Cache stats

[Thor77]

Suggested by this forum-post

[Thor77]

This is a complex topic. My current implementation plan (mostly written down here for myself, but maybe someone can suggest another approach):

• tsstats.log._parse_details should return a ChangeSet of actions to be applied on a Clients-object instead applying them directly
• the actual caching will take place in tsstats.log.parse_logs: the caching-status of a TimedLog-object will be checked before calling _parse_details (by comparing a hash saved in cache and the current file hash => new lines in a log file will lead to parsing the file again)
  • if cached: get ChangeSet from cache and apply it to the Clients-object
  • if not cached: call _parse_details returning ChangeSet and apply it to the Clients-object

[djmattyg007]

A potential idea for caching the data is to store it in an sqlite database. Python has first-class support for sqlite, and it's a simple reliable data structure that anyone can work with.

[Thor77]

Yeah, sqlite is an option. My parsing rewrite is (nearly) finished, I just have to fix some tests: parsing_refactor Event-objects should be easy to safe in a sqlite-database.

[djmattyg007]

Sounds good!