Duplicated ware and building entries

[GunChleoc]

Some ware and building entries are duplicated in the wiki, e.g. here:

https://wl.widelands.org/encyclopedia/barbarians/wares/

I am assuming that the reason for this is the renaming of folders that was done for the tribes recently. So, the solution is probably to delete the "tribes" folder of the website's Widelands installation and get a fresh copy.

[SirVer (sirver)]

the problem is that the wiki is not updated anymore. The creation of these help texts is broken since ships were introduce.

[SirVer (sirver)]

aahh, some more detail: the data for the pages is created in python (in the widelands-website) project as it has to work with the websites database abstraction. This code parses the conf files and writes the data into the database. Unfortunately this broke with ships (which cannot be parsed by this python abstraction). With the one tribe change this will be even more broken and another way must be found to extract the information. I hope for the online help.

[GunChleoc (gunchleoc)]

Since we have this information on the in-game help now, I'm wondering whether it makes sense to remove this code from the website rather than spending a lot of time on parsing the new Lua files.

[kaputtnik (franku)]

In general i like the idea to have the encyclopedia on the website. So a player could retrieve informations without starting the game.

If we decide to held the encyclopedia on the website, we should make an export function like we have with the developers.The export function would be part of widelands and should store the buildings/wares/workers infos in a json file. This would be much easier to parse and there is a defined structure for this file. If something changes in the game (f.e. the adding of ships), this could be handled much better than parsing conf- or lua files.

[GunChleoc (gunchleoc)]

We actually have it the other way around for the developers: The source is in JSON, not in Lua. So, a Lua2JSON exporter will have to be programmed from scratch. Rather than do this one in Python, a C++ application might be the better option, because we already parse the input there.

[SirVer (sirver)]

The eternal true source of widelands data is the engine - it will know how to parse everything. The only right approach is therefore to provide a library or an exporter from the engine.

We already had a second parser of the engine's data format before (in python) which broke badly and quickly and had a different set of bugs to the parser in the engine. That is not a great place to be in.

Writing a c++ exporter binary that writes out JSON is the work of a day I think. This could then be run on the server on a nightly basis.