Add non-const char* overload to LookupByKey

[jtdavis777]

Fixes #8557

[jtdavis777]

the "better" way to fix this might be to SFINAE the StringType based template overload, but that seemed more ugly and this functions just fine.

[aardappel]

Yes, this doesn't seem a great solution since the problem is the overly generic template version that will just take any type.

Given that the users of these functions are mostly generated code, I wonder if we can't simply improve how this is called in generated code instead. Do they even need to be overloaded? What string types can this possibly be called with to warrant the overly generic template?

How does the original reporter call this to cause an error? Are they calling this function directly?

[dbaileychess]

Why can we not just make this the only version? I believe const char * passed into a char * will work.

[jtdavis777]

Thats a great suggestion. Will implement.

[aardappel]

const char * passed to char * generally won't work?

[dbaileychess]

Sorry, I meant to write it the other way around.

[jtdavis777]

I knew what you meant :) it just depends on if the argument match considers it close enough.

[jtdavis777]

oh, no, this is the root of the issue. the original version is a const char* only -- the issue is that a const char* maps onto the templated version template<typename StringType> int KeyCompareWithValue(const StringType&)

So the two options are either

1. providing an explicit overload for char*, or
2. adding SFINAE conditions to the template deduction (which is a bit gross in C++11)

[dbaileychess]

Also add a unit test for this. Find out where we do a call to KeyCompareWithValue in one of our tests and pass in both a const char* and char*.

[jtdavis777]

test added.