howtoaddevaluators.md

How to add evaluators.

1. Fork this repository.

2. Pick a 3 letter abbreviation for your evaluator. Use that abbreviation everywhere you see 'xxx' below.

3. Make a subdirectory for your evaluator. (Use a long name, not the 3 letter name). Change to it.

4. Create a get.sh. Make it executable.
   It must contain commands to fetch the evaluator from the web, and place it in a subdirectory named source. (Use git clone, wget,...) See existing directories for examples. Test it. Add get.sh to a commit.

5. Create a XXX_eval.h This file will map the specifics of your evaluator's implementation to the generic test interface.

• The file should #include "..\pet_utils.h
• The file must define the following 2 types: Card, EquivClass
• It must #define XXXHAND which contains the count of Card sized objects needed to represent a Hand
• It must contain these 6 functions/macros.
  • int XXX_init()
  • Card XXX_makecard(in)
  • void XXX_addcard(Hand,Card)
  • EquivClass XXX_evaluate(Hand)
  • int XXX_rank(EquivClass)
  • pet_rank XXX_ranktran(int)
    See the Interface document for more details.
    Add this file to the commit.

6. Edit PET_evaluator.h to add your code. This is the file that redirects calls to PET_* to your evaluator. Just copy and paste the 3 lines starting with elif TESTXXX, replacing XXX with your abbreviation.
   Add this file to the commit.

7. Edit the Makefile. Copy an existing entry as an example. It is important to change "-DTESTXXX=1" and "-oxxx" for your evaluator. Also add your abbreviation to the "all" line.
   Run make all and ensure your code builds cleanly. Run ./xxx to ensure it runs without errors. Add the makefile to the commit.

8. If there is a data file, Also add a Makefile line to create the file if needed, and to copy it to the root with name xxx_tables.dat. Make sure the XXX_init() function uses this file.

9. See How To Run Tests to compare your evaluator to the rest.