Add network_latency_ms field to ALTS's NextHandshakeMessageReq.

[matthewstevenson88]

Already internally-reviewed (see cl/589894002 and cl/589974172).

[ejona86]

(FWIW, it tends to be useful to copy the CL number here so we can follow things a bit easier. Especially further in the future and trying to figure out what we did.)

[matthewstevenson88]

(FWIW, it tends to be useful to copy the CL number here so we can follow things a bit easier. Especially further in the future and trying to figure out what we did.)

Sure thing, done. :)

[ejona86]

optional is not allowed? We should able able to do that, although we may find some places that need upgrades. It looks like the CI here is just using too old a protobuf.

optional in proto3 provides presence information, and it looks like you would want that. How much do you care?

[matthewstevenson88]

optional is not allowed? We should able able to do that, although we may find some places that need upgrades. It looks like the CI here is just using too old a protobuf.

optional in proto3 provides presence information, and it looks like you would want that. How much do you care?

It's not a big deal in this case, because I only care about the positive values anyway, so "zero" and "not set" are equivalent.

If there's an easy fix to allow optional so that the next person doesn't run into this issue (e.g. maybe just update the @com_google_protobuf dep to the newest version?), I'm happy to do it in this PR or in a follow-up, whatever you prefer.

[ejona86]

I'm happy to do it in this PR or in a follow-up, whatever you prefer.

Thanks for the offer, but I think I have it. We're going to need to merge #123 and then bumping past 21.x required a non-obvious line and bumping to 25 required further changes to bazel. I've fought this before in other circumstances.

[matthewstevenson88]

Thanks Eric!