cordova-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michal Mocny <mmo...@chromium.org>
Subject Re: CLI build failure
Date Tue, 29 Oct 2013 12:27:17 GMT
I think you can also install plugman into node_modules *before* running
cli's npm install and it will not try to fetch from npm since there is
already a compatible version there.

-Michal


On Tue, Oct 29, 2013 at 8:15 AM, Carlos Santana <csantana23@gmail.com>wrote:

> David,
>   One thing you can can try is after you clone cordova-cli
> edit cordova-cli package.json to point to a Git URL instead of npm for
> plugman, this is before running npm install on cli
>
> from this:
> "plugman": "0.14.x",
>
> pick one: base on [1]
> "plugman": "apache/cordova-plugman.git#0.14.x",
> "plugman": "git://git.apache.org/cordova-plugman.git#0.14.x",
> "plugman": "
> https://git-wip-us.apache.org/repos/asf/cordova-plugman.git#0.14.x",
>
> git clone cordova-cli
> edit package.json
> npm install
>
> no need to run "npm install" a second time.
>
> [1]: https://npmjs.org/doc/json.html#Git-URLs-as-Dependencies
>
> --Carlos
>
>
> On Tue, Oct 29, 2013 at 8:06 AM, David Kemp <drkemp@google.com> wrote:
>
> > There is an interesting issue with detecting change though.
> > The failure is detected easily because a commit to CLI triggered a build.
> > The only reason I knew it was fixed was because a commit to JS caused a
> > rebuild on the Master branches (which worked).
> >
> > Ideally, updating the plugman NPM module would have been detected an
> > triggered a build. I am not sure I can track that...
> >
> >
> > On Tue, Oct 29, 2013 at 7:56 AM, David Kemp <drkemp@google.com> wrote:
> >
> > >
> > > The CI always uses master cli and master plugman using this procedure:
> > >
> > > git clone CLI
> > > npm install (installs the 'wrong' plugman)
> > > delete the node-modules/plugman directory
> > > git clone plugman (into node-modules/plugman)
> > > npm install (plugman)
> > >
> > > When CLI refers to a plugman that does not exist, the first 'npm
> install'
> > > fails and aborts the test.
> > > If plugman was the last reference, I suppose that the error could
> > possibly
> > > be ignored and move on anyway, but I am not sure that the npm install
> > would
> > > really be complete even then (post install tasks?).
> > >
> > > If there is a better way to override the npm install behaviour, I would
> > be
> > > happy to give it a try.
> > >
> > > The commit that fixed the problem was about midnight, about an hour
> > before
> > > Michal checked it.
> > >
> > > David
> > >
> > >
> > > On Tue, Oct 29, 2013 at 1:10 AM, Michal Mocny <mmocny@chromium.org>
> > wrote:
> > >
> > >> Just pulled latest cli/plugman to check that the version numbers &
> deps,
> > >> and seems that they are.  So im guessing its a tooling version
> mismatch
> > on
> > >> the CI machine (using dev cli with released plugman).
> > >>
> > >> -Michal
> > >>
> > >>
> > >> On Tue, Oct 29, 2013 at 12:57 AM, Michal Mocny <mmocny@chromium.org>
> > >> wrote:
> > >>
> > >> > If you are using both versions off master, why are you getting that
> > >> error
> > >> > message?
> > >> >
> > >> > Seems it may happen if using master CLI and running npm install
> > without
> > >> > linking plugman first?
> > >> >
> > >> >
> > >> > On Mon, Oct 28, 2013 at 9:52 PM, Steven Gill <
> stevengill97@gmail.com
> > >> >wrote:
> > >> >
> > >> >> That is because I pushed plugman + cli to master but not to npm
> yet.
> > >> That
> > >> >> will go away right when they get published to npm. After some
more
> > >> views
> > >> >> on
> > >> >> the review of the blog post I will publish them.
> > >> >>
> > >> >>
> > >> >>
> > >> >> On Monday, October 28, 2013, David Kemp wrote:
> > >> >>
> > >> >> > Our CI is failing with the message:
> > >> >> >
> > >> >> > Error: No compatible version found: plugman@'>=0.14.0-
<0.15.0-'
> > >> >> > npm ERR!
> > >> >> >
> > >> >>
> > >> >
> > >> >
> > >>
> > >
> > >
> >
>
>
>
> --
> Carlos Santana
> <csantana23@gmail.com>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message