Joshua Haberman

Ideally this could be done at the Ruby level, with a method alias like some of the ones we have here: https://github.com/protocolbuffers/protobuf/blob/main/ruby/lib/google/protobuf/message_exts.rb Does that seem viable? That would save us...

Does it work to replace `NAN` with `0.0 / 0.0`?

This seems reasonable. I agree we could accept a contribution if it was done in a backwards-compatible manner. By the way, `protostream_objectsource.cc` is not part of our repo anymore. It...

> Is there an alternative other than compile time option or source code change, by increasing kDefaultMaxRecursionDepth, for protobuf users? No, that is what this issue is proposing to fix....

@LisaFC looks like this is worth investigating. By the way, @JakeWharton do you have your own `.proto` file parser? Would be curious to know the use case for that.

With C++11 we could use `constexpr` functions to strip the leading directories at compile time. I think this will be hard to fix until we have C++11 though. The log...

I wonder if we should do something more drastic and more immediately obvious: 1. Move all files into an `archived/` top-level directory, so the file list is short and the...

The only downside of this is that clang-format doesn't know about this convention, and will reformat it as: ```c typedef struct { PyObject_HEAD PyObject* pool; // We own a ref....

I haven't (yet) been able to find a way to get the `.clang-format` config file to be respected in the google3 source tree.

We have an automated system that should take over now and submit the change internally first, with it propagating to GitHub after.