[Zope3-dev] Re: Zope X3 3.0.0 alpha 1 released

[Stephan Richter]

Bogdan,

On Thursday 20 May 2004 03:54, Bogdan M. Maryniuck wrote:
> On Wed, May 19, 2004 at 02:54:41PM -0400, Stephan Richter wrote:
> > It is more stable than Zope 2 ever was, since everything is unit tested.
>
> Well, session stuff in Z2 is also unit tested, but it still crashes very
> ugly.

Then it is not tested enough. Zope 2 was not built on top of a testing 
culture; Zope 3 is. We try very hard to test every aspect of the code.

> But I am not talking about stability and architecture (where I have a 
> questions also), but just only about ABILITY to FINISH it one day.

For filesystem Python programmers a first version is finished this summer with 
3.0.0 final hitting the wire. We are specifically not aiming at the TTW 
developer in this release and removed many packages that are aimed at this 
group in the releases. In fact, we even removed some packages from the 
distribution that were voted as "not polished" or "well tested", such as the 
workflow package.

> I remember Zope folk pushed a lot of PR of about "how wonderfull Z3 is" and
> "release will be at autumn of 2003". 

Where?

Yes, we planned to release it around this time, but there were some design 
flaws in some of the core components and it was a **community decision** that 
said we would rather wait a bit longer. Any delay in releases is due to me, 
since I am in charge of the releases! In fact, the reason we did not have 
another milestone release was also my decision.

> But nor Z3 is so much wonderfull: 
> no DTML (arrgghh!!)

That tells me you have not even looked at Zope 3, since DTML is there, since 
about 2 years. And I know that for a fact because I ported it.

> connection to RDBMS architecture seems like mostly the 
> same as Z2 (it's crappy in general)

Here you offend me directly as I have made the initial design decisions on the 
RDB connection code. The Zope 2 version is not at all crappy! It has very 
good concepts on which we have built on. Many long-time Zope 2 users were 
involved in the design. We have corrected the things that the Zope 2 version 
got wrong and replaced it with something better, like unified and 
standardized connection strings.

Just saying that it is "crappy" is not constructive at all and is just a flame 
bate. If you think it is so bad, why don't you improve it?

> neither ZMI 

Huh? We clearly have a ZMI. It is not polished, but that's a goal for later 
releases. Currently there are no UI people helping out and we do not want to 
make it pretty the wrong way.

> and stable first release.  

As I pointed out, a stable release will be here soon.

> :| Seems like Digicool have no money left to develop it or so...

It is **not** Zope Corporation that is developing Zope 3. It is the Zope 
**community**. At least 60% of the contributions are not from ZC. The reason 
Zope 3 is so late is that a large part of the community did not get involved. 
And trust me, there are plenty of non-coding or Zope3-specific tasks out 
there. It is **not** solely ZC's responsibility to develop Zope 3, but the 
people using Zope to make money, which includes you!

> The only I *heard* that Z3 has better architecture in general. But it is 
> good news for geeks, not for customers... :/ 

This is a flat out wrong statement. A more solid architecture for Zope is good 
news for any customer. I have other people comment on that in more detail...

> So the question is: where (really for sure) Z3 will have first stable
> oficial release with backward compatibility 

I don't know. It depends how much people will help out. However, Zope 2.8 and 
2.9 will contain more and more of the Zope 3 code for people to use in their 
Zope 2 projects.

> (I don't think e.g. Plone guys  will re-write from the scratch its twice 
> bigger monster than Zope)? 

You are totally mis- or uninformed. I have talked to Alan already about Plone 
3, which will be a rewrite. There are already plans emerging to develop Plone 
3.

> > 1. Zope 3 will definitely not die. Too many people have invested too much
> > in it. Also, it is used in commercial projects already.
>
> Strange. Really strange. This could be two things:
>   a) Projects are not so serious (can be down, might crash, no support
> etc). 

Of course these are not *huge* installations, but they are serious. Ask Steve, 
he did a project using Zope 3. Also ZC and Sidnei did use it already in 
critical projects. Do not make statements that are not true.

> b) If not a), than developers are young crazy and customers should 
> pray hot.

How come that some of the best programmers in the Zope community use it then?

> Because I am strongly against unfinished things, which is could be very
> nice in general, but yet are still deep Alpha or milestone. My customers
> could "kill" me if they get know I use something unstable for banking
> solutions.

So why don't you help finishing it? BTW, Zope 3 is currently acquiring the 
Common Criteria Security certification. So it is not as unfinished as you 
think. Also, I would not claim that Zope 3 is ready for everybody.

> Also I don't fully trust to unittest. I got so much
> unittested code which is works buggy in practice and long run (NOTE: those
> guys who made it really good pythoneers and really know how to unit test
> things). And you can not unittest everything, e.g. ZODB or RDBMS connection
> things. But yes, I agree that unit test is much better than none.

But unittests are not everything. There are also integration and functional 
tests, which we do as well. Before the final version is released I expect 
many more functional tests to be written.

In general I have to say you have no clue what you are talking about and you 
offend a lot of people by making dubious statements like this. I CCed the 
Zope3-Dev mailing list, so that other offended (or not) people have also a 
chance to respond.

If you want to criticize Zope or any other technology, then do so in a 
constructive way and point out specific issues and problems you see.

Regards,
Stephan
-- 
Stephan Richter
CBU Physics & Chemistry (B.S.) / Tufts Physics (Ph.D. student)
Web2k - Web Software Design, Development and Training