On Wed, Jun 07, 2000 at 04:10:03PM +0100, Chris Withers wrote:
> Michael Bernstein wrote:
> > As I recall (from a WebObjects demonstartion I sat through a
> > long time ago), WO creates the most godawful mangled URLs
> > that I had ever seen. As a result, a WO site cannot be
> > spidered by search engines, or converted to a static copy
> > using wget.
> Zope's not too hot on this either what with index_html instead of
> index.html the problem of "is /something' a 'file' or a 'folder' from
> wget's point of view?"
Well, to be honest, that was the thing that got me interested in Bobo at the
first place: I had a few looks at WebObjects when it was hot stuff in the
NeXT scene, but Bobo's concept of mapping object containment directly to the
URL path seemed much more natural.
WebObjects doesn't try to make the path look readable. A WebObjects
application was just like a NEXTSTEP application, only that the GUI was
rendered in web browsers. WebObjects treated HTTP and URLs like a sort of
alternative communication protocol, just like X11 or DPS--it was quite a
slick concept for a quick start in E-Commerce to be honest.
Bobo applications seemed to be somewhere in the middle: They looked like
traditional static web servers, but had the potential of WebObjects.
> IIRC, it currently would download a zope folder object as a single file
> I've also been bitten by this in Wiki's since links aren't generated
> with absolute_url and so I sometimes get
> /MyWikiFolder/OneWikiPage/AnotherWikiPage which throws up a nasty
> __getitem__ exception (particularly in Netscape, which seems to like
> appending / onto the end of FrontPage in the URL box...)
Yep. Like the "map" link. I wished "Entire XYZWiki Contents" would always
link to the same URL in the root folder of the Wiki. Currently it links to
CurrentWikiPage/map. Not too friendly to offline browsing as well.
AFAICS, the problem is that traditionally in a static web server, if you try
to access a URL like 'http://host/xyz' (without trailing slash), the server
will return you an error '301 Moved permanently' and will point you to the
new location 'http://host/xyz/'. Zope instead (if 'xyz' is a folder and
there's a document 'xyz/index_html') immediately returns the rendered
index_html if you request 'http://host/xyz'.
Now whenever the 301 to the URL with a trailing slash happens, wget takes
this as a hint that this path must have been a directory, and saves the
result as index.html. With Zope, wget doesn't get this hint.
Reading this again, I wonder if this qualifies as a bug in Zope, or is there
a good reason for this behavior ?