Re: [lmi] 'wx/testing.h' not installed?

[Greg Chicares]

On 2014-03-13 17:28Z, Vadim Zeitlin wrote:
> On Thu, 13 Mar 2014 17:23:17 +0000 Greg Chicares <address@hidden> wrote:
> 
> GC> [...] I'm sure I'll need 'Makefile.in', and I'm sure I can
> GC> ignore the compiler-specific makefiles in 'build/msw';
> GC> but do I need 'build/bakefiles/files.bkl'?
> 
>  No, you don't need it, this is the meta-makefile from which the real
> makefiles are generated (by bakefile), but the makefiles are standalone and
> don't depend on files.bkl in any way. So all you really need is indeed just
> 
> diff --git a/Makefile.in b/Makefile.in
> index f1cdfbd..57a01ea 100644
> --- a/Makefile.in
> +++ b/Makefile.in
> @@ -4137,6 +4137,7 @@ COND_USE_GUI_1_ALL_GUI_HEADERS =  \
>         wx/panel.h \
[...]

Wow, that simple patch took me a lot of effort to apply. I tried
just about every 'patch' option except the ones that ignore
whitespace; but somewhere in the chain of email programs the
sequence <sp> <ht> got turned into a series of spaces with no tab.
But now I know one more potential problem to look out for. There's
nothing you could have done to prevent this, of course.

More puzzlingly, I downloaded a context diff from trac.wxwidgets.org
but couldn't get it to apply, even though it correctly distinguishes
<ht> from <sp>.

Suspecting a problem with 'patch', I changed the file with 'sed'...
successfully, once I figured out the <ht> issue; and, consequently...

> (and, well, the first hunk of this patch is not strictly speaking
> indispensable as this header is not used by LMI, but it doesn't hurt
> neither so I'd take it too).

...I didn't apply that hunk (which adds 'wx/preferences.h') because
I didn't want to push my luck too much after I had gotten the
crucial 'testing.h' part to work.