Brian Ward

Results 396 comments of Brian Ward

I just manually edited the included `bernoulli` example to have a second prior on theta and ran the following, with output: ``` In [2]: m = cmdstanpy.CmdStanModel(stan_file="./bernoulli.stan", stanc_options={"warn-pedantic":True}) 12:21:07 -...

Auto formatting in particular is not useful as a part of compilation, because you want the output of that separately. This is why `model.format()` triggers this, same for the methods...

Can you enable debug logging (https://mc-stan.org/cmdstanpy/users-guide/outputs.html#logging) and post the output that yields?

> error: invalid value 'precompiled-header' in '-x precompiled-header' This is not an error I've ever seen before. Could you try running `os.environ['PRECOMPILED_HEADERS'] = 'false'` and trying again with the debug...

Huh, that's a new one for me. Can you try changing the hpp file to use ` stan::promote_args_t` instead of `typename boost::math::tools::promote_args` as the return type? I'm not sure why...

I'm going to ping @rok-cesnovar on this. I believe he has an M1 Mac to mirror your example on

I think it is fine to leave here for now, if we determine it belongs somewhere better like the CmdStan repo we can transfer it then.

I can't reproduce this using gcc on Ubuntu 20.01, it's possible that this is compiler dependent. @rok-cesnovar - do you recall?

I seem to recall this being an issue with gcc being more permissive than clang in terms of multiple declarations. There is a way around this (I believe) if you...

Just as a warning to anyone who wants to try addressing this issue, we've spent a fair amount of effort trying to catch this message and provide guidance in CmdStanPy,...