shema

In a fully typesafe world, it should be pretty hard to derive from the shema this way.

A lot of the registration data will have a defined shema ex: name, email, address.

When it comes to rdbms shema, you want it optimized and enforced from day one, or the price paid later will be significant.

Upon each release I will migrate validated shema-less properties to static columns or even sub-tables.

Metrics are just logs with well defined shema.

I only worked briefly with a nosql 'real' production database but we needed to change the data shema, and migration was a pain (like any migration).

I see some value in that, but in my experience it's actually the queries that inform much of what the table shema and indexes should look like.

So why did this tech disappeared if it was ultimately better than the later alternatives (textual protocols, shema-less serializers, and eventually protobuf which reinstated some form of efficient encoding and type checking).

Naively I find them even worse because you can have multiple shema at once on the same table, so each read on an important document must include a sanity check 'what is the current shema of this particular entry, should we migrate it?'.