|
/stopcasting does what you might expect. On execution it does two things really: it sends a message to the server to stop casting whatever you are casting at the time and it releases the client from locking you out from starting a new cast.
In practice this means if your spell is one-times latency away from completion, the server does nothing of interest while the client allows you to begin a new spell. Since even a latency of 250ms is a significant hit while chain-casting (I think of that as nearly a loss of 10% in dps on fireball spam) one can, with good reflexes produce a nice dps boost. Some situations lead to a two-times latency loss as well, which is of course more significant.
The danger and the downside is that if you hit your little macro (/stopcasting /cast Fireball (Rank omg) for example) too early, you'll lose an almost complete cast. On the converse, if you are too conservative then you might even be slower than just spamming without a macro and eating the latency.
In my opinion, it is all bullshit to be honest. There should be a one-spell deep queue and there should have been one since the original beta. I've said as much for that long and even if the queue were only .5 seconds deep many latency related issues for mages would be moot. Oh well though, my crusading days are done.
EDIT: I should note that there is another odd use for it. /stopcasting occurring after an action that doesn't invoke a GCD kludges in the ability to chain things that otherwise cough up errrors. So in trinket macros and the like, inserting /stopcasting between actions allows one to one-press macros in the present system.
On a historical note, Fastcast used to be the preferred method for dealing with predicted latency but the removal of conditional statements in the API killed that off forever. I feel it was a shame really as I always thought the solution was pretty elegant but I do understand why the conditionals needed to go away.
|