Fix "make doc" error: Character set messup in pdf-scheme.cc

Fix "make doc" error: Character set messup in pdf-scheme.cc

In "make doc", somehow the LANG env variable is cleared, so
glib returned charset=ANSI_X3.4-1968 for the input string. This would
disallow any accented characters in the input string.
However, we know that the input string is by definition always UTF-8,
so the solution is not to use the current locale, but hardcode UTF-8.

[janneke, 2011-04-26 13:19:01]

LGTM

Some nitpicks, see below

http://codereview.appspot.com/4449061/diff/1/lily/pdf-scheme.cc
File lily/pdf-scheme.cc (left):

http://codereview.appspot.com/4449061/diff/1/lily/pdf-scheme.cc#oldcode41
lily/pdf-scheme.cc:41: if (!g) {
Please use gnu coding standards for brace placement

http://codereview.appspot.com/4449061/diff/1/lily/pdf-scheme.cc
File lily/pdf-scheme.cc (right):

http://codereview.appspot.com/4449061/diff/1/lily/pdf-scheme.cc#newcode36
lily/pdf-scheme.cc:36: const char *charset="UTF-8"; // Input is ALWAYS UTF-8!
I think we never start with const: ie use char const * here. Then const always
applies to the left.

http://codereview.appspot.com/4449061/diff/1/lily/pdf-scheme.cc#newcode52
lily/pdf-scheme.cc:52: if (g_without_BOM) { // conversion to UTF-16be might have
failed, too
what about: you avoid casting and add an implicit comment about what exacly is a
bom.
char const* BOM = "\xff\xfe";
strcpy (g, BOM);

[Reinhold, 2011-04-26 20:47:43]

On 2011/04/26 13:19:01, jan.nieuwenhuizen wrote:
> LGTM
> 
> Some nitpicks, see below

New patch is up, which includes all of Jan's nitpicks (in particular, GNU coding
standard for braces, char const*, BOM comment and definition as \xFE\xFF). I
also fixed a memleak by using scm_take_str instead of scm_from_locale_stringn
(which does NOT eventually free its first argument). This also makes more sense
as scm_take_str is meant for arbitrary binary data, while
scm_from_locale_stringn might mess up when the locale handling changes in GUILE.

Cheers,
Reinhold

[Graham Percival (old account), 2011-04-28 12:20:11]

LTGM, please push now.

[Neil Puttock, 2011-04-28 14:14:11]

http://codereview.appspot.com/4449061/diff/4001/lily/pdf-scheme.cc
File lily/pdf-scheme.cc (right):

http://codereview.appspot.com/4449061/diff/4001/lily/pdf-scheme.cc#newcode30
lily/pdf-scheme.cc:30: "Encode the given string to either Latin1 (which is a
subset of "
move space to start of next string

http://codereview.appspot.com/4449061/diff/4001/lily/pdf-scheme.cc#newcode31
lily/pdf-scheme.cc:31: "the PDFDocEncoding) or if that's not possible to full
UTF-16BE "
move space to start of next string

http://codereview.appspot.com/4449061/diff/4001/lily/pdf-scheme.cc#newcode37
lily/pdf-scheme.cc:37: char const*charset="UTF-8"; // Input is ALWAYS UTF-8!
const *charset =

http://codereview.appspot.com/4449061/diff/4001/lily/pdf-scheme.cc#newcode51
lily/pdf-scheme.cc:51: warning (_f("Conversion of string \'%s\' to UTF-16be
failed: %s", p, e->message));
(_f ("Conversion of string `%s'

http://codereview.appspot.com/4449061/diff/4001/lily/pdf-scheme.cc#newcode74
lily/pdf-scheme.cc:74: {
remove { }

http://codereview.appspot.com/4449061/diff/4001/lily/pdf-scheme.cc#newcode78
lily/pdf-scheme.cc:78: {
remove { }

[Carl, 2011-04-30 14:40:06]

LGTM, with Neil's comments implemented.