Javier Ayres

> Looks good! I like the relative simplicity of it. I've got a few comments, and I think we could use some more tests in [`test_histcategory.py`](https://github.com/rcorre/qutebrowser/blob/2da752b47386a847cf74bc7e706d3db6d39cea8a/tests/unit/completion/test_histcategory.py#L200). Thanks. I'm definitely adding...

Hmm, some of the environments now fail because of the UPSERT syntax. It seems I checked all the sqlite versions while thinking "we need 3.8 for upsert", when actually we...

Usually there is a performance improvement when talking to a RDBMS over the network, but I wasn't sure about SQLite so I ran some tests: ```python >>> def insert_update(): ......

Since I was around I fixed the conflicts here. I've been using this patch since writing it by the way, and so far everything works perfectly. @treeshateorcs if you are...

You got it @treeshateorcs!

The upgrade to v2.0 forced me to refactor a good part of this work so I decided to squash all the old commits into one and re-apply it on top...

I was talking about re-creating the actual table (to add the new columns), not only re-creating its records. Unless we make it so `force_rebuild` always drops the table too... it...

@treeshateorcs I noticed now how it may break if you want to switch back to upstream. I guess the change is not backward compatible after all. If you still have...

Hey @The-Compiler, not in a rush here :sweat_smile:, but I'm looking forward to have this merged. What would you say is blocking this? Do we need #6039 first?

No worries! I was just curious of the situation for this one. If you are ready to review though, I should probably have a go at it too because it...