User talk:Tub/Sandbox

A Note about technical decisions

There are three ways to go about a project such as this:

1. Calculate the results on the Wiki-Server, using the technical means already provided
2. Calculate the results on the Wiki-Server, using newly developed technical means (i.e. extensions)
3. Calculate the results somewhere else and import them into the Wiki when finished

I've choosen the third because the first two aren't feasible.

• MediaWiki syntax is good for writing articles, but awful for anything that resembles programming logic. It lacks an editor with syntax-highlighting, auto-completion etc, and if something doesn't work, it doesn't provide any error messages at all, making developing with it slow and painful.
• DPL is very limited, it's not even w:Turing complete (on purpose). It's also SLOW, since it has to handle all the mediawiki-overhead on each query. Even if the required things were doable using DPL (they aren't), they'd put a lot of load on the GWW servers. Remember that development usually involves making lots of tiny changes and re-testing the results, and you'll have a ton of expensive queries running on GWW's live servers.
• The wiki's articles are too inconsistent for nice DPL queries, e.g. try fetching all knockdown-causing skills from the Knock down article. Even filtering by sections (which DPL can do) won't help you, since the interesting skills and the uninteresting skills are separated merely by a bolded sentence.
• Getting extensions installed, much more user-developed extensions, is impossible in a timeframe before GW2's release. It would have to be developed on a local GWW mirror, which won't be provided.

Really, the only sensible way is to do it in a more powerful language and outside of GWWs servers, i.e. get the articles off the wiki, parse the data locally in any scripting language you like (I'm using php) and output text files for later inclusion. Tub 00:16, 3 December 2011 (UTC)

Ok, those are all sensible reasons for doing this using offline programming. I don't think I know enough about SMW to know why gung ho to add it to Guild Wiki as a better option (and he did, iirc, do a lot of clean-up work to make some of the stuff work).

The best reason for requesting the extensions and working live is so that we know how to make this work for GW2W and, out of the gate, follow better syntax, organization, etc. That's not a sufficient reason IMO, but I hope it does get us thinking about how to better organize GW2W before we find ourselves in a similar situation. – Tennessee Ernie Ford (TEF) 01:39, 3 December 2011 (UTC)

Preparing for GW2W is a valid motive for work on GWW, but it's totally not the goal of this project. :) While it's easy to get my support for any suggestions resulting in consistency and cleanlyness, this project here is just about data mining and statistics.

It'd benefit from clean and consistent articles, but it'd be FAR more work to clean up the wiki (including the required discussions, template changes and everything) than it is to just wrangle the existing data in php. Tub 02:35, 3 December 2011 (UTC)

/agree. Data mining is a worth goal by itself. But I'd hate to lose the opportunity to learn from this so GW2W data is mostly clean during the game's first year or three. – Tennessee Ernie Ford (TEF) 02:43, 3 December 2011 (UTC)