Sahan Paliskara

create flag to indicate a program is inside a deploy interpreter

1

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #310 * #308

match pybind with pytorch core and remove inductor test

1

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #310 * __->__ #308

When code fails in a subinterpreter we should have our error point to where the runtime error occurs in python.

Currently torchinductor is broken with deploy as it fails to find cpython symbols. We need to do some creative linking to fix this.

update pybind to 2.10.2

6

Updated to unblock https://github.com/pytorch/pytorch/pull/91248#issuecomment-1402327940 and the refcount debug utils will be useful :)

We have this support, but its non-obvious how to use it.

Remove dead includes in multipy

2

Differential Revision: D42602001

torch dyanamo in c++ test

2

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #299 * #298 Differential Revision: [D42551441](https://our.internmc.facebook.com/intern/diff/D42551441)

Add lintrunner support to multipy

2

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #223 * #222 Add [lintrunner](https://github.com/pytorch/pytorch/wiki/lintrunner) support to the repo. All linters are copied over from pytorch core. Differential Revision: [D40607774](https://our.internmc.facebook.com/intern/diff/D40607774)

This issue can be reproduced using docker with https://github.com/pytorch/multipy/pull/233, otherwise, one a) needs to be running the unit test on a cuda enabled machine + on python 3.7/3.8. This error...