening with arsDigita
and OpenACS very recently and thoroughly review both
the ACS4 documentation and code and what interest DC might
have in a solid turnkey ecommerce package and an escape route
from Oracle in it's commercial contracts.

It certainly can't be answered by waiting to see if someone
*else* stands to gain enough from this job being done to
take up the mantle of actually carrying it through.

All I'm going to do is make suggestions. I'll certainly
help if there *is* a fishbowl project, but I'm in no
position to get one started. I doubt that *anyone* other
than DC "stands to gain the most benefit".

Likewise the various people who have been working on
various Zope based ecommerce projects would be very
likely to help but have shown no likelihood of being
able to deliver what is needed to carry it through
without some sort of coordinated effort sponsored by
DC. You *don't* need to wait to see whether a
grassroots effort is the solution. There has been one
for more than a year in the zcommerce list and it
very obviously is *not* the solution. Has DC gained
any "tangible monetary benefit" from that at all?

vvvvv
> What more could a Zope guru want to persuade them to take a
> look at the possibility of demonstrating how
> much better that fits with Zope than with java or Tcl?

Manpower.  ;-)
^^^^^

Well manpower is one of the major benefits you could get
from taking that look.

You've already got grassroots efforts doing "presentation"
of ecommerce with Zope and there's no big problem doing
the catalog side using ZODB (though an ORDBMS has
advantages there too).

But your core competency does not lie in ORDBMS SQL and
neither does that of the grassroots Zope community.
(Nor in the simpler matter of protocols communicating
with payments gateways for that matter - wampum is
still "pre-alpha" and there is nothing for any
service except cybercash - which is in deep trouble).

An entirely separate company has put major resources
into an SQL data model that is freely available open source.
That saves an awful lot of manpower, just as many
companies using Zope have been saved a lot of manpower
by the resources DC has put into it.

An entirely separate open source community is right
now putting major grass roots resources of highly
skilled SQL programmers into porting the SQL so
that data model can be used with *both* Oracle
*and* Postgresql 7.1 and separating it entirely
from the Tcl and java - mainly for convenience
in doing that - but with the obvious consequence
that it will be much easier to move it from the
relatively clumsy web application framework
they have got, to Zope. That saves an awful lot
of manpower too.

A third open source community has put major
resources into turning postgresql 7.1 into
an industrial strength RDBMS quite capable
of replacing Oracle in many situations,
which is also about to get python as a
built in backend procedural language.

That's not so much a savings in manpower
as something that couldn't be done with
any amount of manpower from DC, but is
now available to DC.

Now DC seems to have had to put quite
a bit of resources into maintaining an
Oracle adaptor for python, which I suspect
is not entirely within your core competency.

I would guess that there would be
at least *some* contracts that DC depends
on for revenue where the customer spends
at least as much on Oracle licenses as
they do on consulting, so you ought to be able
to figure out that there exists at least
a possibility worth investigating that
being able to demonstrate that what they
want works with Zope *both* on Oracle
and on a free ORDBMS could result in
some tangible monetary benefit to DC
(even if they only have to buy
*less* Oracle licenses with Zope/ZEO
acting as an intermediary between
postgresql backed web servers and
Oracle stuff working with internal
systems).

I would also guess that there would
be at least *some* contracts that
DC has *not* gained because the
reason the customer wants the
sort of things Zope does provide
is closely connected with also
wanting to charge for various
goods and services that go with
it and they decide they would
prefer to deal with consultants
that have a better track record
on ecommerce.

I'm just guessing of course. Only
DC can determine whether these,
and/or other matters might
result in tangible monetary
benefit to DC.

But the direct benefit to DC that I
highlighted was the benefit that
is also a direct benefit to the
whole Zope community that I am
more interested in.

The more widely Zope is known
and used the stronger it grows
and the more consulting revenue
DC gets, since people who can
afford to pay for consultants
are quite happy to do so, but
like to have heard that what
they want specially tailored
is widely adopted and mainstream.

ISPs already have web servers
and don't need Zope for what
they are currently doing
(though if *they* had manpower
to take a close look, which
they generally don't, they'd
see some good reasons for
using Zope too).

They also need to put in
"commerce servers" and come
up with some quite bizarre
solutions using perl and
MySQL or else have to pay through
the nose.

If Zope had an industrial
strength turnkey commerce
server it would be widely
used by ISPs and would become
"mainstream" like Apache -
the Zope community would
benefit greatly and DC's
consulting revenue would
grow greatly.

Now how much do *I* get paid
for trying to get DC to take
a look?

If you win this argument you
aren't going to get any tangible
monetary benefit for DC either.

If somebody *does* take a look
you just might. Think about it.

-----Original Message-----
From: zope-admin@zope.org [mailto:zope-admin@zope.org]On Behalf Of Chris
McDonough
Sent: Tuesday, April 03, 2001 7:25 AM
To: Albert.Langer@Directory-Designs.org; 'Michael R. Bernstein'
Cc: 'Mayers, Philip J'; zope@zope.org; zcommerce@codeit.com
Subject: Re: [ZCommerce] RE: Philip leaves Arsdigita (was: Re: [Zope]
kerberos ? + LDAP + ecommerce + ZEO replication etc)


> Now back to the main story...
>
> ACS4 SQL is even using a Domain Object Model pattern with properly
> OO separation of a meta-data knowledge layer from operational
> layer in the RDBMS, with an access control system acquiring
> permissions from context, sophisticated Party/Group/Roles
> relationships that can express the whole UML concept of
> associations and based on accountability patterns
> from Fowler's "Analysis Patterns", a powerful petri net based
> workflow engine etc etc. Arsdigita has stopped Tcl development
> and moved to java.
>
> What more could a Zope guru want to persuade them to take a
> look at the possibility of demonstrating how
> much better that fits with Zope than with java or Tcl?

Manpower.  ;-)

> Oh yes - one more thing - an important Open Source community that
> has RDBMS skills somewhat lacking in the Zope Community,
> currently faced with upheavals in Arsdigita and actively working
> on a demonstrably viable escape route from Oracle and
> welcoming lots of new involvement.
>
> Seems that isn't enough though. No sign of interest from DC.

Lots of interest, not a lot of time.  This is a great fishbowl idea and
community development effort idea.

> Anybody listening?
>
> See my original posting:
> http://lists.zope.org/pipermail/zope/2001-March/086365.html
>
> Plain fact is that Zope *still* lacks an industrial strength
> RDBMS based ecommerce system that could have a huge impact
> in getting Zope widely deployed at ISPs, which is critical
> to getting lots more interest. Postgresql is now
> *fully* "industrial strength" (outer joins, better
> performance than Oracle and a python procedural language
> much better than Oracle java or PL/SQL just released).
>
> Right now when *lots* of people are getting involved in
> OpenACS and are doing a port of ACS4 for *both* Oracle
> and Postgresql 7.1 the opportunity is wide open for
> getting real momentum behind adding the strengths of
> the ACS RDBMS datamodel to the strengths of Zope. All
> the SQL previously mixed in with Tcl cruft is just
> sitting there waiting to be Zoped.
>
> The reasons for deferring it earlier ("it would take
> at least a week to study" and "got to resolve API
> documentation first") cannot possibly still apply.
>
> Everything else that could possibly be going well for
> Zope is obviously going amazingly well, and will continue
> to do so if somebody did spend a week or two on taking
> a look. Yet this key aspect for getting wide deployment
> is still being neglected.
>
> I haven't been hassling anybody about this for nearly a
> year, so I guess it's time for some CCing.
>
> Sure hope somebody who can do something about it reads
> the above link to previous posting concerning the
> ACS4 developments - and the links within it and then
> assigns some actual *resources*.
>
> On the first anniversary of having first pointed out
> that a viable ecommerce package would not happen
> without DC involvement and without checking out ACS,
> I go on the war path. That's not far off ;-)
>
> You don't get turnkey ecommerce packages
> without actual resources being put into a coordinated
> effort. It's the coordination and serious interest
> and commitment from DC that is lacking - there's no
> lack of prototypes showing that Zope is a perfectly
> suitable platform for the UI web side of ecommerce
> as well as content management (when and only when
> used together with an industrial strength RDBMS such
> as postgresql - nobody in their right mind does
> serious ecommerce without that). A project like
> the CMF could get this done *fast*.
>
> BTW another recent posting not responded to re
> Zope SSL support is pretty relevant to this as well:
>
> http://lists.zope.org/pipermail/zope/2001-April/086937.html

I think the question that needs to be answered is "How does it immediately
help DC to reimplement the ACS ecommerce module inside Zope?"  The answer so
far hasn't been compelling enough for a project to be created.  That's not
to say it won't become so in the future or that it's not a worthwhile
effort.  But DC has limited resources, and making a committment to port the
ACS ecommerce module can't possibly be done without being able to see a
tangible monetary benefit to DC as a company.

OTOH, look at ZPatterns.  DC literally had (and still has) nothing
whatsoever to do with ZPatterns.  However, lots of people use it, it's quite
popular.  I think the same could be done with a grassroots effort to get the
ACS module moved to Zope.  DC did provide help to Phillip and Ty in the way
of implementing changes to Zope that made it easier for ZPatterns to do what
it does, because we understand that it's a good long-term investment to do
so.  I think the relationship with DC and a group of folks who wanted to
port ACS ecommerce to Zope would probably best follow the same pattern.

I suggest that this project be fishbowled and those who are interested and
who stand to gain the most benefit take up the mantle of actually
implementing it.

- C



_______________________________________________
Zope maillist  -  Zope@zope.org
http://lists.zope.org/mailman/listinfo/zope
**   No cross posts or HTML encoding!  **
(Related lists -
 http://lists.zope.org/mailman/listinfo/zope-announce
 http://lists.zope.org/mailman/listinfo/zope-dev )