From: Daniel Glassey (danglassey@ntlworld.com)
Date: Thu May 22 2003 - 06:09:52 EDT
On Thursday 22 May 2003 3:09 am, Andrew Dunbar wrote:
> Hi Daniel.
> This is much needed work, but I'm very confused by the
> code I've seen so far.
> You seem to be making explicit calls to the xxxA or
> xxxW functions.
> If we are now *dependent* on unicows, shouldn't we
> just define the preprocessor symbol UNICODE and call
> the unadorned xxx functions? This way we can have both
> ANSI and UNICODE builds until the UNICODE build is
> stable enough to take over. It will also result in
> better code IMHO.
The reason I haven't gone for a full unicode build to start with is that all
the gui resource strings are char* but in a unicode build they must be passed
in as UCS2.
I started to try doing that but it got too messy and obscured the initial
problem I was trying to deal with (unicode input with Keyman).
> In other places, including my own code, we have
> explicit xxxA or xxxW calls, but in these cases we
> check at runtime for the presence of the xxxW calls
> so we avoid the unicows dependency.
Could you give me an example of that please - I didn't see them? (if it
doesn't end up being messy it would be worth doing that for these bits as
well)
> It's my opinion that the combination of the
> dependency,
> the loss of the ANSI-only build, and the explicit
> xxxA / xxxW function calls is the least elegant way
> to go about Unicodizing AbiWord.
The reason for the explicit xxxW calls is so that unicode input can be
accepted even on a predominantly ANSI build.
It is also only a step along the path to a full UNICODE build rather than
being the endpoint,
> I would like to hear your and others' opinions.
Regards,
Daniel
This archive was generated by hypermail 2.1.4 : Thu May 22 2003 - 06:25:57 EDT