hppa2.0n-hp-hpux11.00 libstdc++-v3 locale fails

rbrown64@csc.com.au rbrown64@csc.com.au
Tue Feb 18 03:41:00 GMT 2003

Summarized locale -a output from some accessible machines.
Note that for aix4.2 and osf4.0e at least it appears that only a small set
installed by default.

C       UnixWare7.1.0 aix4.2 hpux{10.20,11.{00,11}} solaris2.5.{6,8}
POSIX   UnixWare7.1.0 aix4.2 hpux{10.20,11.{00,11}} osf4.0 solaris2.5.{6,8}

de      UnixWare7.1.0 solaris2.5.{6,8}
de_AT   UnixWare7.1.0 solaris2.5.{6,8}
de_CH   UnixWare7.1.0 solaris2.5.{6,8}
de_DE.ISO8859-1 sco5.0.2 solaris2.5.8
de_DE.UTF-8     UnixWare7.1.0 solaris2.5.8
de_DE.iso88591  hpux{10.20,11.{00,11}}
de_DE.iso885915@euro    hpux{10.20,11.{00,11}}
de_DE.utf8      hpux{10.20,11.{00,11}}

en_AU   UnixWare7.1.0 solaris2.5.{6,8}
en_CA   UnixWare7.1.0 solaris2.5.{6,8}
en_GB.iso88591  hpux{10.20,11.{00,11}}
en_GB.iso885915@euro    hpux{10.20,11.{00,11}}
en_IE   solaris2.5.{6,8}
en_NZ   UnixWare7.1.0 solaris2.5.{6,8}
en_US   UnixWare7.1.0 aix4.2 solaris2.5.{6,8}
en_US.iso88591  hpux{10.20,11.{00,11}}

If ja_JP.eucjp was chosen to exercise double byte character set support,
maybe de_DE.{UTF-8,utf8} would provide some testing so


if Irix supports one of the UTF-8 varients, otherwise maybe add a standard
ISO-8859-1 == Latin1 based locale.

Push the list of locales to try into
and lose the explicit locale from run_tests_wrapped_locale? Or put in a
const char *locale_list[] = ... into testsuite_hooks?

David Billingshurst wrote:
We have a similar problem with irix.

-----Original Message-----
From: John David Anglin
Sent: Tuesday, 18 February 2003 6:03 AM
To: Rodney Brown
Subject: Re: hppa2.0n-hp-hpux11.00 libstdc++-v3 locale fails

> On hppa2.0n-hp-hppa11.00 and maybe Dave Anglin's hppa64-hp-hpux11.11
> the wrapped_locale.cc execution tests fail because there is no=20
> ja_JP.eucjp locale

Yes.  I also saw some locale test fails because we don't have setenv.

> on the machine. The following patch executes ok - for one test.
> Should libstdc++-v3/testsuite/testsuite_hooks.cc:run_tests_wrapped_locale
>  take
> a list of locales to try, falling back to C or POSIX so tests don't
>  fail from a missing or uninstalled locale?

Seems like a reasonable idea if such a list can be defined.


More information about the Gcc-bugs mailing list