Use ReadOnlySpan<byte> instead of byte arrays

[vcsjones]

A few more changes to use the compiler's ReadOnlySpan<byte> optimization trick instead of static byte arrays.

[ghost]

Tagging subscribers to this area: @bartonjs, @vcsjones
Notify danmosemsft if you want to be subscribed.

[tannergooding]

@stephentoub, do we have an issue for an analyzer for this and sbyte (and the similar, but not quite as efficient, trick for non byte types)?

[GrabYourPitchforks]

One cleanup comment, but otherwise LGTM.

[GrabYourPitchforks]

Since we're taking spans instead of entire arrays, does it make sense to clean up this call site by removing the dwDataLen parameter and inferring the value from the span length?

[vcsjones]

Seems reasonable; so changed. The callers of this always passed in array.Length anyway.

[stephentoub]

do we have an issue for an analyzer for this and sbyte (and the similar, but not quite as efficient, trick for non byte types)?

Not yet implemented, but it's on our list: #33780

[stephentoub]

Would it be better to just do:

if (encodedOid?.Length == 2 && encodedOid[0] == 6 && encodedOid[1] == 0)

?

[vcsjones]

Also a good idea. I changed the method's parameter from byte[] to ReadOnlySpan<byte> to remove the need for the null check and switched to AsnValueReader. The only caller of this method is here

runtime/src/libraries/System.Security.Cryptography.Pkcs/src/System/Security/Cryptography/Pkcs/Pkcs9ContentType.cs

Lines 48 to 51 in 2d1f7b0

	if (rawData == null)
	return null;
	string contentTypeValue = PkcsHelpers.DecodeOid(rawData);

And already did the null check, so the implicit conversion doesn't come in to play.

[vcsjones]

I also considered making a similar change to pSpecifiedDefaultParameters but since it is a negative condition it seemed to make things less readable.

[GrabYourPitchforks]

One CI leg appears to be stalled - even on rerun. Not worried about it.