Lighttpd

If no other handler declares that it will handle the request, then the filesystem handler checks if the file exists in the document root. If it is a directory, then lighttpd redirects the request to http://example.com/foo/bar/ with a trailing slash. On the subsequent request (from the redirect), again if no other handler is configured to handle the request, then mod_dirlisting will handle the request, as long as mod_dirlisting is activated.

What is handling the request in your case? If it is something like mod_fastcgi, mod_scgi, or mod_proxy, then those can be configured to handle the request prior to the filesystem handler.

.

Note: it would be more straighforward and more in line with lighttpd and other web servers if you simply handled the request as http://example.com/foo/bar/ with the trailing slash, rather than attempting to do it without.

What is handling the request in your case?

I'm using the filesystem handler, with index-file.names = ("index").

Note: it would be more straighforward and more in line with lighttpd and other web servers if you simply handled the request as http://example.com/foo/bar/ with the trailing slash, rather than attempting to do it without.

That's true, but it doesn't seem future-proof. If /var/www/foo/bar/ is a directory at the moment, that works fine, but if I later decide it doesn't need a whole directory and I make /var/www/foo/bar an HTML file, then visiting http://example.com/foo/bar/ instead of http://example.com/foo/bar will break any relative links on the page. In short, the trailing slash exposes an implementation detail that can cause trouble if the implementation changes.

You are incorrect and seem to not understand PATH_INFO. Why don't you try going to http://example.com/foo/bar/baz/and/more and see what happens for your http://example.com/foo/bar/ site?

Anyway, please read the documentation for mod_rewrite

url.rewrite-once = ( "^/foo/bar$" => "/foo/bar/" )

You are incorrect

I don't know why you don't believe me, so here's a demonstration. If you go to http://arfer.net/x/dt/foo , the page's relative link to an image works. If you go to http://arfer.net/x/dt/foo/ , the link is broken (because the browser tries to get http://arfer.net/x/dt/foo/img.png instead of the intended http://arfer.net/x/dt/img.png ).

Anyway, please read the documentation for mod_rewrite

url.rewrite-once = ( "^/foo/bar$" => "/foo/bar/" )

I've read this page, and I'm sorry if I'm being dense, but I don't see how I could use mod_rewrite to remove the slashes. A rewrite only changes how the server treats the request, without changing what ends up in the user's address bar, right? What I'd like, to be clear, is to have http://example.com/foo/bar/ redirect to http://example.com/foo/bar.

Your web page is broken. One solution is to use root-anchored paths instead of relative paths. Another is to do something other than what you're doing on the server side. You are using mod_indexfile, which operates inside a directory, and you want it not to work that way. Your "denseness" is that you're asking for something that works exactly as designed (an index file inside a directory) to not be inside a directory. That's not how mod_indexfile works. Instead, you might try using your fubar links on your page, and send the link to http://example.com/foo/bar, which is internally rewritten (using the rule I gave above) to /foo/bar/, which will then be handled by mod_indexfile, if you have configured mod_indexfile.

Another is to do something other than what you're doing on the server side. You are using mod_indexfile, which operates inside a directory, and you want it not to work that way. Your "denseness" is that you're asking for something that works exactly as designed (an index file inside a directory) to not be inside a directory.

It's fine for it to be inside a directory on the real filesystem; the trouble is that this feature of the filesystem gets reflected in the URL. But if there's a way to accomplish this without mod_indexfile, then by all means, I'm all ears.

Instead, you might try using your fubar links on your page, and send the link to http://example.com/foo/bar, which is internally rewritten (using the rule I gave above) to /foo/bar/, which will then be handled by mod_indexfile, if you have configured mod_indexfile.

I understand, but then how can I also get http://example.com/foo/bar/ to redirect to http://example.com/foo/bar? Trying to do this seems to produce a redirect loop, perhaps because of mod_indexfile.

I think I've figured it out:

url.redirect = ("^/foo/bar/$" => "/foo/bar")
url.rewrite-once += ("^/foo/bar$" => "/foo/bar/index")

Rewriting directly to the index file instead of to /foo/bar/, and anchoring the redirect regex with $, avoids triggering the redirect rule after the rewrite.