Discovered as part of #386 and might be a regression from the ems reworks over the past couple weeks, but not totally sure. Could also just be a long lingered bug because of high(er) precision floating point sizes with crypto markets..?
Seems like there's a variety of "issues":
[x] you can enter more $value of pos then the limit allows (likely a rounding error discrepancy that's caused by assuming int / discrete sizes instead of high precision float?)
for example you can fill 1k fiat worth of some dest asset but if there's like "a buck or 2" more worth of fiat you can get after fill / position gets reported back, you can still fill that miniscule amount
this seems worse for short positions in which case you can keep going forever without hitting a 0 size condition on the staged order?
[x] position info reloaded from a long running emsd seems to be constantly stale and relayed from the first trades relay load which we never seem to update on updates from the brokerd..
pretty sure this is due to new processing / symbol matching logic in the order mode startup? => ended up with a fix for this part in 5c8c5d8 on #386
Discovered as part of #386 and might be a regression from the ems reworks over the past couple weeks, but not totally sure. Could also just be a long lingered bug because of high(er) precision floating point sizes with crypto markets..?
Seems like there's a variety of "issues":
int
/ discrete sizes instead of high precisionfloat
?)emsd
seems to be constantly stale and relayed from the first trades relay load which we never seem to update on updates from thebrokerd
..