In certain stressful conditions (e.g. 10 links, no slowdown factor, moderately powerful computer), I see occasional missing Fragments at the end of a run when testing a 1-process and a 3-process system. (The 3-process system is RU, DF, and TRG.) These tests use the FakeCardReader.
I found that using a different order for Stop commands to the 3 processes resulted in fewer missing fragments.
In certain stressful conditions (e.g. 10 links, no slowdown factor, moderately powerful computer), I see occasional missing Fragments at the end of a run when testing a 1-process and a 3-process system. (The 3-process system is RU, DF, and TRG.) These tests use the FakeCardReader.
I found that using a different order for Stop commands to the 3 processes resulted in fewer missing fragments.