[release of 1.0] Check for FALL maps to concepticon in all datasets

[LinguList]

Some datasets have a strange bias linking "fall" in the meaning of "autumn" to "fall = descend"! I suspect TLS, Bowern's data, and maybe some other datasets. Sun1991 as well. But it is very strange... So please all data maintainers, quickly check whether this is normal. Check for the auto-produced subgraph for "summer" for referenceof this error.

[xrotwang]

Here are all datasets that have AUTUMN at all:

$ grep 1276 */cldf/parameters.csv
cals/cldf/parameters.csv:autumn,autumn,1276,AUTUMN
ids/cldf/parameters.csv:14-770,autumn,1276,AUTUMN
northeuralex/cldf/parameters.csv:HerbstN,Herbst::N,1276,
suntb/cldf/parameters.csv:autumn,autumn,1276,AUTUMN
tls/cldf/parameters.csv:fall,fall,1276,AUTUMN,
wold/cldf/parameters.csv:14-77,the autumn/fall,1276,AUTUMN

[LinguList]

It looks like it's only TLS, as this should be linked to "fall" = descend (or not, @tresoldi ?). Strangely, I find this only in the app, not in the SQLITE, so the creation of the app still seems to be extremely buggy...

[LinguList]

Or I was just not paying attention to another step for the standalone app creation ?

[xrotwang]

Hm. Must be the graph creation, already, I'd say.

[xrotwang]

I'm seeing the same thing, only in the subgraph, though.

[LinguList]

I'm re-running the code for subgraph now. Maybe it's just a problem of syncing. The numbers are also wrong, showing some 1417 words linked for "fall" and "autumn".

[xrotwang]

No, it seems already db loading is buggy:

sqlite> select distinct f.dataset_ID from formtable as f, parametertable as p where f.parameter_id = p.id and p.concepticon_id = '1276';
lexibank-bowernpny
lexibank-ids
lexibank-northeuralex
lexibank-satterthwaitetb
lexibank-suntb
lexibank-tls
lexibank-tryonsolomon
lexibank-wold

[xrotwang]

datasets are loaded with forms for AUTUMN, even though the datasets doesn't have AUTUMN in its parameters.csv.

[xrotwang]

Hm. that's a bit scary. I'm investigating.

[xrotwang]

Ok, Not as scary, hopefully:

sqlite> select distinct f.dataset_ID from formtable as f, parametertable as p where f.parameter_id = p.id and f.dataset_id = p.dataset_id and p.concepticon_id = '1276';
lexibank-ids
lexibank-northeuralex
lexibank-suntb
lexibank-tls
lexibank-wold

So maybe this is just a buggy db query in pyclics, which doesn't take into account that we always have composite primary keys (id, dataset_id) in the db. (See @SimonGreenhill - it's not easy)

Will investigate after lunch.

[LinguList]

Okay, relieved this is NOT the data! I was getting scared...

[xrotwang]

The first paper retraction in LT! Yeah! :) Johann-Mattis List <notifications@github.com> schrieb am Do., 28. Juni 2018, 12:57:

…

Okay, relieved this is NOT the data! I was getting scared... — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#27 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AA1HKM5evfJCAgz3FdV0KRqzaVcbCK7yks5uBLadgaJpZM4U7GDY> .

[xrotwang]

All good again: 6c41d5d#diff-f98b9409af68c1f168b8712d59d026d8

[xrotwang]