User:Gangleri/tests/google bugzilla:707

[edit] preliminary

Working with different variations of gogle links and using these in templates I found some very special kinds of parametrisation:

I have never read google specification It is long time since I used boolean edxpressions at http://www.leit.is/ another serch engine. Before we take advantage of this documented or undocumented features we should know about if they are stable. Who can look at this?
The "features" mentioned above are used at ro:Template talk:Lincuri/google/selectiv. I agree they are complex - some may say they are complicated ;-) I used similar at templates at eo: for a working list to find easily lots of pages from all wikipedia generating "odd" links see bugzilla:1512.

Despite the fact that en:Template talk:Lincuri/google/selectiv has the same code it generates different links because:

Whatever is used in [[XXXX:YYYY:ZZZZ]] is translated diffrently both for the special pages, for titles (articles, discussion pages etc). at a Latin-1 wiki as [[:en:]] and a UTF-8 type wiki as ro:.

Both templates work the same if all XXXX:YYYY:ZZZZ consists of ascii characters only.
First differences can be seen if characters as ÄÖÜäöüß ÀÁ etc. are contained XXXX:YYYY:ZZZZ.
Due to the fact that UTF-8 characters are stored as &amp#nnnn; on Latin-1 type wikis XXXX:YYYY:ZZZZ will translate to this and will generate wrong (in order not to call it useless) code for example google links. Other links generated by the parameter refering to titles will fail, but this is a known fact.

... for references to [[google:YYYY]] other wikis and also for parametrisation of templates?

As Brion mentioned at bugzilla:707#comment_text_6 "One might add a field to the interwiki table listing it, or perhaps some funky alternate replacement: currently the interwiki URLs contain $1 as a placeholder for the link with underscores; another placeholder for spaces perhaps?" there are some alternatives: "We" (hopefully there is an interet on this) should evaluate if $2 and other values could be used for a subset of the following requirements:

to use / to generate UTF-8 code (or equivalent translations) regardsless of what type the wiki is
translate spaces in "-" if this would be the the most suitable for google
alternatively to other characters if there is a requirement / a real need for it
to use lowercase letters only (this would allow the usage of the same function at eo: too)