Project generator

[lpil]

Such that one can run a command and have a working project with a sensible amount of bells and whistles.

Depends on the recommended project structure being decided: #18

Should this belong in wisp or another dev-only package?

[bhankas]

Should this belong in wisp or another dev-only package?

I'd vote for former. Its easier on new folks to have stuff in one place. Github search is often enough to ream through such repos.

[lpil]

I'm not sure it should be a package at all now. Let's figure out what the design is and then we can decide where it would be appropriate for it to live.

[GiregL]

Hi !

I'd like to work on this subject if needed.
I'd suggest an embeded CLI, allowing us to do gleam run -m wisp <custom commands>

With, according to the recommanded structure, a way to create a base project (maybe even add classic dependencies such as pog, squirrel, ... ?) and parts of the project (features if domain-driven dev, ...)

[lpil]

One drawback of that is that you cannot use that approach to make a project, you would need to have already made one, added that dependency, and then run it from there. It would mean that how to install the command is easy though!

What would the command do?

[GiregL]

It would generate the basic files (such as the router, the base middleware, .env file for the database, ...) only
It's not that much yeah

The other idea would be to provide with gleam toolings a way to create projects from templates, or to run gleam modules that would be installed system-side. It could be helpful for other projects such as Lustre too

[lpil]

If it's always the same content it seems like a template repository would do find instead.

How would system-wide installations of Gleam modules work?

[JosephTLyons]

Just chiming in with my experience, if its useful at all.

Wisp is really cool and I plan to attempt to use it to build a personal website, but not having a way to bootstrap a project was causing me the most amount of friction, for some odd reason. I put off messing with it for a few weeks solely over that.

I ended up just cloning the examples directory and slimming it down to one project, renaming files to match the app name I wanted, removing comments, etc. That isn't terrible by any means, but not as clean as just a command or something that puts everything in place.

From working on Zed, one of the things I've become intimately aware of is the fact that even if the core experience of something is generally great, many people simply will bail if the initial setup has any amount of friction, even if it seems super silly. I have no insights into whether people are using Wisp or not, but IMO, a project generator definitely would go a long way in getting people to the place where they can experience how nice it is, faster, and increase the rate of adoption.

[lpil]

We would want something significantly more substantial than any of the examples, they don't have very much in them. The aim will be to provide a significant amount of value and convenience.