On 9/1/06, ThomasV <thomasV1@gmx.de> wrote:
Three months have elapsed since the second wave of subdomains creation
by Jens Frank, and little has been done to clean the main site.

Some subdomain admins do not even seem to be aware of the problem.
I do think, however, that we have a problem, when duplicate versions of
a text exist for several months...

In some cases, people imported texts on their subdomains and they did
not blank the original pages, nor did they provide a list of the pages they
imported. In some cases pages have been imported using the import
function, in other cases they were just copy-pasted. Sometimes the
imported pages have been renamed and the redirect deleted, which
makes them difficult to track.

Sometimes people only seem to have imported a fraction of the text
from the main site to their subdomain, whithout explaining the reasons
for their choice. It could be that the texts they left on the main site are
not eligible for wikisource, but how the hell am I supposed to know if I
do not speak the language?

So far I deleted pages on the main site after checking that they were
imported. But a lot of pages remain on the main site, that ough to
belong to subdomains, and I am not able to tell whether they have
been imported or not...

I would like to delete them, because their presence on the main
site is confusing : some occasional contributors do not realize
that a subdomain has been created, and they keep editing pages
on ws.org, and adding new pages there.

In two words : it is a mess.

In the future, I do not want another subdomain to be created without
there being someone in charge, who can supervise the import and
quickly delete pages from ws.org.

For the moment, I am asking what to do with the pages on ws.org
that sould be in a subdomain. Shall we delete them preventively,
keeping in mind that we can restore them later if we realize they
were not duplicates ?

It's very much a mess.  I think it should be a rule before a new language sub-domain is created that not only must there be a certain threshold from the Wikimedia community who support the creation of the subdomain (this will curb inactivity--I mean, how many of the recent languages created in the last batch are actually ACTIVE) all the pages MUST be categorized in anticipation of the page move.
This will make it easier for developers to mass move everything, or for local admins of the new wiki to use the import function and do it themselves.  It will also give us an idea of which pages we can go ahead and delete.  Wikisource.org is, putting it lightly, horridly cluttered because no one indicates which pages have been moved and so we keep them on that wiki for who knows how long.

I say that if we know that the pages ought to belong to an existing WS subdomain, we immediately delete them off the multi-WS.  By now, everybody should have pulled the texts off the multi-WS and into their appropriate language domain; we shouldn't have to wait forever for those pages to be moved.  Also, deleting them will give the languages which don't have their own domain some extra light (which is really the reason we're keeping ws.org as a functional wiki: for the languages which don't have -- or aren't likely to get -- their own subdomain; keeping old pages that should be moved does no good at all).