omake a little slower

[LADSoft]

i did some profiling with omake; one branch was the master branch and the others is a current test branch which has the same code as the master, except the version of omake is before the latest rework by pr.

I basically ran the 'build' script as is, and stopped it right before it started the third compile of occ (after the library got built the second time).

On the master branch the build took about 19 minutes.

On the other branch the build took about 17 minutes.

That is approximately a ten percent difference in build times...

it might be good to look at this at some point.

[chuggafan]

I'm going to bet you it's partially because of my logging (more specifically, the + I left around erroneously adding instead of commas, causing extra evaluations of string functions for no reason).
I know nominally the logging because of the "log level" check should be cheap, but a quick VTune run seems to indicate that some time is spent actually doing string.append.
A further check also points to Parser::ParseLine, more specifically, the chain of "else ifs" that construct and deconstruct a load of std::strings and the construction/destruction penalties.
Perhaps a dynamic dispatch setup where we use std::map and fnv1a again to side-step constructors at the cost of a little more time for a call instruction? Might be worthwhile, areas such as that are the original reason I added that code as I recognized string constructors as a major time consumer.

[LADSoft]

im going to close this since you made pr#1115