@htuch this is pretty ugly, any ideas if there are @type tricks for the text proto? I couldn't find any myself.

There aren't, @jmillikin-stripe has proposed we adopt Any for well-known protos. I'm unconvinced that proto text is desirable in general; it's not properly standardized and we already have a well supported text format (which Google is preferring going forward for proto text representation). I would drop this review for this reason.

+1 I'm not sure we really want to demonstrate this to people. IMO we should push people towards JSON/YAML. It looks much nicer.

I agree that due to use of google.protobuf.Struct the text format is pretty ugly and most people willl prefer JSON/YAML... But if we don't want to support text proto, then we should revert #1593, IMHO. We either support it (and have this trivial test that verifies it works) or not.

Even though using google.protobuf.Struct to encode normal messages is unpleasant, it's good to have a config format that both (1) has comments and (2) doesn't have a rich history of code execution vulnerabilities. Text protobufs are a nicely inert format that can still be documented.

I don't really have a super strong opinion about this. As long as the proto library supports this we can support it. I don't want to really encourage people to use this though. @htuch ?

@PiotrSikora I think you need to push an empty commit to kick the DCO bot.

Let's resolve #1680 first so we can decide where we want to go with proto text.

What's our current thinking on this one? Personally I'm tempted to skip it. Note that @jmillikin-stripe did add test coverage for a basic proto text test, it's just written out in code. @PiotrSikora @htuch @jmillikin-stripe ? (This doesn't mean not supporting it via load, it just doesn't mean publicizing it).

I'm fine with not having an example.

Let's make #1680 happen first, then we can support an example/test which will be readable.

Let's make #1680 happen first, then we can support an example/test which will be readable.

It's not actually clear to me that we should fix that issue, but let's discuss there.

Let's close this out for now, feel free to reopen when we have some progress on #1680.