Define database model

[naota]

Original GenTwoo DB model was:

• User
  • twitter_id, login name, other twitter staffs
  • user configuration
• Package
  • category
  • name
  • version
• Emerge
  • buildtime
  • duration
  • package_id
  • user_id
  • log
  • errorlog
• Comment
  • user_id
  • comment
  • emerge_id (or package_id)
  • timestamp

[naota]

At least I'd like to add "Machine" model for the new version.

• Machine
  • name
  • user_id

and Emerge should have "machine_id" instead of "user_id"

[jan-matejka]

What does other twitter staffs means?

[jan-matejka]

I suppose, a record in Package corresponds to an Installed Package, right?

[naota]

What does other twitter staffs means?

They are Twitter's icon url, screen name, login id or such things. It's there just because the twitter library setup the DB for it.

[jan-matejka]

riiight. I figured, but I have no idea what do you mean by the word "staffs". I'd call this attributes.

[jan-matejka]

Unless you have comments for the model I have drafted in the mvp-spec, I'd close this issue.

[naota]

I'd love to have "repository" within Packages to identify same "category, name, version" but different repository.

Also, if it's possible, I'd like to gather information about package's configuration e.g. enabled USE flags.

[jan-matejka]

Oh right. Repository should definetly be a part of that.

But I think USE flags could be left to next version. I'm not sure they will be needed. The USE flags are used mostly for 1. enabling compile time opts. 2. doing number 1 + requiring a dependency or 3. requiring a RDEPEND, which is frowned upon.

Since we are already tracking all the packages, I'm not sure there is any benefit in tracking use flags per package too.

[naota]

I might missed the point: 0.1.0 is not tracking failure log, right? If so gathering USE is surely pointless for this version.

[jan-matejka]

That's right, only ${P} install/uninstall.