$url->path

Gisle Aas (aas@a.sn.no)
Wed, 7 Feb 1996 16:40:33 +0100 

I am now mostly happy with the current URI::URL interface, but I have
a bad feeling about the behaviour of the path() and epath() methods
and how they are treating the leading slash.

You will find this comment in the code (_generic.pm)

    # 2.4.6
    #
    # RFC 1738 says: 
    #
    #     Note that the "/" between the host (or port) and the 
    #     url-path is NOT part of the url-path.
    #
    # however, RFC 1808, 2.4.6. says:
    #
    #    Even though the initial slash is not part of the URL path,
    #    the parser must remember whether or not it was present so 
    #    that later processes can differentiate between relative 
    #    and absolute paths.  Often this is done by simply storing
    #    he preceding slash along with the path.
    # 
    # so we'll store it in $self->{path}, and strip it when asked
    # for $self->path().  You can test examine if this "/" is
    # present by calling the $url->absolute_path method.

Can anybody explain the rationale behind stating that "/" is *NOT*
part of the path?

In 5b7 we store the leading "/" but we remove it for the value
returned by e?path(). When setting the path value through these
methods we just store the argument value as it is.

This means that:

   $u = new URI::URL 'file:/file/xxx/';
   $u->path($u->path);
   print $u;                # prints 'file:file/xxx/'

will loose the leading "/" in the path. It should not have to be like
this :-(

What can we do about it:

  1) make e?path() return the leading "/"

  2) store the leading "/" as a separate field in the URL
     ($url->{'absolute_path'}) and provide a separate method to get/set
     this value.  Also note that a path is always absolute if 'netloc'
     is present.

  3) make e?path() remove leading "/" from the argument (when setting
     a value)  How do we make a path absolute then?

  4) something else.

I am thinking about doing 1), but perhaps only for epath() since this
should not break old code.

We have the same problem with the path_components() interface
(introduced in 5b7). It is currently not possible to set an absolute
path, but we could define:

   $u->path_components("", "file", "xxx");  # to mean just that

Any advice?

--Gisle