utf8 flag missing

[eserte]

At least the German area names are not properly encoded. Here's a sample (Donauwörth):

$ perl -MDevel::Peek -MNumber::Phone -e 'Dump(Number::Phone->new("+49 906 1234567")->areaname)'  
SV = PV(0x2565280) at 0x242ca10
  REFCNT = 1
  FLAGS = (TEMP,POK,pPOK)
  PV = 0x2121c40 "Donauw\303\266rth"\0
  CUR = 11
  LEN = 16

Correct output should be:

  ...
  FLAGS = (POK,pPOK,UTF8)
  PV = 0x11315c0 "Donauw\303\266rth"\0 [UTF8 "Donauw\x{f6}rth"]
  ...

[DrHyde]

Thanks. Does 5305eac fix it? It looks like it does to me, but I don't really understand all these Unicode shenanigans so shouldn't be trusted!

To test it you'll need to run build-data.sh and have the following additional modules installed:

• XML::XPath
• Digest::MD5
• File::Find::Rule
• Spreadsheet::ParseExcel
• Text::CSV_XS

and the machine running the build script needs git and curl installed and access to the interwebnets, possibly some other stuff that I've forgotten too :-)

[eserte]

Yes, it works. Another possibility (probably slightly more efficient) would be to use utf8; and remove the backslashes before high bit characters, and no map { ...Encode::decode... }.

Another observation: with newest ExtUtils::MakeMaker value of the PERL variable in the generated Makefile is always quoted (even on Unix systems), so the last command in the build script fails:

./build-data.sh: line 110: "/usr/perl5.20.2p/bin/perl": No such file or directory

[DrHyde]

That's odd. Does /usr/perl5.20.2p/bin/perl exist on that machine? Because having extraneous quotes around an executable name shouldn't affect anything ...

$ "/usr/bin/which" which
/usr/bin/which
$ /usr/bin/which which
/usr/bin/which

even in a script ...

$ cat foo
#!/bin/sh
/usr/bin/which which
"/usr/bin/which" which
$ ./foo
/usr/bin/which
/usr/bin/which

[eserte]

But in backticks it's different:

$ `echo '/usr/bin/which' which`
/usr/bin/which
$ `echo '"/usr/bin/which"' which`
bash: "/usr/bin/which": No such file or directory

(Linux and FreeBSD, bash and zsh)

[DrHyde]

I fucking hate Unix sometimes!

[DrHyde]

Should be fixed in 3865fda. Please test!

[eserte]

Tested 62b2ac5 --- the Makefile re-build problem is fixed, and the changed implementation with use utf8 looks good.