Re: unicode strings

[Andrew Gaylard]

Gadi Bergman wrote:
> I was thinking about two different solutions:
>
> 1. Very trivial solution for taking care of Western European languages
> only:
> Arrays of 'unsigned long' and 'unsigned short' will be presentable to
> the user as strings by writing each element that is lower than 0x100 as
> a char with that value and other elements as '?'. For this solution, DDD
> assumes Western European single-byte character encoding of the string.

This is quite easy to do, especially if one assumes ISO-8859-1.

> 2. Very complex solution for taking care of all languages:
> Arrays of 'unsigned long' and 'unsigned short' will be presentable to
> the user as strings by drawing each character using a Unicode font on
> the DDD display section. For this solution, DDD does not assume any
> multi-byte or single-byte encoding and may display a string composed of
> any number of languages.

This proposal doesn't make sense without an encoding.  Remember that
the entire Unicode set won't fit into 16 bits, hence the need for
variable-length characters, as seen in UTF8.  So an encoding is
required to know *which* subset of Unicode a short (i.e. 16-bit)
array entry should index.

> I do not recommend adding a dialog box for asking the user to specify an
> encoding, certainly not for every string the user wishes to display.

Yes, certainly one wouldn't like to have to choose an encoding every
time a string is viewed.  I foresee a scheme where DDD uses $LANG,
unless it is overridden by the user.  All subsequent string
viewing/displaying would then be subject to the specified enconding.

These proposals both depend on the X-server having the appropiate
fonts available, either locally or via a fontserver (which isn't too
hard to do and often works "out-of-the-box" nowadays).  They also
depend on Motif *actually* *using* the correct language as specified by
$LANG, something which is probably very simple but which has eluded
me thus far.

Any pointers on this aspect would be *VERY* much appreciated!  I'm
tearing my hair out...

Andrew

> -----Original Message-----
> From: Andrew Gaylard [mailto:address@hidden 
> Sent: Monday, May 30, 2005 3:09 PM
> To: Gadi Bergman
> Cc: address@hidden
> Subject: Re: unicode strings
>
> Gadi Bergman wrote:
>
> > Hello Andrew and thank you for your very quick response,
> >
> > I do not expect the protocol for X to be Unicode-based. However,
> > programs written in Java do display their characters correctly for all
> > languages when running under X. I assume that they are drawing their
>
> own
>
> > characters with built-in fonts. That would be hard to expect from
> > text-based GDB, but since DDD is a graphic application then it is much
> > more natural.
> >
> > Thanks again,
> > Gadi.
>
>
> OK, then what's needed is a dialog box for DDD to specify the encoding,
> and a button in "examine memory" which will unpack the bytes returned by
> gdb in the encoding specified.  Yeah, that would do the trick.
>
> Feel like contributing a patch?