Paul Smith

Or use https://httpie.org which is sooo much nicer than curl (IMO)

Yeah good point, I think we should start moving the generators to manually defined routes to start out with! That should be an "easy" change

Could we get the same (or similar) info by asking for the `shard.lock`? I imagine that's what `--framework-versions` would use under the hood.

That seems like it'd be really good to know. Since if they're using a specific commit it may not be the officially released version. But I may be thinking about...

@matthewmcgarvey This is a fantastic idea, I think this can be done using the `memoize` macro. I can make a PR for this :)

The original idea was to have one assertion method that worked for everything. But you make some very good points, especially since Lucky is very much about safety and not...

Yes definitely. We will work on this! In the meantime I'd suggest adding a fake User in your spec_helper ``` class User def id 0 end end ```

Yeah, that huge stacktrace would be nice to shorten up. I think using a `while` or `loop` in https://github.com/luckyframework/lucky_flow/blob/324116d8ba696eebabbeb196cd915bdccf381547/src/lucky_flow/find_element.cr#L13-L24 might work better. But I haven't tried it yet so it...

Thanks for the issue and especially for the thorough examples. **I like the idea of making it configurable so people can use whatever they like!** **I'm not sold on `data-hook`**...

I think this would be quite complex since we'd also need to change Avram to support different transactions per thread, and then use a session or something (as @matthewmcgarvey mentioned)...