[tlbuild] No go.

[Martin Gieseking]

Dick,

that's great news. I'm relieved that it works now. The required header
file is /usr/include/xlocale.h which should be available even if X11 is
not. Xlocale.h with capital X seems to be something different.

The configure script should detect the availability of xlocale.h and
define HAVE_XLOCALE_H accordingly. Does your config.h from the latest
build contain #define HAVE_XLOCALE_H? If yes, it's probably
HAVE_CONFIG_H that's not defined. In the latter case it's sufficient to
drop the guards around #include <config.h>.

Martin


Am 03.02.2017 um 20:33 schrieb Richard M. Koch:
> Martin,
> 
> Surprisingly, that did it. Everything compiled and all tests were passed.
> 
> Go figure!
> 
> I just looked at /usr/include. This is a system location, containing many files including
> locale.h and xlocale.h. (The X11 installation is in /opt/X11 and includes its own xlocale.h).
> 
> The file locale.h does NOT define locale_t. The file xlocale.h DOES contain locale_t.
> So your explanation is probably correct, and the question is why HAVE_XLOCALE_H
> isn't set. At any rate, I'm happy.
> 
> Incidentally, on OS X, X11 in't included by default. It is provided by a third party open
> source project. They provide headers in /opt/X11/...  There is an Xlocale.h header,
> but it DOES NOT provide locale_t.
> 
> Maybe what happened is that code somewhere else loaded the X11 Xlocale.h, and
> your modification below finally got the Mac to load the system's copy of Xlocale.h?
> 
> Dick
> 
> 
> 
>> On Feb 3, 2017, at 10:48 AM, Martin Gieseking <martin.gieseking at uos.de> wrote:
>>
>> Dick,
>>
>> according to [1], xlocale.h should provide locale_t and the
>> corresponding functions on OS X. Could you check if the build succeeds
>> if you add
>>  #include <xlocale.h>
>> without the surrounding
>>  #ifdef HAVE_XLOCALE_H ... #endif
>> to libs/ff-woff/fontforge/splinefont.h? Maybe it's just the macro that
>> doesn't get defined for some reason.
>>
>> Martin
>>
>> [1] https://github.com/mgieseki/dvisvgm/issues/65
>>
>>
>> Am 03.02.2017 um 19:31 schrieb Richard M. Koch:
>>> Folks,
>>>
>>> Sorry. Removing my patch, but keeping Martin Gieseking's patches, solves
>>> nothing on OS X.
>>> The errors in the log file are listed below. I believe they are unchanged.
>>>
>>> Dick Koch
>>
>