Ulf Wiger
Ulf Wiger
I see no reason to restrict group names to atom. (Judging by the code, perhaps I did once, but in such cases as these, a good memory is unpardonable, as...
The README.md file is actually generated from doc/overview.edoc, using `edown`.
Sorry for not digging into this yet, but I've been a bit swamped. One thing to check when this happens is whether the locks themselves are consistent (e.g. by checking...
Thanks, that's interesting (and sorry for missing your reply for so long). So A and C are unaware of each other. I will see if I can figure out what...
Not sure if the above solves anything, but it closes a possible loophole, and adds test cases. I have not yet been able to reproduce the problem.
FYI, I've been running your test program. The thing that seems to happen, at least some of the time, is that a node detects that it's no more the leader,...
Before pushing, I let the `locks-test` program run continuously for 10 minutes, and observed no hickups or pauses.
Thanks for being so tenacious! :) I will try to find time to take a look at that again. At least the window seems to have been significantly reduced, albeit...
> The locks app addresses all these asynchronous communications disbalance doing the job pretty well, but the locks_leader rolled it a bit back... Could it be possible to use internal...
A possibility, albeit clumsy, might be to have a timeout on the safe_loop, where the `locks_leader` polls the agent for new info. Perhaps this could e.g. be limited to a...