Complains of string not array, repoman eject in a non-repoman folder

Currently I'm using my steelbrain/publish to publish my npm packages.

It has some lifecycles like checking if certain files are ignored before publishing etc, if we could add support in repoman for those, I could switch all my open source repo management to repoman

So we have an up to date idea of what's happening

• publish to publish, optional org, optional org/package, handles bumping versions
• link <org> links together an orgs files (multiple orgs?)
• exec -- npm run watch good, but what about if some dont have watch? watch maybe?
• cd helps you cd into things, preferably with fuzzy match repoman cd chunky

Major questions:

Linking into monorepo requires some sort of knowledge of the monorepo and all its sub-packages which is an unfortunate complexity... any way to simplify???

• link <...orgs> --into mono-repo --recursive could do it

Finally, not sure why we have dependencies key in config, that should be handled flexibly without needing a config.

Questions on having a repos(.json)? inside ~/projects/*:

• Can share with team to have consistent dev env
• Where do you store it? Github?
  • Perhaps it could just point to a remote file! the whole file could just be:
    • motion/repos/repos.json
    • then on clone it could pull everything from that?.... (complex)
• How do you keep things in sync?
  • repoman clone-set []
  • repoman clone motion/repos/repos.json smart?
  • repoman new set => prompts for location

Naming for "sets" or "groups" or "orgs":

• repoman sets => show list of sets?
• repoman set clone [] => commands for a set? but this is weird...
• repoman clone set [] => better
• repoman set motion publish => prefix with set?
• repoman publish motion => or postfix? seems less flexible

repoman sync in a dir to see

This will give a more accurate linking experience

Making so I don't forget

Lets make this a bit more focused to start. There are two high level directions, lets do the second:

• Make managing lots of github repos easier
• Make managing lots of small node packages easier

Goal is to be able to go from 0 to published package easy, and make ongoing maintenance easy. It's basically lerna sans mono-repo, maybe even more opinionated.

Goals for v1:

1. Track directories
2. Exec commands easily in all directories
3. Share scripts between all repos easily (and have a default set that uses babel + presets-steelbrain)
4. Be really good at linking stuff together thats tracked by it (lerna-style)

Bonus points:

• See nice status of all repos
• Publishing helpers