But couldn't it be the case that with this bug you could sweep private keys from server's memory if they happen to be in there, because bitcoind is using them at the moment?
Yes, I would put that under the "implications for Bitcoin web services that use HTTPS" category. For context, the deleted question asked about the "Bitcoin blockchain" so I was responding to that specifically, with the added caveat for Bitcoin services that use HTTPS.
