[Techtalk] directory handling with apache server

[John Clarke]

On Sun, Oct 26, 2003 at 04:48:48 -0800, Berenice Chong wrote:

> How does apache know that a requested url is a directory? Is it
> because of the trailing slash at the end?  The FAQ says trailing

no.  you can give it a directory url without a trailing slash and it'll
still work.  the system call stat() can be used to find out whether a
give path is a file or directory (or symlink or socket or device or
fifo).  see the stat man page for more detail.

> slashes are needed for relative URLs to work properly, but I don't
> understand why.

that's within a document, and the slash is needed for the browser to
build the correct url.  if you type in the url:

    http;//kirriwa.net/john

the server (apache) knows that /john is a directory, but the browser
doesn't.  a relative url within /john/index.html, e.g.:

    <p><a href="photos/index.html">Photos</a></p>

has to be converted into an absolute url by the browser.  without a
trailing slash, it will strip "john" and add "photos/index.html",
giving "http://kirriwa.net/photos/index.html" which is incorrect, but
the browser doesn't know that because it has no idea that "/john" is a
directory.

fortunately, apache sends a redirection to the browser when "/john" is
fetched:

    [johnc at dropbear ~]$ telnet kirriwa.net 80
    Trying 203.63.21.216...
    Connected to kirriwa.net.
    Escape character is '^]'.
    GET http://kirriwa.net/john HTTP/1.0

    HTTP/1.0 301 Moved Permanently
    Date: Sun, 26 Oct 2003 22:44:52 GMT
    Server: Apache
    Location: http://kirriwa.net/john/

    [rest of reply snipped]

the browser then fetches the new url, this time with a trailing slash,
and can then build any relative urls correctly because it now knows that
"/john" is a directory.

hope this helps.


cheers,

john
-- 
whois !JC774-AU at whois.aunic.net
GPG key id: 0xD59C360F
http://kirriwa.net/john/