Moving away from Grin

[SquidDev]

@ElvishJerricco. I know that historically you've mentioned you want to stick with Grin to install ClamShell. However I'd quite like to move away from Grin to consider various other distribution methods. My main reason for this is that Grin/Grin-Get, whilst really cool, is rather cumbersome for new users to set up, is rather slow (see Team-CC-Corp/Grin-Get#5) and is broken under CCTweaks.

I'm not sure what I'd prefer to use instead. My instinct would be to ship two files:

• The main clam.lua file (and its dependencies). Probably combined using Howl or something.
• All the programs in tools + argument parsing library. These could be bundled together into an extractable file.

Installation could then just be composed of pastebin get/pastebin run on a couple of files.

Thoughts?

[ElvishJerricco]

It should be possible to create a very simple Grin-based installer, so that people can install ClamShell with just pastebin run .... This was the main strength of Grin sans grin-get.

Remind me, what exactly is broken under CCTweaks?

[SquidDev]

Sorry, for whatever reason I didn't see this earlier. Team-CC-Corp/Grin#1 documents the CCTweaks issue a little better.

I guess it would probably be possible to wrap up the whole Grin/Grin-Get/ClamShell installation into a single pastebin which might make things a little easier.

[ElvishJerricco]

I wonder how hard it would be to package the grin-get API so that ClamShell can be used without installing grin-get at all.

[ElvishJerricco]

I wouldn't mind ditching Grin altogether as long as we move to a more structured alternative than Pastebin. CC needs some kind of package manager.

[SquidDev]

It's worth noting that ComputerCraft now supports require - we could look into adding support in Grin-Get for that and migrate to use it - this way it would be easier to for people to package ClamShell in a non-Grin-Get system, but still use its functionality.