[Blacklight-development] LCSH facets; international chars

[Erik Hatcher]

On Mar 13, 2008, at 11:35 AM, Jonathan Rochkind wrote:
> Sorry for more email, today's another of my "explore blacklight and
> SOLR" days.

What a glorious day it is!

> It looks like you are putting LCSH's into facets without
> "uncoordinating" them. That is, there is a "topic_form_genre" facet  
> (and
> display field) rather than seperating out topic, form, and genre. Yes?
> Any background on this?  (Just didn't get to this; found it
> preferable?).

I created the initial facets based on conversations with UVa's  
(former) head of cataloging and just kinda winged it from there  
pulling out what looked good to me. Since my time on it, they may  
have adjusted some of these.  In general I think it was somewhat  
thought through, but also it was well known that these mappings were  
up for debate and rearrangement.

It's adjustable in scripts/virgo_marc_map.rb:

   :topic_form_genre_facet => [:'650a', :'650b', :'650x', :'655a']


>   Also, the Blacklight demo I have doesn't actually expose
> this "topic_form_genre_facet" in the facets pane, it's not  
> available for
> browsing. Again, any background?

I exposed all facets in my original incarnation, but since then it  
was deemed too much for the UI, so they scaled it back.  I've been  
chatting with Matt about it, and what I'd like to see is the facets  
are selectable by the user via some preferences.  I WANT THOSE FACETS  
BACK!  :)   I love browsing by the headings and see how intermingle  
with other facets.

> Also, I'm seeing some records in the same set that came with  
> blacklight
> that have international characters that are NOT displaying  
> properly. In
> general, do you have proper display of non-ascii chars working?   
> Record
> in example set that gives me improperly displayed records is title  
> "The
> art of the persian santur".

In general, yes, Blacklight worked fine with international characters  
back in my time.  I'm not sure if something broke that or if maybe  
something else is fishy in your environment that is causing it.  I  
had issues with non-ascii characters in my first indexing runs, but  
then discovered the MARC files themselves were mangled encoding-wise  
and I ran them through some yaz tools magic and that fixed things up  
fine.  Maybe the test MARC file needs the same conversion run on it?   
I just confirmed that I see the mangled stuff myself, and I'm  
guessing the MARC file needs some encoding TLC.  Matt?  Bess?

	Erik