Zero sized arrays


Could we quickly align on where the issue of serialzing of zero sized arrays should go to? I am referring to the issue i opened in rstan #371. I understood from Daniel it should go to Stan. Putting it there lead Bob to move it to a new issue in rstan with #373 which is now a duplicate.

I don’t know where it has to go, suggestions?


Ps: this type or confusion where issues belong to will probably get worse once we split things, so we need a clever approach to deal with it if possible.

Pps: I know that this is not the most urgent thing at all right now… so no hurry

Your issue didn’t point to a problem in stan-dev/stan.

Given that CmdStan’s I/O didn’t change in any way, shouldn’t
this just be an RStan issue? Obviously, no need for dupes.

If you want to open an issue against Stan itself in stan-dev/stan,
it’d help to know where there is currently an error in the Stan
code and what should be done to fix it.

  • Bob

Hi Bob,

well, I thought myself that this issue belongs into the rstan bucket; but Daniel advised to file an issue to Stan as it is a problem in dump.hpp. At least this is what I understood. I hope to find some time to follow this up.


Hi Sebastian,

The problem with the issue you filed was that you didn’t include a key piece of information in the issue. If you reread the issue you filed, it says RStan and CmdStan. There is no mention of Stan and specifically, dump.hpp. If you had written that in, it would have been much clearer. I might not have made it clear to you, but I passed that info so you could put it into the issue.

Hope that clears up the confusion. For future issues, please include context that makes it clear why it’s there and what the purpose is for. You also can’t really assume everyone seeing the issues has followed discussions everywhere. It needs to be standalone as much as possible.


1 Like