The postgres protocol does support pipelining of queries - and it can be a huge boon in latency sensitive workloads. They'll get processed in-order on the server side, with results being sent back while the next query is being processed.
The biggest weakness around pipelining is that a fair number of drivers don't support it yet, including the C client interface that is part of postgres (there's a patch being reviewed right now adding it there). The common jdbc driver, .net, and a few other popular ones do support it though.
I've not seen any real efforts to add support for different protocols to postgres. Do you really think just adding protocol level support for e.g. TDS would be useful, if the the SQL dialect still was postgres's? If so - why?
While I am employed by MS, I just work on PG, and I have long before starting at MS. So I just have the open source hacker's perspective on this, not any MS perspective.
The biggest weakness around pipelining is that a fair number of drivers don't support it yet, including the C client interface that is part of postgres (there's a patch being reviewed right now adding it there). The common jdbc driver, .net, and a few other popular ones do support it though.