Fall back to Latin and English section by section

[GoogleCodeExporter]

Currently, if a file in a given vernacular language is missing, the program 
falls back to the corresponding English file, and, failing that, to Latin (this 
is on the assumption that the Latin always exists, so we never have 
translations appearing where Latin ought to).

It would reduce duplication in the data files if instead this fall-back 
behaviour happened on a section-by-section basis: a file for a given office 
could contain only those sections that have been translated, but requests for 
non-existent sections would return the English or Latin. As well as helping the 
maintainability of incomplete translations, this would reduce the need for 
redundant sections which do not need translated, such as [Rank] sections and 
references to commons.

This could be implemented reasonably simply by loading vernacular languages 'on 
top' of the Latin and English hashes.

Original issue reported on code.google.com by igregord on 20 Nov 2011 at 12:21

[GoogleCodeExporter]

r477 implements this on the 'languages' branch, but it exposed some mismatches 
between languages in the data files. I'll fix these problems gradually, but it 
might be that the fall-back strategy will have to be revised somewhat (although 
I'm hopeful that this won't be necessary and that r477 will be sufficient).

Original comment by igregord on 6 Feb 2012 at 1:47

• Added labels: ****
• Removed labels: ****

[GoogleCodeExporter]

This issue was closed by revision r707.

Original comment by igregord on 17 Apr 2012 at 1:44

• Changed state: Fixed
• Added labels: ****
• Removed labels: ****