I was under the impression that the Sign conversion / Modulo change of the signature was immediately obvious, because the default client would only choose positive R and S and there is only one other solution in the available length.
Only if it's immediately obvious that every user is using OpenSSL-linked bitcoin-qt. That's neither obvious nor true!
For example, the transactions I use to move coins in and out of cold storage are generated by a javascript client, since it can run on an "air-walled" laptop with no network connection. A lot of people do this. In that case the sign of the ECDSA signature depends on what javascript VM I'm using
Standards are standards, especially for crypto -- never a good idea to start assuming people are using specific implementations.
Yes, but wouldn't it be beneficial to agree on a canonical signature with positive R and S that gets preferential treatment for mining? Otherwise with your change hanging on confirmations, you could get severely limited in exchanging digital coins which supposedly are the new fast thing.