This is another way of addressing the problem I attempted to fix with
f34c5ec926 earlier today.
Apparently, older versions of compiler plugin-registering packages such as
standard-minifier-css and templating-compiler still depend on older
versions of the meteor package, which may still use path.join to import
fibers/future. This can be fixed by republishing those packages, as I did
in 917b01ac5f, but I'd prefer not to
republish every compiler plugin package.
Fortunately, we can also solve the problem by being more tolerant in the
implementation of Npm.require, which is what this commit does.
Ever since Meteor 1.3 first introduced a module system based on something
other than `Npm.require`, we've continued throwing missing module
exceptions that refer to `Npm.depends` and/or `Npm.require`, even if the
developer called `require` or used an `import` declaration. This commit
fixes that, so that all missing module exceptions look like 'Cannot find
module "module/name"'.
I also noticed recently that `Npm.require` is capable of returning modules
installed in `node_modules` directories completely outside the app, which
is bad news for development/production reproducibility. Fixed that too.
CC @hwillson who has spoken of deprecating `Npm.require` entirely, and
just using `require` everywhere, instead.
This is performance-sensitive code, and I think this extra (uncached) call
to files.statOrNull could be contributing to test timeouts on Circle CI.
Follow-up to #9030.
As outlined in #3854, if `meteor --settings` points to a
directory instead of a file, an application gets caught up
in an infinite rebuild loop. Since it was assumed only a
file would be configured via `--settings`, a file watcher
is created for the specified directory, and that watcher
is then triggered ad infinitum.
These changes prevent a file watcher from being created
on a `--settings` directory.
If a login token is expired, or no longer valid, make sure that Meteor doesn't
just sit there at a login prompt that the user can't see.
This currently only applies if `headless` mode is enabled, though a follow-up
to this commit might consider setting `headless = true` automatically
in a CI environment using environment variables such as: `CI`, `TRAVIS`,
`JENKINS_URL`, etc., as the npm-registry-client does in a similar way.
See: https://github.com/npm/npm-registry-client/pull/129/filesFixes#8839.
The `child_process` documentation indicates that .cmd scripts (such as
the `npm.cmd` script for npm) must be started within a shell, such as
cmd.exe.
While it was tempting to switch this to use `child_process`'s `spawn`,
which supports a `shell` option, it would have required us to buffer
our own stdout/stderr output via `data` event handlers.
Ref: https://nodejs.org/api/child_process.html#child_process_spawning_bat_and_cmd_files_on_windows
This was causing problems with `npm@5`, but realistically it may not be
be necessary anymore since `npm@5` has a much smarter global,
self-healing `cacache`.
Should fix#8988 and #8942.
Now that #8866 is the default behavior, it can take up to 5000ms for
changes to files modified during the build process to be noticed.
Before #8866, when we called e.g. files.writeFile(path), a native file
watcher would notice the change immediately, almost always before the
build process read the file again. This was definitely racy, but we were
getting away with it consistently... until #8866.
I was able to reproduce the problem in #8988 by running
echo some-local-package-name >> .meteor/packages
in an app with a local package of the given name. After debugging the
endless rebuild cycle, I found that .meteor/versions was being rewritten
by files.writeFile during the build process, but the file watching system
was not noticing the change in time to prevent watch.isUpToDate from
returning true. The change was finally detected when restarting the
Watcher responsible for .meteor/versions, which of course triggered
another rebuild, so the same problem kept happening again and again.
