Mailing List Archive

What are the benefits you are hoping for?

I understand that pages in the proposed namespace will not show up in
search results any more, unless I specifically search in this
namespace. While this can be beneficial, it can also have negative
effects. If old pages are excluded users will often get seemingly
incomplete or even empty results, thinking there is nothing to be
found, or the internal search broken. Personally I rarely had issues
with outdated pages polluting my search results. More the contrary. If
old pages show up that's probably because I was searching for exactly
that information.

I hope we are not going to hide this namespace from external search
engines, as this would make this content largely inaccessible.

Another issue I see are broken links between pages. MediaWiki relies
on one mechanism to make sure links between moved pages still work:
redirects. These would need to stay behind. This can become a problem
when people start deleting the redirects, or using them for other
content.

Personally I'm really not sure if a namespace is the best possible
solution. mediawiki.org is more a continuum of content we collected
over the years. Pages are rarely up to date. It would probably be less
cumbersome to consider the existing main namespace an "archive" and
only mark pages that are known to be up to date.

Kind regards
Thiemo
_______________________________________________
Wikitech-l mailing list -- wikitech-l@lists.wikimedia.org
To unsubscribe send an email to wikitech-l-leave@lists.wikimedia.org
https://lists.wikimedia.org/postorius/lists/wikitech-l.lists.wikimedia.org/

Thanks for bringing this up! Some comments below.

On Wed, 2023-05-31 at 12:11 +0200, Derk-Jan Hartman wrote:
> I'm proposing we introduce a new namespace on mediawiki.org.
>
> The namespace name will be Archived (numerical id to be determined)
> and its purpose will be to hold pages like "Subversion" that have the
> template "historical" applied to it. These pages would move info that
> namespace and you would get Archived:Subversion,
> Archived:Manual:Small padlock icon, Archived:Help:Extension:WebFonts
> etc.
>
> This will give us a place outside of the main content namespaces to
> keep information about configurations, manuals, extensions and skins
> that we want to keep, but where it will no longer pollute our set of
> currently relevant information. The namespace will not be part of
> $wgContentNamespaces and $wgNamespacesToBeSearchedDefault namespaces.
> Hopefully this will allow for a more searchable and better
> functioning mediawiki.org when it comes to documentation for new
> users, while preserving history in the spirit of wikis.
>
> I've discussed this with various people at various points in time who
> all seemed to think this was a good idea, but we've never really had
> an open discussion about it that could result into action.

It is sometimes hard for me to find relevant information. Thus a while
ago I asked for deboosting search results on mediawiki.org that include
"Template:Historical": https://phabricator.wikimedia.org/T274082

Still I'd love to have a way to exclude such pages by default.
As time passes by the percentage of obsolete pages will increase. Thus
relevant information become harder to find. I consider this a problem.

FYI, wikitech.wikimedia.org does have an "Obsolete" namespace (id 110;
talk id: 111). Though I do not know how it came to existence and how
regularly it is used (as moving pages requires special permissions).

IMHO a related issue is that authors and editors need to be more aware
of time. I've come across many pages stating "currently" or "next
July". Without going to the page history, readers cannot realize that
these strings were added years ago and that the page *is* outdated.

It may also make sense to show more visibly when a wiki page was last
edited and/or that there is a bigger chance that it might be outdated.
See https://phabricator.wikimedia.org/T247987 for a vague idea.

Cheers,
andre
--
Andre Klapper (he/him) | Bugwrangler / Developer Advocate
https://blogs.gnome.org/aklapper/
_______________________________________________
Wikitech-l mailing list -- wikitech-l@lists.wikimedia.org
To unsubscribe send an email to wikitech-l-leave@lists.wikimedia.org
https://lists.wikimedia.org/postorius/lists/wikitech-l.lists.wikimedia.org/

I'm sure others have better thoughts than I have on the proposed solution,
but I agree this is a problem; I would really appreciate better ways to
sort away things which are not only outdated but only historically relevant
because it's about a tool or extension which is no longer installed or
maintained – as opposed to something still in use, but where the
information was last updated three years ago.

Best,


*Johan Jönsson*Manager, Product Ambassadors
Wikimedia Foundation <https://wikimediafoundation.org/>


On Wed, May 31, 2023 at 12:12?PM Derk-Jan Hartman <
d.j.hartman+wmf_ml@gmail.com> wrote:

> I'm proposing we introduce a new namespace on mediawiki.org.
>
> The namespace name will be Archived (numerical id to be determined) and
> its purpose will be to hold pages like "Subversion" that have the template
> "historical" applied to it. These pages would move info that namespace and
> you would get Archived:Subversion, Archived:Manual:Small padlock icon,
> Archived:Help:Extension:WebFonts etc.
>
> This will give us a place outside of the main content namespaces to keep
> information about configurations, manuals, extensions and skins that we
> want to keep, but where it will no longer pollute our set of currently
> relevant information. The namespace will not be part of
> $wgContentNamespaces and $wgNamespacesToBeSearchedDefault namespaces.
> Hopefully this will allow for a more searchable and better functioning
> mediawiki.org when it comes to documentation for new users, while
> preserving history in the spirit of wikis.
>
> I've discussed this with various people at various points in time who all
> seemed to think this was a good idea, but we've never really had an open
> discussion about it that could result into action.
> _______________________________________________
> Wikitech-l mailing list -- wikitech-l@lists.wikimedia.org
> To unsubscribe send an email to wikitech-l-leave@lists.wikimedia.org
> https://lists.wikimedia.org/postorius/lists/wikitech-l.lists.wikimedia.org/

On Wed, May 31, 2023 at 3:37?PM Thiemo Kreuz <thiemo.kreuz@wikimedia.de>
wrote:

> Personally I'm really not sure if a namespace is the best possible
> solution. mediawiki.org is more a continuum of content we collected
> over the years. Pages are rarely up to date. It would probably be less
> cumbersome to consider the existing main namespace an "archive" and
> only mark pages that are known to be up to date.
>

I'm also on the skeptical side. I think whether something is relevant is
rarely clear-cut. I think most pages where the information is in some sense
old, and cannot be fixed just by updating the page, fall into one of these
categories:

- The page documents a thing. The thing still exists, but isn't
important anymore. Maybe it's not used by Wikimedia, but used by others.
- The page documents a project. The project is finished. The page is not
outdated (since the project finished, there is no need to change its
description) but not that relevant to present-day capabilities.

In both of these situations, it's not that obvious whether the reader would
want the page included in the search results or not.

What I'm hoping for is that we will have information that is relevant,
show up better than information that is no longer relevant, especially
for users of mediawiki.org that are newer and less versed in the history,
terminology and mechanics of MediaWiki.
At the same time we would still have all information available to anyone
more experienced and willing to look deeper, with a simple click of
"All" in the search page.

DJ


On Wed, May 31, 2023 at 3:37?PM Thiemo Kreuz <thiemo.kreuz@wikimedia.de>
wrote:

> What are the benefits you are hoping for?
>
> I understand that pages in the proposed namespace will not show up in
> search results any more, unless I specifically search in this
> namespace. While this can be beneficial, it can also have negative
> effects. If old pages are excluded users will often get seemingly
> incomplete or even empty results, thinking there is nothing to be
> found, or the internal search broken. Personally I rarely had issues
> with outdated pages polluting my search results. More the contrary. If
> old pages show up that's probably because I was searching for exactly
> that information.
>
> I hope we are not going to hide this namespace from external search
> engines, as this would make this content largely inaccessible.
>
> Another issue I see are broken links between pages. MediaWiki relies
> on one mechanism to make sure links between moved pages still work:
> redirects. These would need to stay behind. This can become a problem
> when people start deleting the redirects, or using them for other
> content.
>
> Personally I'm really not sure if a namespace is the best possible
> solution. mediawiki.org is more a continuum of content we collected
> over the years. Pages are rarely up to date. It would probably be less
> cumbersome to consider the existing main namespace an "archive" and
> only mark pages that are known to be up to date.
>
> Kind regards
> Thiemo
> _______________________________________________
> Wikitech-l mailing list -- wikitech-l@lists.wikimedia.org
> To unsubscribe send an email to wikitech-l-leave@lists.wikimedia.org
> https://lists.wikimedia.org/postorius/lists/wikitech-l.lists.wikimedia.org/
>

I like the idea of using
https://www.mediawiki.org/wiki/MediaWiki:Cirrussearch-boost-templates more
extensively, for both positive and negative boosts.
E.g. compare that to Wikitech's more extensive usage:
https://wikitech.wikimedia.org/wiki/MediaWiki:Cirrussearch-boost-templates
(Docs: https://www.mediawiki.org/wiki/Help:CirrusSearch#Boost-templates )

Perhaps we just need to apply {{historical}} more widely?
Or maybe we need a new Header-template that's distinct from {{historical}}
and {{outdated}} and {{fixme}} ?
The current list of related and widely-used templates is under
https://www.mediawiki.org/wiki/Documentation/Style_guide/templates#Fixme

I think tagging pages is slightly preferable to moving pages, because it's
less complicated to undo, and still leaves content accessible to a basic
search (just further down the results), and to different use-cases (per
Gerg?'s comment), and doesn't get technically semi-blocked for translated
pages with >500 subpages.
However, it is more labor-intensive for pages with many English sub-pages
(i.e. needing to tag each individually), although I'm not sure how to
evaluate that issue, without preemptively seeing the massive list of the
pages that ought to be deprecated!

I'd recommend: Tag a few dozen pages, and simultaneously create a listing
of pages where tagging doesn't seem sufficient, so that we can discuss more
specific examples/buckets of examples.
E.g. you mentioned [[Help:Extension:WebFonts]] as a candidate for Archival,
but perhaps that could be handled in the same way as [[Extension:WebFonts]]
itself was (blank and tag it with {{archived extension}} )? Plus I just
checked, and that page has over 500 subpages (from translations) so it
would require dev assistance to move!

The other semi-related problem this all makes me think of, is the cleanup
needed in https://www.mediawiki.org/wiki/Special:PageTranslation to
"discourage" any outdated pages. I'm not sure how extensive that problem or
overlap is, but ideally anyone doing cleanup of one problem could also keep
in mind the other.

Hope that helps!
Quiddity.