Responder: Re: [PATCH 1/2] tools/make_unicode: Implement full Unicode character decomposition.

---- Huw Davies escribió ----
> On Tue, Apr 10, 2018 at 04:25:54PM -0500, Sergio Gómez Del Real wrote:
> > Signed-off-by: Sergio Gómez Del Real <sdelreal(a)codeweavers.com>
> > ---
> >   libs/port/mbtowc.c |  14 +-
> >   tools/make_unicode | 423 +++++++++++++++++++++++++++++++++++++++++++++--------
> >   2 files changed, 370 insertions(+), 67 deletions(-)
>
> What this patch is doing is to add a 'compat' flag to wine_decompose()
> and then always call that function with that flag set.  So essentially
> you've added a bunch of dead code (the compat not set case).
>
> The patch needs to make some material difference to the Wine's
> implementation of the Win32 api, preferably demonstrated by a test; it
> currently does not do that.
>
> Similar comments could be made about patch [2/2].
>
> Huw.

Yes, it effectively replaces wine_decompose with wine_unicode_decompose_string adding, among other things, the flag that tells it to use canonical or compatibility decomposition.

In these patches, only the existing code in mbtowc.c use it; it already was using wine_unicode. Because it uses only canonical decomposition, there is no case in these patches for compatibility.

The thing is that the third patch I have, which implements NormalizeString (and make full use of the mentioned flag) wouldn't work in Wine without first running the make_unicode script. I could send the whole patchset with this in mind.