On Thu, Jul 9, 2009 at 1:14 AM, Dmitriy Sintsov<questpc(a)rambler.ru> wrote:
that. Synchronizing back is a very difficult task, and you
find yourself in deep trouble very soon. If you
don't do proper
replication conflict resolution, you'll have either junk on your side
on the Wikipedia side. In the later case,
you'll probably get blocked
rather soon, in the other case your users will get frustrated because
their edits don't get through.
At one wiki site (not a Wikipedia) an improved version of
Special:Ancientpages is used to manually synchrolize between localhost
and webhost clone wiki sites. This version of Special:Ancientpages
allows to select another exportable namespaces besides NS_MAIN and
optionally submits these to Special:Export. By default, the
functionality of Special:Export is very limited (at least it was last
time I was cheking it).
A better approach probably would be introducing an --date option to
The important problem though is, that during XML import not all hooks
are called and extension-specific data is not saved/restored correctly
:-( I think it's major problem that these hooks doesn't allow to pass
their own XML tags into the dumps.
Wikitech-l mailing list
Tags like <ref> which do not appear in core are dumped. You just have
to have the extension installed for them to do anything after you've
imported the content. If you don't have Cite and ParserFunctions (at a
minimum) installed, don't expect much of Enwiki to work after import.
That being said: I've found that having Cite and/or ParserFunctions activated
*while* importing to slow down the process (and occasionally cause it to
halt). It's better to import and then activate the extensions.