Use classic score in total score processor

[bdach]

As per #134 (comment)

• Depends on Add GetDisplayScore() extension for SoloScoreInfo osu#24957
• Depends on Read scoring attributes from the new table #155 for mergeability
• Depends on Update classic scoring formula to closer match stable score V1 osu#24924 for adjusted test values to be correct (dependency can be removed if need be).
• Depends on nuget release of osu.Game

[smoogipoo]

Should we not be increasing ScoreStatisticsQueueProcessor.VERSION as a result of this? And shouldn't this support revert of both versions? Though I believe it is the first time VERSION is being used in this way, I can't imagine any other purpose for it.

Edit: Actually it already has a check for previousVersion >= 2, so it's not the first time this is being done.

[peppy]

Nah, I won't be running any kind of revert operation on already applied scores. Is what it is at this point.

[smoogipoo]

As long as you understand the implication is not only that you will not be doing it, but that if you ever decide to do it it will become extremely difficult to do and blindly doing it will cause those users to have negative total scores.

[bdach]

Edit: Actually it already has a check for previousVersion >= 2, so it's not the first time this is being done.

That's the typical "this processor did not exist on prior versions" check, isn't it? Slightly different.

As long as you understand the implication is not only that you will not be doing it, but that if you ever decide to do it it will become extremely difficult to do and blindly doing it will cause those users to have negative total scores.

I mean... if we're bringing this up as a concern, then any change to any scoring algorithm is potentially a concern. If the scoring algorithm silently changes on the game side and then the processor version gets incremented, and the processors begin to reprocess scores, everything that touches score is now potentially not reverting what it originally applied. For this to work in a 100% controlled manner you would need a score versioning scheme etc. Which I will do, if you think that's best, I'm just not sure that it is.

[smoogipoo]

It's exactly the same thing at the end of the day.

switch (previousVersion)
{
    case < 2:
        return; // Processor didn't exist before this version.
    case < 7:
        totalScore -= standardisedScore;
        break;
    default:
        totalScore -= classicScore;
        break;
}

Anywhoo... #159

[peppy]

With added context and IRL discussion, I think the above is a valid concern (should increase VERSION and add per-version revert logic).