Re: [Semediawiki-user] PROBLEM: imported pages not ack'ed until manual 'activation'

[Emanuele D'A. <ma...@gm...>]

On 7/11/07, agwilson <agr...@gm...> wrote:
>
>
> I've created an XML doc and imported some pages into MediaWiki v1.9.3.
> (I'm also using Semantic MediaWiki v0.7)
> The import XML file is clean.
> And the site accepts the imported pages just fine.
> But - when I go to <ask> for those pages... not one of them is returned
> from
> a query.... UNTIL I go to each page and save it.   PAINFUL!!!


I think I've started to understand why this is. This is not just related
to the import process, it also happens when changing a template
that uses Semantic tags: one needs to edit/save each page using
that template for the Semantic data to be refreshed in the database.

The problem lies in the fact that a page needs to be parsed at
least once for the semantic tags to be found by the system and
stored in Semantic MediaWiki's database tables. I suspect the
import process does not parse the page.

It sounds to me like an architectural limitation of MediaWiki
rather than SMW. Maybe there are work arounds though.

So, it seems that Semantic MediaWiki has a problem with imported pages like
> this in that they must be 'activated' before the system acknowledges them
> as
> 'active' pages that will be returned in an in-line query.


More than activated the semantic data stored in them must be read at
least once, something that happens only when they are edited/saved.
Which is a sensible thing because you wouldn't want this to happen,
say, every time the page is displayed. I guess it could be possible to
change the import process to acknowledge SMW tags. But that would
be a change in MediaWiki's code, not SMW's. Unless the SMW dev
team comes up with its own import tool.

Yaron had pointed me to SMW_refreshData.php, but I can't run that script
> since my site is hosted and they prevent manual script executions on the
> server.


Maybe this script could be modified to run from the browser,
ideally as a special page and ideally with some limiting mechanism,
so that if it isn't necessary not all pages, but only a subset is
refreshed. Alternatively, maybe a bot would do?

Hope it helps.

Ciao!

Manu