Posted by Alan J. Flavell on 12/25/67 11:26

On Sun, 11 Sep 2005, Dan wrote:

> WD10 wrote:
> > On Thu, 08 Sep 2005 18:20:01 -0700, Dan wrote:
> >
> > > On an Apache server it can be accomplished with a .htaccess file.
> >
> > Why is that better than the meta tag? Isn't an .htaccess file a last
> > resort in a situation because it increases work for the server?

Talk about this contributor missing the point! OK, from the point of view
of a *server administrator*, the use of a .htaccess file (although
convenient) might be seen as a "last resort" compared to the use of the
server configuration, considering the extra bit of work done by the
server.

On the other hand, for those who handle only a portion of a server
hierarchy, it would not be wise to give them unrestricted access to the
server configuration, whereas it's rather harmless to allow them to create
..htaccess files.

Furthermore, .htaccess files take immediate effect, whereas changes to the
configuration need a server reload (which you're hardly going to grant to
someone who only handles a portion of the server hierarchy, what?).

None of this has any relevance to sticking meta http-equiv into HTML
files, though. And how do you suppose you're going to influence the HTTP
server configuration for non-HTML files (images, plain text, whatever).

> A meta "http-equiv" tag is intended as a substitute for a real HTTP
> header, for cases where for whatever reason you are unable to send one.

Indeed.

Originally the idea was that the server might pick these up from the
source and use them for constructing real HTTP headers; however, that
seems to have been rapidly dropped, probably for the reason I just
mentioned - it's useless for non-HTML files, and, once you've developed
something for non-HTML files, then you might as well use the same thing
for HTML too.

Later, client agents got the idea of evaluating these http-equivs for
themself. But there are significant numbers of things that you simply
cannot do via http-equivs - those need real HTTP headers.

> It's more logically pure to send the right HTTP header in the first
> place than to substitute for it later.

Agreed.

The one exception I found was that MSIE (when I tried it) did not really
implement the HTTP "refresh" header - it only implemented the "meta
http-equiv". But this header isn't in the HTTP specification anyway -
it's rather amusing, in a way, that it was introduced as a meta HTTP
equivalent to an HTTP header that didn't exist. Oh well, that's another
topic, I suppose, really.

[Back to original message]