[Search-l] Need for a proper bugzilla
Balinny
balinny at gmail.com
Mon Jan 7 21:34:00 UTC 2008
We urgently need a proper bugzilla system to achieve bug reports.
Using a wiki page going public is unmanegeable. Specially if
jer will have to fix them all (is anyone else working on it?).
I see those hundreds of bug reports and really miss an option to close
as duplicate ;-)
Although i have to admit, making "Report a bug" a link to action=new
doesn't encourage to check if it has been reported before...
For instance, i would summarise
http://search.wikia.com/wiki/Prealpha_problems_with_character_encoding
as:
-Nutch doesn't support utf8 encoded input.
-alpha.search.wikia.com is utf8 encoded* so the query is sent in utf8 to
nutch.
-As nutch doesn't decode it, no results are found.
-JSON then reencodes the UTF8 query as entities, so à becomes Ã
and so on...
-The browser gets the entities and shows the utf8_encoded as plain text.
-User shouts and screams at the system, and ultimately goes to Google or
Yahoo.
Interestingly, as a workaround you can change the page codification to
another one.
Then the ajax query is not in utf-8, nutch understands it and the
results are provided.
*Or so says the meta tag. Web server doesn't provide charset.
Related to that, when the query is not in utf-8, the miniarticle
sections fails at the first no ascii-7 character (mediawiki would send a
redirect to the utf8
url, but that system seems not integrated on it).
Cheers
More information about the Search-l
mailing list