this with addition of an index should dramatically reduce query processing time and increase performance.
Next step is probably to join all queries together and parse in Python afterwards.
TODO:
Start working on rendering again
Try to scale with a hundred hastags.
The number of mysql conenctions should decreqse now.
TODO:
Future thoughts to dramatically increase this : Use a database pool.
Threads would simply add new write/commit requests and only one entity actually accesses the database.
This would probably reduce the amount of problems of concurrent connections.
Adds unicode support in mysql database connexion.
Removes all try/catch in commits that were used to avoid SQLite limitations.
FIXME: There is still some problems with special characters handling in MySQL, but it doesnt cause any problem for the leaderboard printing so far.
It has to be fixed though.
TODO:
Add separate encoding parameters in basic.ini
ConfigParser is used and will replace data.py's current system.
Used config file is basic.ini, but might be upgraded later (using command line arguments)
Streamer on it's way to be fully compliant with new system.
Should be better now, while still far from error proof.
To solve most of the database issues, it might be way simpler to change to another engine.
New problems may arise when clients will start connecting through the web interface.
We'll see.
Activates autocommit and deactivates autoflush to sessions.
This should allow higher concurrency on SQLite.
If things work, I should remove my poor tentative to activate autocommit by myself
Puts all test stuf into tests repo.
Puts all crashes and log into log repo.
I should really start some testing some day :s .
And remove all FIXMEs.
=> This is next step as soon as I got my code running on python anywhere
It is possible to create the same hashtag several times in the databse. => Add existence checks
A bug should not be created twice any more in the trendy list.
FIXME:
Fix waas done in the plane with no internet connexion nor ressource.
Code is really bad and not pythonic :s.
Think about doing better.
database has to be cleaned from hashtags to remove duplicates now.
Running command.py generates a database problem. Apparently due to passing session/connection between threads.
Sessions have been separated in HashtagLogger to avoid session shaing between threads.
TODO:
Invetigate how many concurrent sessions can be created, and/or how to avoid this problem.
