Re: [Enigma-devel] Enigma universal build

[Ronald Lamprecht]

Hi Johannes,

Johannes Fortmann wrote:
> > I commited the patch with r150.
> >
> > Additionally I moved the Mac OS X system location detection code from
> > main.cc to ecl_sys_localename.cc. Could you please check and test this
> > change.
>
> Won't work:
> - it still uses string, even though that's not included. Well, that's
> no problem, but
> - it allocates a string on the stack, then returns a pointer to its
> c_str(). I'm pretty sure that means instant death. Personally, I'd
> revise the whole function to return an std::string; however, since my
> build system seems hosed just now, I won't get to do that 

You are right. I first moved the Mac code to

std::string ecl::DefaultMessageLocale ()

where we have no problem with the return value. Then I decided to drop
one of the two methods and moved the Mac code to

const char * sys_message_locale_name ()

After a break I forgot to drop DefaultMessageLocale() and to adjust the
return value of  sys_message_locale_name(). The fix will follow with my
next patch.

We will still need the fix of the includes.

BTW you wrote:
> - I had a problem with running texi2html. Mine chokes on -output=.
> which causes the build to fail

Your patch causes the output to be put in a subdirectory. This causes a 
minor incompatibility with the current install/distribution procedures 
for Linux  and Windows. Thus I would first like to ask you to check your 
texi2html and to analyse the reasons for the failure. Version 1.76 is 
required and should work. I will temporarily revert this single patch 
until we know the reasons of your problem.

> (for the next week, since I've got my exams and have to learn)...

Take your time with the fixes.

- Ronald