Thomas Holenstein

This is an issue I'm working on in a larger context. Code will look something like this: ``` KeysetHandle.Builder builder = KeysetHandle.newBuilder(handle); builder.addEntry(KeysetHandle.newEntry(KeyTemplates.get("HMAC_SHA512_512BITTAG").withRandomId().makePrimary()); KeysetHandle keysetHandle = builder.build(); // Now you...

I suspect not, but it will depend when the next release it. I would assume it will take minimum 6 months.

Note that while this is still in alpha, you can in principle use it in 1.7. However, while I don't think it should be needed, there is a small chance...

This is, indeed, an issue we eventually need to address. However, at the moment I personally tend to the variant that we have an always incrementing ID in the keyset...

Yes, one can indeed make arguments for both, and I'm interested in hearing from people. I just look at a lot of code in other libraries which do similar things...

In principle, Tink should not require a proguard config at all. If it does for part of it (even the protos), I might actually check and see if I can...

Thanks Thai, I forgot we had to make this change. I would still check the log file to see what the classes d.g.c.a.a and d.g.c.a.s are, they are obviously classes...

Yes, the issue is that you are building an HMAC key (HmacKeyFormat.newBuilder()...), but you want to use it with an AEAD. HMAC keys are made for the Mac primitive.

I'm not happy that this might be needed. Of course, the Aaron's point is that it is needed :/ However, at the moment I would still somewhat prefer not to...

I'm closing this, as I don't think eeSeeGee is interested in doing the changes we wanted. Also, one should check whether escape analysis works, and keep the code if it...