Explain why we check hashes before signatures

[trishankatdatadog]

An attempt to fix #138

Does this look good @tedbow?

Signed-off-by: Trishank Karthik Kuppusamy [email protected]

[lukpueh]

Thanks for clarifying this, @trishankatdatadog. I wonder if we can make it more concise.

It is safe to check the hashes before the signatures, because the hashes comes from the timestamp role, which we have already verified in the previous step

feels a bit redundant to

hashes ... listed in the trusted timestamp metadata

Maybe it's enough to just expand what's already there, e.g.

hashes ... listed in the above/previously/already verified and thus trusted timestamp metadata

?

On the other hand, it seems worthwhile to elaborate on the

quick way to reject bad metadata

What about something along the lines of

preliminary integrity check before performing a more expensive signature verification

?

Besides, should this be part of the spec or is it rather a contender for secondary literature (#91)?

[trishankatdatadog]

Sounds good, I will address your comments. In the meantime, I think we should add this to the spec, because that's where the question came up for php-tuf while implementing, but I'll let @tedbow decide.

[joshuagl]

I agree that the spec is a good place for this, at least until we have the secondary literature. There are other parts of the spec where we explain why things are they way they are, such as the recently introduced fixed update start time.

[trishankatdatadog]

@joshuagl @mnm678 PTAL 👀

[joshuagl]

LGTM, thanks Trishank