I do think that they are missing a fairly simple solution for the majority of use cases: move things like `urllib2` to use `libcurl` underneath rather than OpenSSL. The CURL project has already solved this problem by being portable to things like Apple's CommonCrypto library and Window's security libraries.
I should also note that the article's suggested workarounds are less then ideal as both use versions of OpenSSL that do not consult the system keychain for root certificates, and so create strange problems that most people will not understand.
That doesn't sound very simple at all. For one thing, it means the python runtime loses control over the http implementation it ships. It's also not obvious that libcurl provides an API that can do everything something like http.client can do. I think this is somewhere between 'not simple' and 'non-starter'.
I expect the issue will be solved by linking and bundling a recent openssl, which is relatively trivial to do for only one platform.
