Cache encodings of known OIDs

Cache encodings of known OIDs

[ekasper, 2013-12-02 13:29:57]

This makes decoding certs a little faster because name attributes are full of
common OIDs. I'm paying an advance for making it slower again when I add more
X509v3 extensions.

[Ben Laurie (Google), 2013-12-04 11:28:27]

https://codereview.appspot.com/35920043/diff/1/src/python/ct/crypto/asn1/oid.py
File src/python/ct/crypto/asn1/oid.py (right):

https://codereview.appspot.com/35920043/diff/1/src/python/ct/crypto/asn1/oid....
src/python/ct/crypto/asn1/oid.py:149: return tuple(components)
Would it make sense to add this to the dict here?

Would it make sense to just use the dict as a cache, rather than explicit
initialisation?

[ekasper, 2013-12-05 19:33:21]

https://codereview.appspot.com/35920043/diff/1/src/python/ct/crypto/asn1/oid.py
File src/python/ct/crypto/asn1/oid.py (right):

https://codereview.appspot.com/35920043/diff/1/src/python/ct/crypto/asn1/oid....
src/python/ct/crypto/asn1/oid.py:149: return tuple(components)
On 2013/12/04 11:28:27, Ben Laurie (Google) wrote:
> Would it make sense to add this to the dict here?
> 
> Would it make sense to just use the dict as a cache, rather than explicit
> initialisation?

That's a very cool idea but it'd then be prudent to at least have an
upper/configurable limit. I played with it a bit but it quickly gets more
messy/thread-unsafe/slow than the little optimization is worth.

But I've meanwhile realized this is Python, so I've removed the hard-coded
second dict.

[ekasper, 2013-12-10 19:21:24]

Ping (no particular rush but I figured you might have simply missed it).

[Eran, 2013-12-11 10:35:58]

Drive-by LGTM.

[Ben Laurie (Google), 2013-12-11 10:38:48]

LGTM