Re: [PATCH 1/6] ssh signing: extend check_signature to accept payload metadata

[Junio C Hamano <gitster@pobox.com>]

Fabian Stelzer <fs@gigacodes.de> writes:

> On 24.10.21 01:13, Junio C Hamano wrote:
>>
>> One possible downside is that sigc has been mostly output-only
>> structure, and turning it into a structure that also has some input
>> members might make it too confusing.  I dunno.
>
> However if everyone is ok with changing the struct to be used for input
> as well then i would adjust the function to have it as the first parameter.
>
> The sigc struct already has a payload member (only used for verbose
> output) which is populated by the check function with a xmemdupz. I
> would then change it to a const char, add the length var and use it to
> pass the payload into the function as well. This way we also avoid the
> unnecessary mem copy.
>
> The function would just become:
> check_signature(struct signature_check *sigc, const char *signature,
> size_t slen)

I do not offhand think of a huge downside going that route myself.
We seem to end up with unusually large number of folks on the CC
list for some reason, so hopefully somebody will stop us if it is a
stupid idea ;-)

Thanks.