[libvirt] [PATCH 0/3] Remove website search, just use google

[Daniel P. Berrangé]

On Thu, Apr 04, 2019 at 10:25:52AM +0200, Andrea Bolognani wrote:
> On Wed, 2019-04-03 at 18:26 -0400, Cole Robinson wrote:
> > Andrea's ChangeLog patches reminder me that I attempted something
> > similarish once, and also tried to drop our custom website search:
> > 
> > http://www.redhat.com/archives/libvir-list/2016-May/msg01616.html
> 
> Glad to be an inspiration for you! :)
> 
> > danpb recommended I just make our search bar use google instead but
> > I never made the attempt. 3 years later here it is.
> > 
> > Patch #1 makes the switch but doesn't remove anything
> > 
> > Patch #2 drops index.py, the search indexer
> > 
> > Patch #3 drops search.php and all the references to it.
> >     libvirt.org/search.php is still a valid URL so this may break some
> >     links, even though at the moment the page prints virtually nothing
> >     except some boilerplate text. Maybe we can redirect it to the
> >     homepage or something, ideas welcome
> 
> I like the idea, and I like the diffstat even more O:-)
> 
> I would, however, suggest a slightly different implementation than
> what you have here, where using the search bar would still send you
> to https://libvirt.org/search.php?query=X, and that page would
> contain three links:
> 
>   * "Search libvirt.org"
>     https://google.com/search?q=site:libvirt.org+X
> 
>   * "Search the libvirt wiki"
>     https://wiki.libvirt.org/index.php?search=X
> 
>   * "Search the libvir-list mailing list archives"
>     https://google.com/search?q=site:redhat.com/archives/libvir-list+X
> 
> The advantages of this approach are that existing links pointing to
> search.php will keep working, people will be able to search the
> mailing list archives as well (this is a feature the current search
> implementation is supposed to have) and search the wiki without
> having to go through an extra hoop, and finally that extending the
> search interface to include more sources would become as simple as
> adding another generated link to the page.
> 
> What do you think?

Sounds reasonable, though I don't think it should be a search.php page.
It can be done with a plain search.html page and small amount of
javascript.

Avoiding php has the benefit that it works for locally installed docs
from the RPM.

I'm not fussed about broken links because search.php is long broken and
so any bookmark to it is largely useless already.

Regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|