[libstdc++ testsuite][patch] many locale tests only SUPPORTED on linux, start making these portable

[Jonathan Wakely]

On 16/10/15 11:21 +0200, John Marino wrote:
>There are a few issues with the locales portion of the libstdc++
>testsuite.  These issues stem from the fact that much about locales are
>not standardized, such as the names of the encodings, how to handle
>abbreviated versions of named locales, and modifiers (e.g. @euro,
>@preeuro).  Compounding the problem is that many of test themselves are
>Linux specific.
>
>The attached large patch makes a dent with an effort to resolve this
>issues.  This patch has been scoped to only standardise the names of the
>locales.  There is no net effect on Linux, but it allows many more tests
>to be considered on other systems, including *BSD.
>
>Some types of changes are:
>1) convert abreviations to full names (e.g. "de_DE" => "de_DE@ISO8859-15
>2) Use ISO-8859-15 over ISO-8859-1 in European locales when possible
>3) Use "UTF-8" over variations like "utf8"
>4) Use "ISO8859" over "ISO-8859"
>
>on these last two, Linux is case-insensitive and hyphen-insensitive with
>respect to named locales where other systems like BSD have only one
>version that must be explicitly used.  Since encoding names are not
>standardised, we've picked a lowest common denominator that works.
>
>5) case changes for eucJP
>6) convert tests that specified "@euro" modifier to a new ISO_8859 macro
>can produces a named locale as a function of system (currently
>FreeBSD/DragonFly/NetBSD and everything else but the tailoring is not
>limited).  This fixes systems that don't support @euro modifier
>
>There is a lot more work that cant be done to open up libstdc++ locale
>tests to non-Linux systems, but given the size of this patch, I wanted
>to limit it's scope to only modifying to the name locale specification.
>There are many more modifications that could be made to the test
>themselves to make them portable to non-linux systems but I think they
>are better made with separate patches.
>
>Andreas Tobler has done a lot of review and testing on both FreeBSD and
>linux (to verify no regressions), and Jonathan Wakely has done an
>initial review and said it was ready for a broader review now.

I asked some colleagues to look at this too (a few weeks ago now,
before I went on vacation).

The only significant comment was:

e.g. `"de_DE" => "de_DE@ISO8859-15` should be `e.g. "de_DE" => "de_DE.ISO8859-15"` ? Since it's an encoding, @ is used for a modifier.
FYI, https://www.ietf.org/rfc/rfc4646.txt