So when we (in Stan 2.x days) talked about 'parameters {real y;} model
{y ~ normal(0,1);}' as "model code" (which typically was associated with
a model name like anon_model) we should instead (in Stan 3) refer to
this as "program code" associated with a "program name"?
Correct. Any real code is a program. I was very sloppy
with this originally, but Andrew has reformed me :-)
For example,
the (Python) signature for the StanProgram constructor (née StanModel)
should be StanProgram(program_code, program_name). If so, is the goal
eventually to have this reflected in the C++ API as well (model_name
occurs in 13 files under the stan tree)?
Yes. Feel free to open an issue and assign it to me or
just go in and clean it up yourself in the C++ code (on
a branch named after an issue, of course).
Issue by ariddell Friday Jan 29, 2016 at 02:00 GMT Originally opened as https://github.com/stan-dev/stan/issues/1759
From the mailing list:
Related protostan issue sakrejda/protostan#32