Jamie Gaskins

I added 990e3c5 to avoid raising the error on missing env vars, which should at least let you run your migration without working around this feature. It's not an ideal...

I've been meaning to add this for almost a year and a half, ever since I wrote [this post](https://forum.crystal-lang.org/t/having-a-blast-with-crystal/5759). I finally got around to extracting it from that app this...

It just occurred to me that this shard may not be the right place for this feature. It requires adding [other shards](https://github.com/jgaskins/armature/pull/7/files#diff-cbaa8062ee0fcfb5af8adfd98f69af4766b9ad5bb5e79158abe8da34f6fd7de0R12-R15) as production dependencies that don't make sense. I've...

This looks like a great start. A couple areas of focus: - Ranges can be `nil` on either end - It would be awesome to see some tests — this...

Tests *and* specs? In this economy?

This looks great! Thanks for improving on my PR!

@BlobCodes Are you still working on this? If not, I'll go ahead and pick #14604 back up.

> This would be akin to literal types in languages like [TypeScript](https://www.typescriptlang.org/docs/handbook/2/everyday-types.html#literal-types), where if a value appears as a type restriction, it represents the type consisting only that value. Thus...

This is a good point. Should we instead open this issue on the Crystal compiler? I opened it here because for some reason I hadn't considered that other `X::Serializable` mixins...

That's a solid point. Scaling Postgres (especially for writes) is far more challenging than scaling out app servers when you hit a CPU threshold in Crystal app code.