For weeks now I've been working with the CLI as it goes from version to version
and a few weeks back a problem arose where the lazy load of the default www
project fails or the lazy load of the Android or iOS project. I've documented
this repeatedly, my most recent JIRA ticket is here:
https://issues.apache.org/jira/browse/CB-4322.
What's happening is that the CLI is trying to download an archive for some of
the files it needs, but fails somehow. Using -d I can tell a little more about
what is happening, but it is not always clear why its failing. I'm not sure if
the download is not happening or the extraction of the download is failing, but
the end result is that the CLI thinks it already has the files it needs (when
it actually doesn't) because the folder it's looking for already exists.
The CLI needs to report whether or not it is able to download the files and/or
report that it's failing on the extraction - then act accordingly. Checking to
see if the folder exists is not a valid check in this case as if the lazy load
fails, the folder the CLI is checking for still exists - which breaks the
process and leaves the CLI in an unworkable state. The target folder should not
be created (or it should be deleted on fail) unless the files have been
extracted to it.
This particular problem is affecting my entire team, there are many of us here
being affected by this. I thought at first that this must be a proxy problem,
but I've now gotten back to my home office and I'm experiencing the exact same
problem on my Mac Mini which isn't using a proxy and has no proxy settings.
I've experienced this problem on our company network as well as in two
different hotel networks and now my home office network. This is with three
different computers on 4 different networks.
On my MacBook and my windows laptop, I've configured npm and git with the
appropriate proxy settings for my work network to no avail. With or without the
settings, it fails.
Two of my colleagues (in England and Germany) are experiencing the same
problems; they're on different networks and have different proxy servers. They
were able to get around this by manually forcing the proxy setting in the
lazyload.js file:
request.get({uri:url, proxy:'http://some_proxy_server:8080'}, function(err,
req, body) { size = body.length; })
I would prefer to not have to modify the code to make this work. Unfortunately,
the end result is our developers are ready to give up on the CLI since we can't
make it work reliably simply for creating new projects or adding platforms to
existing projects (both use lazy load).
Is there a recommended fix for this? Anything I can do to help document this
better?
John M. Wargo
Twitter: @johnwargo
