(#14522) Don't read /proc/self/status on Solaris

[haus]

/proc/self/status isn't a standard file on Solaris and can't be read as such.
Attempting to read it when using ruby 1.9.3 produces errors such as 'invalid
byte sequence in US-ASCII'. According to
http://www.ruby-forum.com/topic/163804, /proc/self/status is a sparse file with
a map at /proc/self/map to indicate which parts of the file to read. As a
stopgap measure, this commit simply returns false on Solaris, rather than
trying to read the file.

[puppetcla]

CLA Signed by haus on 2011-07-19 21:00:00 -0700

[jeffmccune]

Haus and I looked at this together and it's not the File.open that's the problem, but rather the comparison of raw data inside of a string Ruby thinks is encoded as UTF-8 being compared against a regular expression that is also UTF-8. When this happens the invalid UTF-8 sequence in the raw data causes the exception.

I think we can filter the raw data and make this work in a general way if we do this:

txt = File.open("/proc/self/status", "rb").read
txt.encode!('UTF-8', 'UTF-8', :invalid => :replace)
return true if txt =~ /^(s_context|VxID):[[:blank:]]*[0-9]/

[joshcooper]

@jeffmccune @haus make sure to check if the encode! method is available, as it won't be in ruby 1.8

[jeffmccune]

summary: Merged into master as 2cadc7c. This should be released in Facter 1.7. Thanks again for the contribution!

-Jeff