[Zope3-dev] Re: [Zope3-checkins] CVS: Zope3/lib/python/Zope/App/OFS/Services/SmtpService/Views - __init__.py:1.1 configure.zcml:1.1

Eron Lloyd elloyd@lancaster.lib.pa.us
Mon, 28 Oct 2002 21:08:42 -0500


I guess I agree with Guido here, but can see some of the other points as =
well.=20
After 2.0, Zope began evolving into an army-knife anyways (database, Web=20
server, security framework, templating language, scriptable API, SQL=20
interface, search engine, WebDAV server, FTP server, DOM API, etc., etc.)=
, so=20
it's understandable that the desire to continue to extend this architectu=
re=20
is there. For instance, we now have a good XML-RPC interface. We could al=
so=20
build in a SOAP interface, and go even further and add a Jabber layer (wh=
ich=20
is basically an XML transport protocol like SOAP). Where do we stop? Why=20
should we stop? What about XML & Web services? How can I build a complex =
Web=20
application like MS SharePoint portal without messaging components? Just=20
because Jabber isn't as critical as SQL support doesn't mean it won't be=20
soon.

Zope is already having an indentity crisis, though (besides the Content=20
Management vs. Application Server). In a sense, we already *have* a gener=
ic=20
protocol framework in place, that handles HTTP, FTP, WebDAV, and friends.=
=20
When Zope 3 began, there was talk about scraping Medusa (& asyncore) in f=
avor=20
of a more robust solution; many mentioned Twisted, yet it was then told t=
hat=20
a similar server framework was being developed (Shane H.?), and that was=20
that. Do we even need an integrated Web server to begin with? We're told =
it=20
makes it easy to set up "out of the box". Well, I'd like to setup a Web m=
ail=20
system with IMAP & SMTP support then, plus iCalendar too!

I guess what I'm saying is that if we _do_ in fact want to see Zope grow =
10X,=20
then perhaps we do need to expose Zope as extremely extensible platform. =
I'd=20
still like to see Twisted become built into the Z3 core, to prepare for=20
future protocols and be equipped with a more robust network foundation in=
=20
general. I see that being key to being able to survive.

Cheers,

Eron

On Monday 28 October 2002 08:36 pm, Guido van Rossum wrote:
> > Whether someone specifically needs SMTP integrated as a protocol in
> > Zope 3, I would imagine would really be up to individual
> > requirements, however I would like to think it is possible and
> > fairly straight forward to do such a thing.  Being able to "plug in"
> > (probably not the correct term as plug-in is often misused ;-) new
> > protocols, or at least implement new protocols in Zope 3, in a
> > straight forward and well documented manner would have to be a big
> > attractor and feature of Zope
> > 3.
> >
> > I think we do need to have a protocol framework, and sure many
> > protocols aren't shipped with Zope 3, (ie Jabber support)
>
> I agree that it should be possible to plug in new protocols easily.
> But whence this desire to support "many protocols"?  Have you all been
> watching too much Twisted lately?  Protocols need to make sense for
> Zope.  Why would you want Zope to support Jabber?
>
> Twisted has different goals than Zope; it is a framework for writing
> arbitrary net applications.  Zope limits itself (mostly) to that
> subset of web applications that require mostly-read access to
> persistent objects, with added services like access control,
> authentication, multiple views, cataloging, and so on.
>
> --Guido van Rossum (home page: http://www.python.org/~guido/)
>
> _______________________________________________
> Zope3-dev mailing list
> Zope3-dev@zope.org
> http://lists.zope.org/mailman/listinfo/zope3-dev
> ---
> [This E-mail scanned for viruses by Declude Virus]