Wikipedia:Software Phase IV

出典: フリー百科事典『ウィキペディア（Wikipedia）』

過去の文書：この文書の内容は古く、現在のMediaWikiソフトウエアについての参考にはなりません。代わりにm:Help:Contents#For MediaWiki hackersにある文書を参照してください。

The actual real software is called MediaWiki (formerly called Wikipedia:Software Phase III). There are also other lists of potential features, see m:Paper Wikipedia, m:Internet think tank(s), m:Wikipedia4, m:T1wiki. Specific features are discussed in m:Metalingo, m:machine-friendly wiki interface, and m:simple ideology of Wikitax. These should all be integrated with the list below, and files describing features be named more consistently.

このソフトウェアの真の名前はメディアウィキです（以前はWikipedia:Software Phase IIIと呼ばれました）。将来的に実装される可能性のあるリストはm:Paper Wikipedia、m:Internet think tank(s)、m:Wikipedia4、m:T1wikiを見てください。特定の用途の機能はm:Metalingo、m:machine-friendly wiki interface、m:simple ideology of Wikitaxで議論がされています。これらはいずれ下にあるリストに統合されるべきで、一貫した命名法で機能を記述したものをつづるべきです。

There are a number of ideas floating around for further improvements to MediaWiki (note: the notion of Phase IV is now replaced by the idea of incremental improvements to MediaWiki) (see also m:T1wiki and m:Wikipedia4 and m:simple ideology of Wikitax for more to be integrated here).

メディアウィキにはいつか改良したいもののアイデアが渦巻いています。（注意、Phase IVという考え方は今ではメディアウィキの暫時的な改良と言ふものに置換っています。m:T1wikiやm:Wikipedia4、m:simple ideology of Wikitaxをご覽ください。）

Stability:

achieve stability before adding any more features?
finish porting all wikis to Unicode character set with UTF-8 encoding? (this work is under way -- is anyone tracking it?)

Correctness:

Formally define Wiki-syntax "m:Wikitax"
Generate valid HTML/XML under all circumstances

Performance:

a more efficient parser (using formal Wiki-syntax)?
better caching of rendered page content?
parallelize architecture?
rewriting SQL queries to be more efficient (under way)
decouple SQL queries from PHP code?
ensure only standards-compliant SQL, to allow portability between different database implementations
splitting page server and database (done!)
supporting multiple page servers (easy?)
supporting database clustering (how hard??)

Integration and decoupling: pushing the boundaries around:

support all languages with new software?
integrate the user database across languages?
decouple page design, layout and skins from main engine

Tuning and monitoring:

to be written

Scalability:

scalability to millions of articles and millions of hits per day
parallelize architecture?
controlling algorithmic complexity is vital: use indices, avoid O(N) or worse operations
avoid lock contention
full support for transactions: no "lost edits"
- move away from MySQL to Postgres or something else?
- using a cluster of page servers (easy)?
- using clustered database software (hard!)?
- supporting multiple server farms (hard, and difficult to administer)
need to eliminate the recent changes bottleneck that limits edit rate?
- multiple categories for articles? -- is this a good idea, or a bad idea?