-
Notifications
You must be signed in to change notification settings - Fork 95
dependency issues #71
Comments
Hey, happy to hear you are considering using karma to test mocha :) To be honest I'm not a big fan of peerDependencies in general, especially in npm@2. We had a lot if issues with them in development as we use it for specifying the karma version. With npm@3 they are much better I think. For using it in mocha I would suggest having an afterInstall step that deleted node_modules/mocha and symlinks it to ../ so that karma-mocha uses the actual moch version. We do the same thing at the moment on the karma main repo to work around the auto install of peerDependencies. We should change that |
Currently npm v3 throws a max call stack exception if a package is symlinked to itself. I'm not sure if that's a real bug; at the very least they should have a friendlier error message. I'll inquire over there. |
I think warning better than README.md. Users don't read README. For testing mocha you can manually link mocha as node_module in mocha directory. Thanks |
Already fixed in this PR |
Hi,
I'm exploring using
karma-mocha
for testing Mocha itself. I'm concerned aboutpeerDependencies
. With npm v3, an unmet peer dep will result in a warning only. This is fine, but until npm v3 becomes more commonly used, ifkarma-mocha
becomes a dev dep ofmocha
, it's pointless for Mocha tonpm install
itself.(This will also potentially cause problems with forks of Mocha that do various weird things, but I don't know if anyone's complained about that.)
How does the karma-runner org plan on handling the
peerDependencies
issue? Perhaps it's prudent to removepeerDependencies
altogether and note thatmocha
should be installed withinREADME.md
?As discussed in qunitjs/js-reporters#1, the event names which Mocha emits will eventually change. Requiring version
*
of Mocha may be dangerous.The text was updated successfully, but these errors were encountered: