## Not signed in

Want to take part in these discussions? Sign in if you have an account, or apply for one below

## Site Tag Cloud

Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.

• CommentRowNumber1.
• CommentAuthorUrs
• CommentTimeOct 22nd 2009
This comment is invalid XHTML+MathML+SVG; displaying source. <div> <p>Somebody sends me an email telling me he runs into the following problem when trying to read the nLab:</p> <blockquote> many of the links (internal) do not work, I get the message: "Published version of web 'nlab' is not available". </blockquote> <p>I have encountered this problem once before . At that time it just went away after a while. I think Andrew said it was something about the cache.</p> <p>Here it is kind of annoying, as that person encountering this problem is somebody I would dearly hear an opinon from about some page, and don't want to bother too much with technical problems.</p> </div>
• CommentRowNumber2.
• CommentAuthorUrs
• CommentTimeOct 22nd 2009

Ah, I think I tracked down how the error occurs:

it happens when

• first you follow a link from the nLab to a (my) personal web .. that ends up displaing the "published" link instead of the "show" link on the personal web

• then when the personal web points back to the nLab, one is sent to the "published" link version of the given nLab entry -- which doesn't exist!

• CommentRowNumber3.
• CommentAuthorUrs
• CommentTimeOct 22nd 2009

here is concretely how to stably reproduce the problem:

• then from there click on, say the link "(oo,1)-topos", which points back to the main nLab.

That procedure ends up requesting a URL of the type ".../published/..." instead of ".../show/..." and produces the problem.

• CommentRowNumber4.
• CommentAuthorAndrew Stacey
• CommentTimeOct 22nd 2009

I've looked at the code and this does look like a bug so I've referred it "upstream". If I'm right, it should be a simple fix in which case it'll probably get corrected in the next update. I'll keep you posted.

• CommentRowNumber5.
• CommentAuthorTobyBartels
• CommentTimeOct 22nd 2009

There are really two bugs here:

• The link from nlab/show to schreiber sends it to published (because it exists) even though show is available (and is better).
• The link from schreiber/published to nlab sends it to published even though it doesn't exist.

You could fix both at once by changing the default from published to show, but it would be prefectly reasonable for schreiber/published to link to nlab/published if that existed, so it's really two separate problems.

• CommentRowNumber6.
• CommentAuthorUrs
• CommentTimeOct 22nd 2009

Right. And I would think "show" should be default (meaning if it exists it should be preferred over "published").

• CommentRowNumber7.
• CommentAuthorTobyBartels
• CommentTimeOct 22nd 2009

But perhaps, if you come from published and both exist, you should go to published? Especially if both are available where you came from!

Either way you answer that, I would not consider it a bug. But going to published when it is does not exist is definitely a bug.

• CommentRowNumber8.
• CommentAuthorAndrew Stacey
• CommentTimeOct 23rd 2009
• (edited Oct 23rd 2009)

Damn, but it's hard to get this right. I think I've finally done it, though. We'll see ..

This change got into our branch half an hour ago (6am UTC). There may be the odd unexpired cache though so please report pages where there's still a problem.

The relevant code is:

action = web.published? && !(web == @web && mode != :publish) ? 'published' : 'show'


Where web is the target web, @web is the source web, and mode is the current view type (publish, show, s5, tex, etc).

• CommentRowNumber9.
• CommentAuthorTobyBartels
• CommentTimeOct 24th 2009

If I'm reading this correctly, it takes published as the default (possibly overridden when a web links to itself), but doesn't actually go there when it doesn't exist. And it never goes to s5, tex, etc.

That's probably not what I'd have done, but it'll work!

• CommentRowNumber10.
• CommentAuthorUrs
• CommentTimeOct 24th 2009

Yes, it will work. But I agree with Toby that this is not what seems logical. Because, isn't it true that if the "show" version exists, then the web is explicitly allowed to be edited by anyone. So it is probably desired and intended that everyone is explciitly invited to edit it. Therefore there should be no reason to send anyone to a view-only page on such a web which doesn't offer a link to edit it.

So I would think "show" should always have recedence over "published".

Or am I misunderstanding this?

• CommentRowNumber11.
• CommentAuthorUrs
• CommentTimeOct 24th 2009

To amplify: in that bug I reported, the big problem was that coming from my personal web back to the nLab produced an error. But also the previous step was not what I would intend: that people coming from the nLab to my personal web are sent just to "published" not to "show".

I am not requesting anyone (i.e. Andrew) to change anything, i am jut trying to see if we agree on how things should work, ideally.

• CommentRowNumber12.
• CommentAuthorMike Shulman
• CommentTimeOct 24th 2009

Why is it even possible for a web to be both publically available (i.e. non-password-protected) and have a "published" version? Under what circumstances would you want that?