[ZWeb] Zope.org - take 3?

[Chris Withers]

Paul Winkler wrote:
> I wasn't going to respond to this thread, but hey, you insisted ;-)

yay :-)

>>4. Aim at building the site with as simply as possible, using no fancy 
>>new software, and sticking to the absolute bare minimum to make it work 
>>fast. If fanciness is needed, let it be in error reporting and site 
>>logging domain, rather than "cool new features".
> 
> dunno what you mean. Do you think zope.org has too many features?

No, I'm talking purely software here. I want to see zope.org on good, 
solid robust, SIMPLE software, so people can pitch in and help when they 
want. To put it another way, I want Zope.org to look exactly like it 
does now, work exactly like it does now, but be faster running and 
easier to maintain...

>>5. Have an SVN checkout of the instance home of the storage server and 
>>each of the app clients (the client would probably be branched off a 
>>common base, with only the config files being different) so that people 
>>can checkout the software on a local machine and develop stuff using the 
>>usual branch-and-merge model.
> 
> somehow i thought that something like this existed
> already, but I don't recall hte specifics.

It does, but I remember running away when I heard it tried to do things 
like compile python, etc. The instances should be simple enough that 
they run on any Zope 2.7.4 and any Python 2.3.4, if anything more 
complex needs to be specified, the end result will be brittle and 
impossible to understand.

>>6. Enable all "system" software (eg: python, zope, cmf) to be easily 
>>upgraded as needed, so the latest bugfixes can be used as soon as possbible.
> 
> +1, but I'd be wary of upgrading too eagerly unless the intention
> is for zope.org to serve as a dogbowl.
> Consider e.g. the recently reported problems with AccessControl under 
> zope 2.7.4.

Yep, I'm not advocating going with the "latest and greatest", that's 
what got zope.org in this mess in the first place. However, being stuck 
on 2.damn.old and requiring MyWierdProduct 
0.1.some-beta-pre-release-rubbish and nevner being bale to move off them 
seems somewhat foolhardy. How many security exploits is Zope.org still 
vulnerable to? ;-)

>>7. (maybe) a functional test suite that actually tests all the 
>>functional aspects we're aiming to support. Does anyone have lots of 
>>experience producing functional test suites?
> 
> not me, but ZopeTestCase should help.

Functional != UnitTest.

> If resources allowed me to devote time to this,
> I would bump this up from "maybe" to "definitely" and do it first 
> before changing anything.

Well yes, but we're all in that boat, hence the maybe...

cheers,

Chris

-- 
Simplistix - Content Management, Zope & Python Consulting
            - http://www.simplistix.co.uk