Closed dhull closed 7 years ago
Thanks David,
These look good.
I suspect the timer is a remnant of when I was first building the websocket hijacker. There's definitely no reason to keep it there.
You also may be right about the apply_mask dialyzer warning. I honestly can't say why I added that specific header clause. It's been so long since I wrote it, and I didn't comment on it like I should have. I'll try taking it out and see what happens with the websocket tests.
My gut tells me that when I was building it (and timing it), I was probably passing in binary strings for testing if it worked, and using that clause to convert the binaries to integer, so it's likely also a remnant from testing.
I'll give this a few tests and make sure it's all good, but I'm pretty confident this is fine :)
Thanks for the PR!
Merged. Thanks!
Oh, and by the way: I removed the apply_mask header after testing that its removal was unaffected by the autobahn test suite, and I removed the timer.
This pull request contains a number of small fixes for simple_bridge 2.0.1 revealed by dialyzer, in order of controversy.
The first two commits fix genuine bugs that dialyzer uncovered: an incorrect conversion, and a function call with the arguments in the wrong order.
The third commit converts a
now()
call toos:timestamp()
. Dialyzer warns thatnow
is deprecated, and theos:timestamp
call is equivalent (and faster).The fourth commit removes a case clause that dialyzer says is unnecessary. I think dialyzer is correct, but even if it isn't the only difference is a slightly different error is returned from
simple_bridge:inner_make/2
.The fifth commit silences another dialyzer warning, where is says that a function head is never matched. I believe that dialyzer is right, but rather than removing the code, I added a call from a dummy function to silence the warning.
Feel free to take as many or as few of these commits as you want, and to modify them as you see fit.