On Mon, May 05, 2014 at 10:57:37AM -0300, Antonio Terceiro wrote:
> On Sun, May 04, 2014 at 05:00:58PM +0200, David Suárez wrote:
> > Hi,
> >
> > El Domingo, 4 de mayo de 2014 16:06:50 Christian Hofstaedtler escribió:
> > > Upstream said something about LE vs. BE:
> > > https://github.com/sparklemotion/sqlite3-ruby/issues/128
> >
> > If you look at s390x build log:
> >
> > > Expected: "foo"
> > > Actual: "\u6600\u6F00\u6F00"
> >
> > and then, reverse the bytes on each unicode codepoint like;
> >
> > \u0066\u006F\u006F
> >
> > you get: foo
> >
> > > I currently don't have the time to look into this any deeper, so
> > > this is a call for help.
> >
> > Maybe something like:
> >
> > ++ test_encoding.rb
> > - db = SQLite3::Database.new(':memory:'.encode('UTF-16LE'))
> > + utf16 = ([1].pack("I") == [1].pack("N")) ? "UTF-16BE" : "UTF-16LE"
> > + db = SQLite3::Database.new(':memory:'.encode(utf16))
> >
> > ++statement.c
> > if (UTF16_LE_P(value) || UTF16_BE_P(value)) {
> >
> > But I'm unable to test it on my i386 machine :)
>
> I tested these changes on s390x and they don't fix the problem :-(
After some extra modifications I got it to work and I am uploading a
fixed package. Thanks!
--
Antonio Terceiro <terceiro@debian.org>
Attachment:
signature.asc
Description: Digital signature