>I'm also dubious about the "format" used - the python tool we had been using, and the new shell tool I wrote to replace it, both just read regular SQL files in a series of directories (one called `up.sql` and one called `down.sql`).
This is something I considered, whereby the revisions themselves would be stored in `up`, and `down` directories beneath the revisions directory itself. It would make things even more transparent I suppose when it comes to integrating with other SQL tooling, instead of having to rely on some ad-hoc DSL. However, I chose to use the special format because for me, I found it simpler to work with, each revision is its own SQL file, and for the size of the revisions I was writing this suits me fine.
>As with the python tool we had been using, this doesn't appear to support SSL connections for mysql/postgresql?
Yes, unfortunately. I will need to look into testing this, as it was something I overlooked. For my use case specifically I was simply using it against local databases, and yes that does make for a weak excuse as to why SSL is not supported.
>Anyway, good luck with it. Ultimately these tools usually come about to solve some problem or scratch some itch, so I hope your itch is scratched!
Essentially, I wanted something simple that read in plain SQL and kept a log of what had been performed. Though with what you said about using separate `up`/`down` files for the respective directions would make things far simpler than they currently are.
For this to work the revisions would need to be changed from standalone SQL files, to actual directories themselves. Not that big of an issue, but perhaps something I will look into. Ideally I want this tool to play nice with other tools in regards to the data it produces.