Open rspeer opened 14 years ago
The database fix is taking an interminably long time, but meanwhile I have worked around this bug in the web API.
The RawAssertions table has too much depending on it to make the large change that this requires. It simply stalls forever.
Would it work to take the database down for a day, and have the only connection to the database be the one that fixes this bug?
A bug in csc.conceptnet.models allowed equivalent RawAssertions to exist, differing only in their creator. The underlying code has been fixed, but the API has not been updated to use the fix, because the database is now inconsistent with the code.
This affects all languages to some extent, but particularly Chinese, whose entire import was done after this bug was introduced. It also particularly affects people who add assertions through the API.