Vadim Gritsenko wrote, On 26/06/2003 14.38:
Jeff Turner wrote:
(moving to cocoon-dev..)
On Thu, Jun 26, 2003 at 08:43:12AM +0100, Upayavira wrote:
There are quite a lot of new features in the Cocoon CLI that Forrest
isn't using, for example the option to switch off mime-type checking,
and to only scan pages once (i.e. not using the link-view) to follow
links.
But I like the link-views! ;) It's one of those design elegancies that makes Cocoon unique. Adding a don't-crawl-these-links option to the new CLI may solve the same problem, but IMHO it's a hack in comparison.
IMHO it's the link-view that's a hack. How it's done now, it's completely mixed between external and internal needs.
I mean, what we have is an environment that to work has to have a certain view specified in the sitemap...
The argument against link-views is that it's slow: two requests instead
of one.
Actually for the original method it's three.
Isn't the correct solution to fix caching so that 99% of the
processing between foo.html and foo.html?cocoon-view=links is shared?
It's not so. The links are in the content view, and after that there is a lot of processing.
If caching worked properly, why would requesting the link view take much more time?
Because they do not share 99% of the same stuff.
Have you tried caching point pipeline; and/or pipeline with caching hints? IIUC, you need to hint sitemap to cache data of the pipeline fragment common to both normal processing and links view.
Thgis alleviates something, but it will never probably get down to the speed of the new method.
Please Jeff, take a moment and read all the thread that brought to the creation of the new method, it's fully explained and discussed in detail.
--
Nicola Ken Barozzi [EMAIL PROTECTED]
- verba volant, scripta manent -
(discussions get forgotten, just code remains)
---------------------------------------------------------------------