Dependencies tracking

[apraga]

Hi,

A few ideas about tracking dependencies. I think the cleanest approach would be to have a separate table with a task and one of its dependencies. I'm not sure SQL can be constrained to avoid circular dependencies but that should be checked.
Also, the priority should take dependencies into account.
Just my two cents :)

[ad-si]

Yeah, this will definitely need a new table. Maybe something like:

CREATE TABLE relations (
  source_task TEXT NOT NULL,
  target_task TEXT NOT NULL,
  relation TEXT, -- blocking, duplicate, related, successor, …
  FOREIGN KEY (source_task) REFERENCES tasks(ulid) ON DELETE CASCADE,
  FOREIGN KEY (target_task) REFERENCES tasks(ulid) ON DELETE CASCADE,
  PRIMARY KEY (source_task, target_task, relation)
) STRICT;

Checking circular dependencies is a good point. We'll have to be careful when writing any code that traverses relations over several steps.

More information: #14