Lars Gullik Bjønnes wrote:

> Why cannot the converson functions be allowd to stay in one place?
> (unicode.Ch)
> After all we are not supposed to use them all the time.

They can also go to unicode.[Ch]. I did not do that because that would mean
to add lots of #include <unicode.h> during the transition process that will
vanish afterwards.
I did neither use the names utf8_to_ucs4 etc, because I want to be able to
easily grep for the temporary functions.
If you prefer I can put them to unicode.[Ch], but then I would like to
#include <unicode.h> in docstring.h for the transition period in order to
save some typing.

> | Then I changed the type of FuncRequest::argument to docstring, made it
> | private (as it always should have been) and added an accessor.
> | Then I added a lot of lyx::to_utf8() to fix the compile errors. Of
> | course we do now have a lot more conversions than before, but these will
> | either go away when more strings are converted to docstring, or can be
> | changed to lyx::to_ascii() when we know that only ascii contents is
> | stored.
> 
> Why/Where do we need to convert _to_ utf8?

When we are interfacing with parts of the code that are not converted to
docstring yet, e.g. MathNestInset::interpret(). As i wrote: I expect that
almost all of these calls will go eventually.

> | I will put this in tomorrow unless I get objections.
> 
> I'll look it over later tonight.

Thanks.


Georg

Reply via email to