Problem with balancer

Issue #79 resolved
Justin Schwartz
created an issue

"Changing to that frequency would make the teams too uneven." When it's baloney using a custom balancer. Every player's metric is 1 yet cannot enter on a team where the max matrix is 6 and only 5 players exist.

Comments (9)

  1. Justin Schwartz reporter

    Wow this issue's been up for this long?

    Here's the summary for people out of the loop: basically freqman was rewritten for 1.5.0 but that rewrite was again re-engineered to use even more advisers and that has been up in HZ and HS for a few months. I'm not certain it's 100% stable but I guess it's at a point where it can be merged into the core--I think this particular bug is fixed. I will need to clean it up though.

  2. Justin Schwartz reporter

    this is a revamp of the new freqman code that dr brain wrote. I wrote it in ACE originally (cleaned it up a bit here, but there are still some noticeable artifacts) basically I rewrote the thing almost from scratch (though I'm pretty sure there are parts that are verbatim from brain's code and/or fm_normal before it) the goal was to make the code a little cleaner, fix some fundamental problems with the workflow, fix a memory leak/failure-to-LLRemove that could not be isolated, and add some additional functionality

    the interface should be binary-compatible with modules written for the original freqman rewrite. cb_preshipfreqchange should not be used by custom modules. it was solely to allow freqman to preempt other modules in getting a shipchange callback due to load order snafu

    this module ended up replacing the default branch's freqman for what Hockey Zone and Hyperspace uses for over a year now, it seems fairly stable but it seems to have a bug with the balancer code, where if two opposing teams are uneven, favoring the lower frequency, and the balancer settings are normal, it will still put players on the lower frequency if it's not full. however this code should repair the other balancer issue from the other freqman which was a memory leak (#79)

    addresses issue 79

    680016ddb3dc

  3. Log in to comment