[TLS] Re: 2nd Working Group Last Call for The SSLKEYLOGFILE Format for TLS

[David Benjamin <[e-mail]>]

Accepting both labels gets super messy because then we have to make a bunch
of decisions like whether you output both labels on the logging side.

But we can just do a bit of research here:
- In IETF land, EARLY_EXPORTER_MASTER_SECRET dates to the start of the I-D,
but...
- The shorter EXPORTER_SECRET name for the non-early secret dates to the
earliest proposals for TLS 1.3 here:
https://bugzilla.mozilla.org/show_bug.cgi?id=1287711
- BoringSSL does not output this label
- OpenSSL does not output this label
- NSS outputs this label but uses EARLY_EXPORTER_SECRET
- Wireshark consumes this label but uses EARLY_EXPORTER_SECRET

So I think EARLY_EXPORTER_MASTER_SECRET was just a typo and should always
have been EARLY_EXPORTER_SECRET. Unless there's any evidence that someone
actually relies on the EARLY_EXPORTER_MASTER_SECRET label (very, very
unlikely given both the history of early exporters and the history of this
SSLKEYLOGFILE integration), I think the answer is clear: No, we should not
accept both labels. We should simply fix it to say EARLY_EXPORTER_SECRET
and move on.

David

On Fri, Feb 7, 2025 at 1:33 PM Salz, Rich <[e-mail]> wrote:

> The question is really "should we accept both names?"
>
>