Initial support for demo stats parsing backend (WIP) #609
Conversation
|
I think its ok to leave the migrations in here. Id like to get going on this fairly soon and should have some cycles free to help with this more. Is there something I can help with on the golang/frontend side perhaps? How stable do you think the data shape is from tf2_demostats is at this point? |
|
I have some WIP local changes to tf2_demostats for A) shots/hits tracking for more weapon types B) more medic stats and C) playtime-per-class and playtime-per-team. I also have limited cycles but I should be able to find time to send those PRs in the next 2 weeks.
I filed leighmacdonald/tf2_demostats#18 for the only structural change left. If you make a decision there I'll implement it. Otherwise I'm happy with the current output -- remaining work will at most add more fields to existing objects.
I'm totally paused on this branch -- please take it/discard it/iterate however you'd like. My only real opinion on the gbans side is that I think the DB schema should store stats per-demo with the tf2_demostats version that was used to analyze each demo. Even if we don't reference that version anywhere for the MVP, it'd be good in the future to be able to reanalyze demos and/or delete bad stats if we have a buggy tf2_demostats release. |
Sounds good, I agree about storing the raw output & versions makes sense for potential reconstruction later on, makes sense, at least for now. |
# Conflicts: # internal/match/match_service.go # internal/tests/match_test.go
2 participants
Still WIP
This is a rebased version of #592 with a bit more wiring and demo JSON format updates.
Not yet ready to land as we don't actually create matches yet and there's some more work to do on tf2_demostats first; will continue to update. We can pull any DB migrations out separately if you'd like.