Performance improvements, documentation cleanup #21
Conversation
|
|
|
Travis tests seem broken but tests do complete successfully when I run them locally. |
traildb/traildb.py
Outdated
|
|
||
| if self.event_filter_obj: | ||
| if lib.tdb_cursor_set_event_filter(self.cursor, event_filter_obj.flt): | ||
| raise TrailDBError("cursor_set_event_filter failed") |
There was a problem hiding this comment.
This indentation looks wrong. How is it possible to reach this code immediately after raising an exception? @knutin
There was a problem hiding this comment.
It's probably just misindented. I'll pull that indentation back a bit.
traildb/traildb.py
Outdated
| if lib.tdb_cursor_set_event_filter(cursor, event_filter_obj.flt): | ||
| raise TrailDBError("cursor_set_event_filter failed") | ||
| else: | ||
| self.event_filter_obj = None |
There was a problem hiding this comment.
Indentation uses 3 spaces on this line
traildb/traildb.py
Outdated
| raise TrailDBError("Failed to initalize trail in cursor") | ||
|
|
||
| if self.event_filter_obj: | ||
| if lib.tdb_cursor_set_event_filter(self.cursor, event_filter_obj.flt): |
There was a problem hiding this comment.
event_filter_obj should be self.event_filter_obj
traildb/traildb.py
Outdated
| Apply a whitelist of the given uuids so that only | ||
| events with the given uuids are returned by the | ||
| cursor. (Empty trails are still returned for other uuids.) | ||
| Applies a whitelist UUIDs to TrailDB so that further calls to |
There was a problem hiding this comment.
I think we're missing an of in this sentence
traildb/traildb.py
Outdated
| Apply a blacklist of the given uuids so that | ||
| only events without the given uuids are returned by the | ||
| cursor. (Empty trails are still returned for the given uuids.) | ||
| Applies a blacklist UUIDs to TrailDB so that further calls to |
| The selected_uuids keyword argument can be used to select a subset | ||
| of trails to iterate over. Missing uuids are skipped over. | ||
| :param selected_uuids: If passed, only go through the UUIDs passed in | ||
| this argument. It should be an iterable that yields hex UUIDs. |
There was a problem hiding this comment.
Should we add an explicit check of each UUID as we iterate over it? (Or do we check already?) If performance is a concern there, then I guess any bad strings just won't be matched and a trail won't be returned for them. Maybe we need to add a comment that entries that aren't found are silently skipped and no error is raised then?
Fixed Conflicts: test/test.py traildb/traildb.py
|
Travis is failing to install libarchive-dev apparently |
Most important performance improvement in this PR is sharing of cursors in
TrailDB.trails()call. Cursors are relatively heavy so this can be a massive improvement in some workloads.Also cleaned up a whole bunch of documentation and added some sphinx configuration for that.