Re: HTTP 404 or 301 - Which ErrorDocument response

In article <Pine.LNX.4.62.0506281204560.19712@xxxxxxxxxxxxxxxxxxxx>,
Alan J. Flavell <flavell@xxxxxxxxxxxx> wrote:
>
> On Tue, 28 Jun 2005, James Taylor wrote:
> >
> > In article <8764w0xb5p.fsf@xxxxxxxxxxxxxxxxxx>,
> > Chris Morris <c.i.morris@xxxxxxxxxxxx> wrote:
> > >
> > > "301 Moved Permanently" seems a little inappropriate, since the
> > > content hasn't been moved, but removed.
> >
> > But at least visual clients, ie. real web browsers, would follow
> > the "Location:" header to the root page automatically.
>
> Yes, but typically that has no effect on any bookmarks which the
> browser has previously saved. (It *could* be done in theory,
> but I don't recall any browser which actually does it).

Astonishing! I think browsers jolly well ought at least offer to update
their bookmarks when they see permanent redirections. I find keeping
my bookmarks updated manually more of a chore than it's worth.

> So if you take this approach, you risk having stale bookmarks out
> there without their owners realising that anything has changed.

Fair point. Perhaps transparency is not desirable in this case.

> > Also, indexing bots would realise the URL was no longer valid and
> > would update their databases (equivalent to a 404).
>
> Yes, but if you display a "gone away" page with a status 4xx (I still
> think Chris's proposal of 410 is the most appropriate choice of 4xx
> for this), they'd do much the same.

I agree that 410 is the best response. It just means I'll have to
modify my 404 script to distinguish between pages that have been
removed and pages that were never there. I didn't want to have to
keep a database of removed pages. Maybe I can fudge it by looking
at the format of the URL, which in this case is fairly specific.

> If the gone-away page also contains a link to another URL, I'd
> surmise that most indexing bots would include it in their scan

Really? I would expect the content of 4xx responses to be ignored.
Is there any way we can be sure?

> > I suppose I could combine a 410 response with a Refresh header.
> > What do you think of that idea?
>
> This is potentially a WAI problem, although if you set a
> sufficiently long delay

You mean slow readers won't be able to read the page before it
switches over to the next page? Of course, screen-reader software
reads at its own pace and might get cut short by a refresh so I can
believe this would be a problem. On the other hand, the title
and heading of the page are clear about the page not being there,
so maybe it's a good thing to redirect quickly so as not to bore
people with the page when it has little practical significance.

In Konqueror you can turn off HTTP refreshes altogether. I would
imagine that anyone with a serious reading difficulty would have
an equivalent feature in their software. Such software might
even follow the refresh URL *after* reading the page. Sadly, I have
no knowledge of what specialist browsing software is available.
In fact, I feel unqualified to judge the real issues affecting the
handicapped, especially as those handicaps take so many forms.

> and warn about it on your 410 response page I guess it's
> not too bad.

The kind of refreshes I hate are the ones that trigger after a
long period of time when you're half way through reading a long
article. Displaying two lines of text and refreshing after say
10 seconds doesn't seem to me to be very problematic.
What's your view?

> You would of course want to repeat the target URL in a link on
> the response page, so that an impatient user can go there as
> soon as they are ready.

In my case the error page contains a brief explanation of the problem
and a navbar full of closely related links. Surely an impatient
visitor would either click on the navbar or on the back button.

> For sure you shouldn't (ever) use zero-delay refresh - this not only
> violates WAI requirements, but also in some browsers risks trapping
> the user into a situation where their browser's "Back" function
> becomes unusable.

Agreed. I hate that.

> hope this helps.

Yes, Alan, you've been very helpful - as usual I might add. :-)

--
James Taylor, London, UK PGP key: 3FBE1BF9
To protect against spam, the address in the "From:" header is not valid.
In any case, you should reply to the group so that everyone can benefit.
If you must send me a private email, use james at oakseed demon co uk.

.