[prev in list] [next in list] [prev in thread] [next in thread] 

List:       koffice
Subject:    Re: OpenParts - Unicode
From:       Simon Hausmann <tronical () gmx ! net>
Date:       1999-04-12 17:01:28
[Download RAW message or body]


On Mon, 12 Apr 1999, Warwick Allison wrote:

> On Sat, 10 Apr 1999, you wrote:
> >Hi Warwick,
> >can you comment on the Unicode conversion question please ?
> >
> >Bye
> >Torben
> >
> >
> >On Sat, 10 Apr 1999, Simon Hausmann wrote:
> >
> >> Hello,
> >> 
> >> I just took a few thoughts about making OpenParts Unicode compliant.
> >> 
> >> Currently GUI related strings in OpenParts (example: part captions, menu
> >> names, and similar stuff ) are defined as "char *" based strings (or in
> >> IDL as "string" ) . This seems not to be "friendly" in regard to Qt's new
> >> QString and that's why I think this should be fixed in OpenParts soon,
> >> especially in regard to the fact that more and more people seem to get
> >> interested in using KOM/OpenParts in their applications.
> >> 
> >> The way how to actually do the conversion is quite clear:
> >> (1) convert "string" into "wstring" in the interface descriptions (where
> >>     it makes sense)
> >> (2) in the implementations convert "const char *" and "char *" both to
> >>     "CORBA::WChar *" where necessary
> >> (3) use CORBA::WString instead of CORBA::String where necessary
> >> (4) provide two smart conversion routines between "CORBA::WChar *" and
> >>     QString
> >> (5) fix a lot of method calls in all applications (...and that's the real
> >>     work)
> >> 
> >> 
> >> About 4) : I wrote two conversion routines and put them for testing
> >> purposes into konq_txtview.cc in Konqueror, they might be useful. But
> >> since I'm no QString<->Unicode expert I'm not quite sure whether the
> >> actual conversion in these routines is correct (but they work fine here:).
> >> Perhaps some Unicode expert could check this?
> 
> Looks correct to me, except an off-by-1:
> 
>   CORBA::WChar *res = CORBA::wstring_alloc( s.length() );
> 	...
>   *tmp = 0;
> 
> No space was allocated for the 0 terminator.

I think it _is_ allocated.

In CORBA::wstring_alloc( len ) the allocation is done by
wchar_t *s = new wchar_t[len+1];

So I think the terminator space is included.

> You probably should also cast CORBA::WChar to ushort to be clear which QChar
> constructor is in use:
> 
>     *(tmp++) = QChar( (ushort) *(s++) );

Ah, yes, that's true. Thanks for the hint.
 
> There may be a slightly more efficient way to implement C2Q, but worry about
> that later.

Ok :-)
 
> The whole idea is certainly important - I don't know what the alternative would
> be - to assume that all Parts are in the same locale (giving the same encoding
> for char*) seems a bit bogus.

Hm, ...
Where/How is the encoding in Qt set, how does Qt find out about the
used locale?
If the information would come from the X-Server directly then I think it
might work since the server/display is shared for all OpenParts objects
=> same encoding, or?

Since I'm not familiar with QString internals: What could this help us?
Does this mean we could still use the good old char *'s in this case?


Greetings,
 Simon
 
> --
> Warwick
> 

[prev in list] [next in list] [prev in thread] [next in thread]