Potential race condition fix?

[Daniel Gutson]

I observed this change in @c4p-n1ck's fork:

master...c4p-n1ck:montydb:master#diff-822d717926e4514a846fad4aecc14e0cfcd9f30b60eca800667359921e6a5022R96

-       self.__conn = sqlite3.connect(db_file)
+      self.__conn = sqlite3.connect(db_file, check_same_thread=False)

Should that be brought here?

[David Lai]

Hmmm, just had a quick look.

I think what we could do is adding that check_same_thread as an config option at here:

montydb/montydb/storage/sqlite.py

Line 237 in 4653538

 def config(cls, journal_mode="WAL", **kwargs): 

Then the config should be used to initialize a SQLite storage engine instance here:

montydb/montydb/storage/sqlite.py

Lines 85 to 89 in 4653538

 class SQLiteKVEngine(object): 
  
     def __init__(self, db_pragmas): 
         self.__db_pragmas = db_pragmas 
         self.__conn = None 

At the moment, SQLite config is being used to keep database pargmas, but it could do more than that. And check_same_thread is a good fit.

[Daniel Gutson]

so it is not a race condition fix? Sorry about my ignorance.

[David Lai]

Don't be sorry! I am not familiar with this either.

But we certainly can add check_same_thread as a connection option though. Since it will affect the behavior of how a sqlite connection reacts to different thread, so we would want it to be optionable. ☺️

[David Lai]

Feature added in 2.5.2, closing this now. 😃

[Daniel Gutson]

u rock!