AndreasMadsen
commented
10 years ago Without knowing anything about co or mongoose I would say that they do some sort of quick pooling. It is particular apparent with co there runs a recursive setImmediate loop, until some condition is meet. Sometimes this condition is meet immediately in other cases it takes a few ticks.
I think the same is the case with mongoose. It seams to manage some connection pool, the size of this pool will obviouse vary. Sometimes there is an available socket and sometimes a socket needs to be opened.
tl;dr You flow may be the same, but the modules you use can have state dependent conditions, in each case different async or sync routes may be seen.
I'd like to point out: it skips kind of a lot of intermediate middlewares in the stack. Is it intentionally?
It shouldn't skip anything, except for:
- 2 lines in each tick there only exists because I hook intro some node monitor API
- all node internal lines because you use clarify.
My guess is that some of this middleware is sync or lazy async, thus co executes it as A(); B(); C(); and you won't see any stack lines from those modules.
iliakan
I'm developing and didn't have the time to leave it hanging yet, but got another thing.
Please consider the video here: http://ilyakantor.ru/screen/2014-09-13_1705.swf
In short, I refresh the page and the stack trace changes. But the control flow is the same. Is it normal?
I'd like to point out: it skips kind of a lot of intermediate middlewares in the stack. Is it intentionally?
Here's the last trace:
P.S. That had both trace and clarify on and stack limit of 1000.