see a practical demostration at http://user-gemeinschaft.wikidot.com/guest:testeintrag

with the links

SeiteMitUmlautÄ
SeiteMitUmlautÜ
SeiteMitUmlautÖ

very telling!

My first idea was - is the example to do not wrong around?

What we could do is to "fix" broken pages by replacing each [[[SeiteMitUmlautÄ]]] with [[[SeiteMitUmlautÄ | seitemitumlauta]]] to keep the links consistent, but this requires at least a bit more work.

Should it be not like this ?:

What we could do is to "fix" broken pages by replacing each [[[SeiteMitUmlautÄ]]] with [[[seitemitumlauta | SeiteMitUmlautÄ]]] to keep the links consistent, but this requires at least a bit more work.

This is a repair of the "existing links" to the (old) existing pages "/umlauta" - where the page names are not changed. ( thinking of all attached files and so on..)
This repair you could do "now" before the new code is implemented .. ?
Because the new replacing link has the same target…

And than - after the implement of the new code .. any new links in the odd manner (..umlautä) to OLD Pages are wrong - because of linking to a new "/..umlautae" paqe - which does not (yet) exist…

I can see the inconsistency .. have we to live with such old error as long as wikidot exists?

The proposal from Franz to search in a fixed sequence is on a long term too exclusive to live with it… I believe.

So - I have no idea - what is better on a long term basis?

Helmut:

The proposal from Franz to search in a fixed sequence is on a long term too exclusive to live with it… I believe.

I felt it was saving all work of renaming pages. "Ä" and ä are very rare characters that could trigger that code, but of course its like natural evolution goes: you put one fix on the top of each bug….not very elegant codewriting, as is our DNA. My only programming experience is HyperTalk. so please forgive me….

BY the Way I see you are very close … Vienna!

I'm not sure how long the query took to determine that 30,000 pages are affected, but I would think a relatively simple search and replace query could be run on the database to resolve this. Of course, with total pages approaching 10 million, parsing through that much data could take some time.

That would fix internal links, but external links (from outside to wikidot pages) might be a problem.Maybe not a real itch.

N o, we ( or better Michal and his team) cannot change the old living page-names in the URL! The existing external Links will find the (not changed) existing URL's ..

All what is possible is to start now with a new code working as expected… no fix of old existing page-URL's!
The search and replace of old (simple) Links in the source of existing pages (30.000 !) would be change a code written by someone in a generall manner..

iI can imagne thtis is a dangerous task and has to be tested very well.
There are some other issues:
a) if someone writes after this a link in the new code-using manner to an old page there will be no warning that the existing page is not found becouse of new translation …only the page does not exist ( if someone clicks on the link).

b) We have to inform the admins of such pages of this change in any way of a global replace .

Not easy to change an existing environmwent !

I think

if someone writes after this a link in the new code-using manner to an old page there will be no warning that the existing page is not found becouse of new translation

Nevertheless, it seems to me that this is the best solution. It avoids having 30,000 pages suddenly unlinked, and it avoids messing up the Wikidot code-base with fudging code.

We have to inform the admins of such pages

A targeted email with links to the affected pages would be very useful. If you are going to make their life difficult then you should apologise and give them all the help that you can.

Naaaah, this would be confusing in the long term. We are thinking about it.

Yes, this makes sense..