Re: [TLS] SHA-1 vs. FNV-1

[Eric Rescorla <[e-mail]>]

On Sun, May 9, 2010 at 2:07 PM, Michael D'Errico <[e-mail]> wrote:
> Eric Rescorla wrote:
>>
>> Really, given that the security analysis for why you don't need a secure
>> hash needs to be written in any case, and that we're NOT using MD5
>> but rather SHA-1, I don't see that the argument you're presenting
>> for deviating from our standard SHA-1 using practice is compelling at
>> all.
>
> Won't there be a push to get rid of SHA-1 from protocols just as there
> is for MD5 today?

Yes, there will, and we'll send those who suggest we remove it from this
location the relevant security analysis and that will be that.


> It seems prudent NOT to use SHA-1 precisely for the
> same reason we are not using MD5.

Not really. We are (well, were before Stefan changed it) using SHA-1
because it's what everyone uses. Which, incidentally, is precisely the
argument for using it instead of FNV-1, which, by comparison, practically
nobody uses.

-Ekr