Authors CurseForge

it's the droprate in percent multiplied by 100(?) iirc.

At least the dataminer is not fixed, there are lots of other items that are reported to be crafted by tradeskills now ( eg Leatherworking: item vs spell )

/€dit.....
I've run the dataminer on most of the tradeskill sets myself now, resulting in tons of changes.
Maybe you are using the cache functionality of the dataminer and it somehow gave you old data for some sets?

So r289 should now have really fixed it.

Yeah, I had the problem, too. At least since the time when the enchant API was converted to tradeskill API, it has always been that spell ids were stored as negative numbers to easily distinguish them from item ids though a couple of times positive spell ids slipped into the data.
For tradeskills, the only time spells and thus negative ids where used was when there was no item being produced, eg for enchanting or profession specific enchants like fur lining.

So this isn't a problem with Mendeleev handling the numbers incorrectly but with the dataminer storing the wrong data in the wrong format.

Wasn't the itemid used in order to be able to map items to tradeskills?

I don't think talents are in PT3.1 at all (besides those for druids and the mage's frost tree that give spells). So this isn't about separating talents into those that give spells and those that don't.
It's simply that wowhead doesn't list the first rank for spells learend via talents in their class spell list. Thus these would have to be mined from the talent list but there is no easy way to select exactly these talents.
That's why the dataminer has the ability to force-insert additional data into sets which is currently used for these talents but is only ste up for druids and the mage's frost tree.
So if you can provide me with a list of talent/spell ids from each talent tree that are rank 1 of some class spell, I can add them to the dataminer in order to get them into the classspell sets.

I'm currently in the process of redesigning some datasets (eg instance loot) and the dataminer so I don't have time to do it myself :/

You said it yourself:

Quote from Shadoweric

Database -> Spells -> Class Skills -> Paladins
This one won't have Divine Sacrifice.

And no one took the time to add all the talent spells to the dataminer by hand, yet. (only druid and frost mage are in I think).
Feel free to provide a list though (spell id and the earliest level to learn the spell/talent).

The class spells are automatically mined from wowhead. Unfortunately they suck... spells you learn from talents aren't included in their list...

Yeah, I'm basically waiting to see how wowhead/wowdb/... are going to handle that.
The idea to use the numeric constants for the difficulty makes sense, as does using 0 (or maybe 5) for 40 players.

A similar problem exists with the tier sets. I used to refere to them as x.10 and x.25 where the community just used x and x.5 :/ and now we got even three tiers for the sets. Though it seems like 10heroic doesn't provide the 25 sets so I could just name the third heroic. I want to stick to x.10/x.25 instead of x/x.5 in order to have the correct tree layers.

done :)

You could try trinkets crafted by jewelcrafting. There is no single list for them.

Well, there isn't much to maintain in Mendeleev, since it mostly depends on PT3.1 being up2date...

Well, the problem with the Arena sets is that they aren't that much season related anymore. OK, a new season adds a new set but since when the switch to season 5 happend there have also been two low-tier sets added. So what we currently see as arena sets isn't limited by the season buth rather by the current maxlevel introduced in WotLK. Although we can't be sure that a new addon will always increase the maxlevel, so we could also go with lvl80 instead of WotLK, as it is already done with the lvl60 sets.

I'm currently thinking about renaming "GearSet.PvP.Arena5.Tier x" into "GearSet.PvP.WotLK.Tier x". Season 6 has just provided one additional tier on top of the existing three, but naming the new sets GearSet.PvP.Arena6.Tier 4 would mean two changes in the key which feels somehow wrong to me. Any opinions on that? I think currently only Mendeleev is using these sets directly (ItemDB just iterates over all keys) so it wouldn't be hard to pass that change.

yeah, in theory, these numbers should be automaticaly increased when commiting to the svn. maybe the keywords have been lost when moving to the new svn structure and noone noticed?

Exactly. I put these files on ignore for packaging in order to keep people from using the files from the packages for commits. Since that would break some keywords and such.