Replace 'stanc' with 'compile' about protostan

[ariddell]

This is a bit more involved than I thought. We need to compile protostan's tests without relying on Stan's makefile. In principle people using protostan will be working with a different Stan tree and will not have libstan.a on hand (I think this is right).

[sakrejda]

Could you explain why this is tied up with Stan's makefile? I don't see the connection to not using
"stanc" as a name.

[ariddell]

There's a rule in the Makefile which uses libstanc.a -- I'm pretty sure this is an artifact which is there for cmdstan from the days when cmdstan was part of the stan repo. I may be wrong about this.

[sakrejda]

I thought PyStan would also need to do the compilation required to make something like libstanc.a, no? Do you really recompile everything for every model?

[ariddell]

PyStan compiles a hybrid Python/C++ thing for compilation separately.
When models are compiled they don't need to use the stan/lang stuff.

For testing I think it makes sense to leave libstanc.a the way it is.
But I do think it's a good idea to replace 'stanc' with 'compile'
elsewhere. That's the function signature after all. Without knowledge of
cmdstan I think 'stanc' doesn't have a clear reference.

On 01/04, Krzysztof Sakrejda wrote:

I thought PyStan would also need to do the compilation required to make something like libstanc.a, no? Do you really recompile everything for every model?

---

Reply to this email directly or view it on GitHub:
#18 (comment)

[sakrejda]

Ok, works for me. I can take care of it, I think there's some proto
file/c++ namespace issue involved.
On Jan 5, 2016 9:04 AM, "Allen Riddell" [email protected] wrote:

PyStan compiles a hybrid Python/C++ thing for compilation separately.
When models are compiled they don't need to use the stan/lang stuff.

For testing I think it makes sense to leave libstanc.a the way it is.
But I do think it's a good idea to replace 'stanc' with 'compile'
elsewhere. That's the function signature after all. Without knowledge of
cmdstan I think 'stanc' doesn't have a clear reference.

On 01/04, Krzysztof Sakrejda wrote:

I thought PyStan would also need to do the compilation required to make
something like libstanc.a, no? Do you really recompile everything for every
model?

---

Reply to this email directly or view it on GitHub:
#18 (comment)

—
Reply to this email directly or view it on GitHub
#18 (comment).

[ariddell]

I've got a PR ready, let me send that.

[sakrejda]

Ok
On Jan 5, 2016 9:39 AM, "Allen Riddell" [email protected] wrote:

I've got a PR ready, let me send that.

—
Reply to this email directly or view it on GitHub
#18 (comment).