GSoC:2008 Collections 2.0 Todo list

bold are the larger tasks.

italic are the not-so-important tasks.

• Attributes
  • Allow general xmmsv_ts as attributes
    • Several values for field, value (except for MATCH), source. To be interpreted as a union.
• Chop history in logical chunks

• Documentation
  • Add documentation for most functions
  • Add documentation on the wiki
• Querying
  • Work on querying proposal
  • Implement querying proposal
• Matching
  • Define pattern-matching for some of the collations
  • Disable for those collations in the meantime.
• Prefix
  • Make prefix-matching work for collations.
  • Fix prefix for non-ascii characters like Å‚
  • Move idlist to the attributes.
• IdlistVirtual
  • Implement an index
• Checking
  • Set default value for type-attribute of idlist to "list"
  • Check for cyclic behavior in jumplist
  • Check source-attribute
  • Check source-preference-attribute
  • Make default values configureable
• Sourcerank
  • Add default sourcerank to more plugins
• Medialib
  • Change schema to use value ids, which then appear in MediaTokens, MediaValues and MediaBackend tables
• Abolish INTERSECTION_ORDERED
• Change attributes to LIMIT (to not use UINT_MAX)
• Fix FIXMEs
• Do TODOs
• Add optimizations to collquery
• Serialize as a true DAG

Pre February 10 2009

Notation: In bold high-priority stuff, in italic optional/low-priority stuff, in normal font normal-priority stuff.

• Finish this todo list

• Prefix does not work with non-ascii characters like Å‚ (in the NATCOLL-collation at least).

• Create a Collections 2.0 operator documentation page
• Work on the querying proposal
• Work more on the querying proposal

• Use strings in the medialib everywhere
• Implement NATCOLL
• Add prefix-matching
  • For BINARY collation
  • For NOCASE collation
  • For other collations (how is prefix defined for INTCOLL, NATCOLL when ending in numerals?)

• Abolish case-sensitive attribute
• sed "s|FULLORDER|COMPARE|g"

• Implement an "index" in the idlist virtual table

• Readd DAG-checking

• Implement an id-wise operator
• Implement the token-operator
• Implement a similar-operator (extra)

• Revisit the "is this id in the collection"-function

• Implement global source-preference
  • Change database
  • Use it for collection-queries
  • Use it for medialib_info-variant
  • Have a config-value system to change the priorities
  • Allow plugins to set default sourcerank values. (Create a function that sets the [pluginname].sourcerank config if it is an empty string (signifying that the user wants the default or never set a value.) or does not exist.)
  • Change the config names from "server.sourcerank", "plugin/[pluginname].sourcerank" and "client/[clientname].sourcerank" to something different.
• Implement per collection-node source preference

• Add default-properties when saving a collection. (still double-checked when querying, not for pshuffle/queue, as those are to be abolished)
• Make these properties configureable

• Change the uniqueness of the medialib table + change the xforms accordingly

• Allow all the funky querying (needs value_t!)
  • Abolish json-like string for requesting. Request using xmmsv_t instead.
  • Better error handling
  • implement query_medialist
  • implement query_clustered
    • Change syntax to avoid ambiguity
    • Allow distinct tuples
    • Allow collations for clustering diction
    • Allow ordering
  • Aggregates:
    • implode
      • only distinct values
    • min, max, avg
    • list
    • tuple
    • named-tuple?
    • dict
      • only distinct keys
  • Functions:
    • url-decode
    • source-name
    • source-rank
• Maybe use a dict for the collection-attributes?

• Add a specials namespace

• Add proper database updating (once all is stabilized) Done for now
• Add some optimizations to collquery.c

• Update collparser
• Do the FIXMEs
• Do the TODOs