I don't think the code reads right. The deadline_unix is null-able in database IIRC.

I don't think the code reads right. The deadline_unix is null-able in database IIRC.

1bafbb6

Related to #32339

Well, I regretted that I approved 32339 ..... and I think we shouldn't add more patches.

If you search "deadline_unix" in the code base, you can see there are more places using it for comparing and sorting.

So the correct fix should be introducing a very large magic unix timestamp for "no due date", but never use it as a real date/time struct

I think magic values are not good and should mostly be avoided ...

... new maintainers just get confused ...
... so the tend to be hatder to maintain
... as the name already implyes with magic, the meaning is not clear at first sight and you have to dig deep into why what appens for each of such values
... that should be documented, but we know about docs being ... just what they are now :/
... if set you need to migrate stuf if you need to change it

but it would make the db query shorter yes ... :/

🤔

I don't like magic numbers and seldom use them.

But for this case, if you can make sure:

• all "deadline_unix" related SQLs are written correctly (fix all other places)
• complex SQLs should be covered by tests
• future developers won't make such low-level mistakes (help them to fix all "order by deadline_unix" and "deadline_unix < ?" bugs)

Then just do it.