Friday, September 14, 2007

How to Pretend Zope Doesn't Exist

I'm doing my best to ignore that I just read Well-kept secrets of Zope, which lists all sorts of things I'm interested in seeing developed and how Zope did it. Zope is our Simpson's to South Park's Python community. All of this ORM wild fire and debates about unicode support and web frameworks seem really petty when someone reminds us that Zope, which most of us never really look at, has been doing some of these things for over a ten years. That's practically a decade in software time!

Will I settle back into a comfortable hole in the ground, cover my years, and chant "There's no such thing as Zope" or am I going to bite some pride and surround myself with Zope? The first talk for a CharPy meeting is going to be on Zope, and I'm going to ask a lot of questions about why it gets ignored and how someone really familiar with Python can get into Zope so late into the game. Now, I am not planning to drop everything else. I still think CouchDB has some interesting ideas that I'm pretty sure aren't part of ZODB. I'm also a huge Twisted fan, where it applies. We've had some segmenting problems in Python with Zope and non-Zope, Twisted and non-Twisted and, more recently, Django and non-Django. We seem to be gaining a habit of frameworks that gain a really large following and are really known to those who don't use them, who continually ignore the developments. If Zope and Twisted played together better, both in code and community, would Django ever have even surfaced to fill the gap? Would Rails have been irrelevant?

We need to build some bridges, so who wants to shake hands?


Technorati Tags: ,

4 comments:

Florian said...

My first python only job was maintaining a zope application for which original maintainers already hat fled the field. I tried to wrestle the thing to a zombie living where it wouldn't missbehave at the soonest opportunity, and failed miserably.

After 6 months of struggling I was at the end of my rope. I kept my job. While wrestling zope a day I started rewriting what we needed by night with the help of turbogears. Then during 2 weeks of x-mas holidays I worked like possessed, and by the first weeks of january I could present a viable exit strategy for me and the company, which would end up with me having a job and the company having an application.

I blame my predecessors (classic), but I also blame zope to no small degree, because I believe that zope lured these people, then encouraged them to be that messy and horrid, and finally made it possible for them to hang on to their jobs much longer then ought to be.

I derive a simple teaching from this story:
1) before the 3rd to 7th personal complete rewrite, any solution you write to a problem domain sucks beyond relief.
2) There is no framework substitute for being an architect for nontrivial applications. Inversly, using a framework to get near the problem domain solution feels good, but dooms your code.

So what now?

Zope bridge Twisted bridge Django and the Age of MegaGigaFrameworks? and that's going to help with either of 1) or 2)? Who're you kidding?

/me *yawns*

Philipp von Weitershausen said...

Florian is right, though I blame Zope's predecessor (another classic, but true). Zope 2 has this awful way of luring people into a development scheme that's just wrong and completely backwards. Fortunately, Zope 3 (which is what the Well-kept Secrets article is about) isn't at all that way. Unfortunately, it has the same name as Zope 2. Today we know that was a mistake, but can't easily change it now.

We're certainly up for some bridge-building, Calvin. Using Twisted was just one step. We now fully support WSGI, and at EuroPython I demonstrated integrating Zope with PasteDeploy, something that everybody else seems to be using nowadays.

It's good and somewhat healthy to be looking over the fence once in a while, or even be inspired by others, just like we were with Grok (http://grok.zope.org) which takes the "convention over configuration" approach to Zope. That's why I think it's good for the overall community to have other frameworks with other approaches: they provides breeding grounds for new ideas. I think the diversity of Python web frameworks is a strength, not a weakness.

Florian said...

Philip, I spend quite a time contemplating what constitutes these "lures" in software.

It's a complex phenomenon, but a few things stick out. The probability of a lure rises with the complexity a piece of software. The lure probability also rises with how easy it makes it for people who know neither the problem domain nor the solution space to coax it into behaving a bit like the desired solution.

Zope3 goodness'n all, it _is_ still zope, a big mother of framework promising you to solve the most vexing of everyday web programming problems with a passing glance.

You'll forgive me if that still smells overwhelmingly like a lure to me?

Anonymous said...

Wow you are a brain surgeon. They have been doing it for 10 years? That's practically a decade. Can't put nothing past you!!!!

I write here about programming, how to program better, things I think are neat and are related to programming. I might write other things at my personal website.

I am happily employed by the excellent Caktus Group, located in beautiful and friendly Carrboro, NC, where I work with Python, Django, and Javascript.

Blog Archive