cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jonathan Stimmel <>
Subject Re: Pipeline conditional model
Date Thu, 08 Jun 2000 21:57:46 GMT
Shoot, I forgot to catch up on this thread before posting to the
sitemap thread... Oh well, a little redundancy never hurt anyone =)

On Thu, Jun 01, 2000 at 07:25:18PM +0200, Stefano Mazzocchi wrote:

> Interesting enough, I played Sam's role in the creation of the JAMES
> sitemap, advocating that flat pipelines are better than tree-like ones
> (and implemented the same thing myself in Ant, even if that is kinda
> different).
> But I changed my mind.

As I stated in my other post, I believe the sitemap should be viewed
as a tree. I honestly hadn't given much thought to the conditional
model, however, and I appreciate the amount of thought you (and others)
have put into it.

> > 2) I readily concede that the <if/> approach scales UP better
> Agreed.
> > 3) I would argue that the "when=" approach scales DOWN better
> I'm still not sure of the impact on this over my girlfriend... I'd
> better test it soon.

Personally, I think "scaling down" is more important. A well planned
site should be relatively simple and won't require a complex conditional
model. Additionally, reusing the XSL conditional model makes for a
shallower learning curve, particularly for small sites where the
designer *is* the administrator.

I would expect that 95% (ok, maybe I'm being optimistic =) of sitemaps
will be layers of choices - a decision tree. As has been observed,
this would allow you to create AND relationships (by nesting decisions),
but unfortunately doesn't provide an OR relationship for the remaining
5%. Assuming the xsl-like model can pass the Girlfriend Test, should
we necessarily abandon it because of that remaining 5%? Combine that
model with named pipeline fragments (<resource>), and you can almost
acheive an OR relationship (it still exists as a branch, but you are
no longer required to copy and paste the actual process).

A question in my mind is, "how many layers of decisions would
a site need?". I personally can't imagine a site that would need
a tree more than 3 or four layers deep. Heck, would you really
want your server making a dozen tests per request? I'm pushing to
develop a cocoon infrastructure for a fairly complex network of
sites ( and 60+ related sites), and I can't
imagine using a sitemap more than a couple layers deep...

> Like I said I'm superimposed...
>  (+/-)1
> Oh gee... now its your turn to quantistically measure me to make my wave
> function collapse on a definitive state :)

Does this mean that if we all stare at you long enough, you'll come
to a decision? ;)

View raw message