httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From d...@ast.cam.ac.uk (David Robinson)
Subject Re: Current status...
Date Thu, 14 Dec 1995 15:46:00 GMT
Ben wrote:
>On the subject of 2), although trapping self-inclusion is a good idea, I
>wonder about the semantics. Seems to me that /a/b/some-ssi/extra/bits is
>morally equivalent to /a/b/some-ssi?extra/bits, and the relative should
>therefore be relative to /a/b/some-ssi. Or shouldn't it? I've heard the
>"breaking existing sites" argument, and I have sympathy - we should make it
>flaggable, as in "SupportBrokenExtraPathBehaviour on", but I do not approve of
>the concept of perpetuating broken behaviour any more than is absolutely
>necessary, and certainly not to the exclusion of correct behaviour. It seems
>obvious to me that a relative path means "relative to me", not "relative to
>whatever path happened to invoke me".

However, if you use 'relative to me' then relative paths parsed on the
server will resolve differently to relative paths resolved by the client.
i.e.
<!--#include virtual="../xxx" -->
and
<A HREF="../xxx">

might point to different objects.

It would be better to forbid extra data on the URL to a SSI document, but
we can't do that for compatibility reasons. However, this seems as good
a reason as any to drop SSI in favour of a newly designed server-parsed
implementation.

 David.

Mime
View raw message