|  | Commit message (Collapse) | Author | Lines | 
|---|
|  | The fallback event should not produce a notification. | 
|  | Logs are now proper sentences ended with full stops. Exceptions are
logged with full tracebacks, and log level are revised to be more
sensible and consistent across the extension. | 
|  |  | 
|  |  | 
|  |  | 
|  | Co-authored-by: Joe Banks <[email protected]> | 
|  | No reason for this to be async. | 
|  | The default KeyError message from dict lookup is just the missing key.
In order to give more context in the log message, we raise our own. | 
|  | The fetch helpers will now raise when the request fails rather than
logging a warning and returning a fallback value.
This allows better error logging as the caller is able to log the
propagated exception while adding its own context.
Additionally, the caller in some cases no longer needs to check for
the None return and raise its own exception. | 
|  | Previously, the event description & duration strings were only stored
on event entry. In the case that the description or duration change
for an on-going event, the cached values wouldn't be updated.
After this commit, the cache is refreshed daily by the daemon. | 
|  |  | 
|  |  | 
|  | Also changed talentpool review cog to post there instead of mod-announcements | 
|  |  | 
|  | The user ID is much more accessible, and is usually what is used to obtain the nomination ID. | 
|  | If it's been over a day overdue for a review, don't reschedule it.
This is done in order to not fire reviews for all nominations which
are over 30 days old when the auto-reviewing feature is merged. | 
|  | This is necessary as otherwise the bot would try to review them every time it restarts | 
|  |  | 
|  | Uncached mentions render as 'invalid' users on mobile, and with the list now showing the user's name we can now just show the ID without many problems. | 
|  |  | 
|  | Co-authored-by: ToxicKidz <[email protected]> | 
|  |  | 
|  |  | 
|  | After previous changes, the docstring was no longer accurate.
See: 1d5625a2f47a1d4d050f9eb0eb7a18e7d6fe171b | 
|  |  | 
|  | The daemon will now perform a sync iteration immediately when started,
and then every UTC midnight.
Previously, it would only perform the initial iteration when started
for the first time, which is odd.
It is also believed that splitting the daemon's logic into three
separate functions is beneficial: before, loop, and main.
This commit makes log and doc adjustments where appropriate. | 
|  | Fixes BOT-XK | 
|  |  | 
|  | `ast.parse` raises a ValueError complaining that source code strings
cannot contain null bytes. It seems like they may accidentally get
pasted into Discord by users sometimes. | 
|  | Fresh stable release, just in time! | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  | This should never do anything, but it's better to be safe.
Values taken from Discord developer docs. | 
|  |  | 
|  |  | 
|  | Knowing which event failed would probably be quite useful. | 
|  | The notification is now sent conditionally depending on whether we're
entering a new event. This prevents sending a repeating notification
in the case of a manual resynchronisation.
A practical example of when this may trigger is when a staff member
temporarily applies custom assets & then uses the sync command to
reapply the current event. | 
|  | Discord.py doesn't await the return value. | 
|  | Now that the boolean flags are propagating from 'apply_asset', we can
present them to the user. | 
|  | The sync command will now be able to use present this information
to the invoking user.
This commit also prevents the cached banner & icon hash from being
overwritten in the case of asset upload failure. As a result, the
daemon will attempt to re-apply the assets the following day. | 
|  | It makes more sense for the init and the rotation to be separate
operations.
In a subsequent commit, the separation of responsibility will allow
the `rotate_icons` function to have a meaningful return value. | 
|  | Remove the mixed line endings pre-commit hook because it is obsolete.
Relying on git to handle line endings means contributors have more
flexibility with which line endings they want to use on check-out.
The settings in .gitattributes only impose which line endings will
be used upon check-in (LF), which should not impact local development;
git will still respect the core.eol and core.autocrlf settings. | 
|  |  | 
|  |  | 
|  |  |