The problem demonstrated in #6065 is that a string has incorrectly been
encoded with US-ASCII even though it contains invalid US-ASCII byte
sequences (any byte with the most significant bit on is invalid in the
US-ASCII encoding).
The thing about doing newline normalization is that it is not actually
sensitive to the presence of US-ASCII byte sequenzes. Additionally, it
is very unlikely that a user will ever be using an encoding where \r\n
is not encoded the same as it would be in ASCII.
This patch first tries the existing method of normalizing the newlines
in the provided script file, if that fails for any reason it force
encodes the string to ASCII-8BIT (which allows the most significant bit
to be on in any individual byte) and then performs the substitution in
that byte space.
Here we implement a naive solution to #5605 which catches the case that
a provided source contains an object which cannot be inspected, because
an object contained within in has an #inspect string that returns a
string that is incompatible with the encoding in
`Encoding.default_external` or a string which cannot be downcast to
7-bit ascii.
The Ruby VM implementation of "#inspect" implements this checking on
these lines of code: http://git.io/vZYNS. A Ruby level override of
this method does not cause this problem. For example:
```ruby
class Foo
def inspect
"😍".encode("UTF-16LE")
end
```
will not cause the problem, because that's a Ruby implementation and the
VM's checks don't occur.
However, if we have an Object which **does** use the VM implementation
of inspect, that contains an object that has an inspect string which
returns non-ascii, we encounter the bug. For example:
```ruby
class Bar
def inspect
"😍".encode("UTF-16LE")
end
end
class Foo
def initialize
@bar = Bar.new
end
end
Foo.new.inspect
```
Will cause the issue.
The solution this patch provides basically catches the encoding error
and inserts a string which attempts to help the user work out which
object was provided without blowing up. Most likely, this was caused
by a user having a weird encoding coming out of one of the sources
passed in, but without a full repro case, it's not clear whether a patch
should be applied to a different object in the system.
Closes#5605.
We gained a ton of improvemnts to WinRM error handling in
https://github.com/mitchellh/vagrant/pull/4943, but we also got one bug.
The new code raises an exception when `winrm_info` does not return right
away. This was preventing us from catching the retry/timout logic that's
meant to wait until boot_timeout for the WinRM communicator to be ready.
This restores the proper behavior by rescuing the WinRMNotReady
exception and continuing to retry until the surrounding timeout fires.
Windows offers no out-of-the-box rsync utility. By far, the most
commonly used external utilities for Windows rsync are built with the
GNU Cygwin libraries. The cost for this convenience is that rsync on
Windows has to be provided paths that begin “/cygdrive/c” rather than
“c:/“ like other Windows-API utilities. Compounding the situation,
rsync doesn’t create paths/to/sub/targets and so the vagrant plugin
code, when performing an rsync, is responsible for creating
intermediate directories in guest paths if there are any. Furthermore,
the mkdir utility in Windows is not another Cygwin utility like rsync
but the routine mkdir of Windows command.com. Therefore, while rsync
needs the /cygwin paths, mkdir uses the Windows paths. Later, the
chef_solo.rp provisioner running within the guest will expect to find
Windows-style paths in its solo.rb configuration file. Due to all this,
vagrant has to keep track of both the original, possibly dirty Windows
guest path and the cygwin-scrubbed guest path.
When the network's type is :dhcp, the :ip option is used to derive the DHCP server configuration, and it doesn't actually indicate the IP that will be received by the VM(s).
The parameter name in the VirtualBox provider source is nic_type, not nictype.
The typo is probably caused by the fact that the parameter maps to nictype args in the VirtualBox CLI.