Ouroboros Chrysopoeia

> you end up with separate files for each partition It turns out that I had to change to using partitions anyway because simple queries such as `select * from...

> So all is well? Yes. This is not a blocking issue for me at this time. It's just a low-priority wishlist issue. Thank you.

More simply, I think it would be nice to have a backend feature to merge an *unsubmitted* new order into an accepted order. This is assuming the direction of `long`...

I am not convinced that this should be a global setting for the account. I currently take the more conservative opinion that it should be API driven.

> Someone from Alpaca team has suggested a method in the py package. I have looked and there is no applicable method that replaces an accepted order. The only way...

By the way, I am seeing two timestamps in the raw quote, both as `msgpack.ext.Timestamp` objects, with keys `t` and `r`. Only `t` is [documented](https://alpaca.markets/docs/api-references/market-data-api/stock-pricing-data/realtime/#quote) whereas `r` isn't. Examples: ```...

> @impredicative, There is no harm, but using the "raw" flag I would expect a "RFC-3339 formatted timestamp with nanosecond precision". I believe it's natively coming in through the socket...

msgpack is ostensibly specified as the desired encoding by the client, and so in my understanding it makes sense for the response to use a msgpack object where it helps....

Well, it is important for the user to have both the integer seconds value and the integer nanoseconds value. If a msgpack object was not used, the two values would...

I see your point. Perhaps the arg name `raw_data` can be changed to `native`.