I looked at the Compiler User Guide, and it also implies that `-d:release` is
sufficient to turn on compiler optimizations. As do your comments above.
Weird...
Here's my environment:
* MacBook Pro, macOS 10.15.5
* Xcode 11.5; `clang --version` reports "Apple clang version 11.0.3
(clang-1103.0.32.62)"
* I used to have Nim 1.2 installed from HomeBrew; this morning I uninstalled
that, installed choosenim, and installed 1.2.2.
I'm building a Nimble package that I created a few days ago, which imports
asynchttpserver and the 'news' WebSocket library. I'm using `nimble build` to
compile. My `nim.cfg` file contains:
--path:src
--outdir:bin
--hints:off
Run
I'll use binary size as a rough indicator of optimization, for now.
* Default (debug): binary size is 752K
* With `-d:release`: 701K
* With `-d:release --opt:speed`: 608K
* With `-d:release --opt:size`: 528K
* With `-d:release --opt:size --stackTrace:off --lineTrace:off
--assertions:off --checks:off`: 273K
* With `-d:danger`: 752K
* With `-d:danger --opt:speed`: 647K
* With `-d: danger --opt:size`: 570K
* With `-d: danger --opt:size --stackTrace:off --lineTrace:off
--assertions:off --checks:off`: 307K
Conclusions: It definitely appears `-d:release` is not turning on
optimizations. And it's weird that `-d:danger` results in a larger binary; it
should be the other way around since it disables generating runtime checks,
right?
