SQLx v0.6 and beyond

[mehcode]

:man_surfing: I'd like to thank everyone for filling out the survey. ❤️ It will help to inform our priorities for 0.6 and beyond. Which, overwhelmingly, it seems, are:

1. Efficient batch insertion
2. Minimal dynamic query builder
3. Asynchronous query pipelining
4. User-defined array types

As expected, over 3/4 of SQLx users use the compile-time type checking. Our dynamic usage is there, but low. This could also be that if you're comparing dynamic SQLx to other database libraries, we currently fall short.

Unsurprisingly, Postgres is used by 3/4 of the SQLx users. This is also likely influenced by the less-than-stellar capability of the other drivers.

As for adding more drivers, our roadmap is definitely:

1. MSSQL (complete)
2. Redshift & Cockroach
3. Oracle

---

Moving on, I'd like to give an update on the potentially proprietary question. We've decided to move forward with this approach for v0.6 and beyond. I'll try and cover all the bases below, but feel free to ping me or another team member with further questions.

1. Open-source, popular, and freely licensed databases will be supported and open-source as long as they are themselves officially supported.

Examples: PostgreSQL, MySQL, MariaDB, SQLite, CockroachDb

If further-backwards support (PostgreSQL 7, MySQL 5.4, etc.) is needed, it can be handled on a case-by-case basis. We'll do something like the "platform tiers" that Rust has where these will be in a "might work but not tested for" tier. Official (paid) support for older versions will be available.

2. Closed-source, proprietary, non-freely licensed, or obscure databases will be closed-source and at a cost.

Examples: Google Cloud Spanner, Redshift (†), MSSQL, Oracle, DB2, MS Access, Amazon Aurora (†)

(†) Connecting to proprietary connectors that mimic the open source protocols may work but will not be officially supported under open source. If it can be made to work by detecting its existence, then PRs would be accepted but it would be the "might work but not tested tier" as discussed above.

Free licenses for open-source or other non-commercial use will be made available for the closed-source connectors on request with a human approval process.

We haven't nailed down how purchasing is going to work but we'll share details as we get closer. With this, we'll be making a big effort towards documentation and a web presence at https://sqlx.rs/ (nothing yet).

We're targeting Q1 of 2021 for v0.6 and https://sqlx.rs/.

If there are any questions, please, don't hesitate to ask in a discussion thread (in General, not announcements please) or on Discord.

---

With that said, work on v0.6 has begun in the next branch on GitHub.