On Tuesday, August 12, 2003, at 4:41 PM, spamcontrol2@xxxxxxxxxxx wrote:
I've noticed that starting with 0.9.14 I have problems with captures
occasionally. Packet summaries do not appear in the Info column, and
in a
separate window I see the following errors:
(ethereal.exe:1328): Gdk-WARNING **: gdk_text_size: gdk_nmbstowchar_ts
failed
(ethereal.exe:1328): Gdk-WARNING **: gdk_win32_draw_text:
gdk_nmbstowchar_ts failed
One set of these messages seem to appear per packet that has a problem.
I've only noticed this in Windows (Win2k and WinXP, and on multiple
systems),
and do not have this problem with 0.9.13. I know that I don't have
these
problems in Linux with 0.9.14, at least with GTK 1.2.10.
I've only noticed this with SMB packets, and I believe it's with all
packets
that contain strings with extended characters (in one case, 0x00e4)
that would
appear in the Info column. These same strings appear to display in
the tree,
except that the suspected characters do not display.
GTK+ 1.2[.x]'s routines for displaying strings take, as I remember,
strings in whatever the encoding is for the font being used - even if
it's a 2-byte encoding. Ethereal tries to get only ISO 8859-1-encoded
fonts, so that it doesn't have problems with 2-byte encodings; this
means strings containing characters other than 8859-1 encodings don't
display well.
GTK+ 1.3[.x], at least on Windows, and GTK+ 2.x, on all platforms, take
strings in UTF-8. This causes strings that aren't UTF-8-encoded not to
display well - and SMB doesn't use UTF-8, it uses the native character
set ("code page") or Unicode.
We should probably eventually have Ethereal use UTF-8, UCS-2, or UCS-4
internally (hopefully they're rich enough to handle all the various
weird PC encode pages and Mac character sets - I think they're rich
enough to handle all the 8859-n character sets, although CJK
unification might cause problems when using various East Asian
character sets, and I don't know if they're rich enough for PC or
pre-OSX Mac character sets), and attempt to do whatever is necessary to
display correctly with GTK+ 1.2[.x] and the UTF-8 GTKs, and to do
something appropriate for printing to printers and to text files (for
text files on UNIX it should probably use the character set specified
by the LANG or appropriate LC_ environment variable; I'm not sure
what's appropriate on Windows - 2-byte Unicode text files, such as
NetMon 2.x produces, or something else?).
