123 reputation
5
bio website acooke.org
location Santiago, Chile
age
visits member for 2 years, 7 months
seen Dec 2 '13 at 23:34

I work on projects that mix computing and physics - I have a pile of experience as a software engineer, but my education, way back in the day, was as an astronomer. And it was research in astronomy that first took me to Chile, where I now live. So I telecommute from Santiago, working with companies in the USA and UK. Currently I'm part of a small team developing software related to geophysics - I guess you could call it a consultancy, but really we're just people that understand what you need and then - importantly! - make it.

Some of my favourite answers:

In every case I learnt something new, which is what's so good about this site :)


Dec
30
comment Future-Proof Versioning and Validation
@JesperMortensen afaik keyczar doesn't work w python 3. if it does, please tell me. incidentally, simple-crypt is secure against that attack and it's documented in the code (not through any brilliance of mine, but because i have been pestering people to review the code and including their feedback)
Dec
28
comment Future-Proof Versioning and Validation
@Thomas oh, an extra HMAC (one that doesn't change with protocol ID), sorry, right.
Dec
27
comment Future-Proof Versioning and Validation
i don't have a specific attack in mind (hence my comment that maybe i need to wait for more specifics). i'm just assuming i may have screwed up. thanks for the comments - it's given me more to think about. i'll wait a while to see if anyone else posts, but that was help enough for a "correct" answer, thanks.
Dec
27
comment Future-Proof Versioning and Validation
thanks. so apologies for not being clear enough that what seems to be the root of the problem is that i want to both (1) automatically support decrypting data encrypted with an older version and (2) somehow avoid attacks that supporting an older version could imply. i guess looked at that way i may be asking the impossible. to answer the final q - the api is ideally two functions, encrypt and decrypt. but maybe i need to consider conversion instead of automatic backwards support. ps the current heder includes what you are calling a protocol id.
Dec
27
comment Future-Proof Versioning and Validation
@dchest - true, but in general corrupting the data will simply abort things. changing the version number triggers a more specific action (update your software) that might be combined with some kind of attack where a fake replacement is provided. i realise i may be taking things too far, but it seemed like it was different...
Dec
27
comment Future-Proof Versioning and Validation
@PaĆ­loEbermann - yes, clarified