[Xfce4-commits] CVS: xfce4/xfdesktop/src xfce-desktop.c, 1.24, 1.24.2.1

[Benedikt Meurer]

Brian J. Tarricone wrote:
> Benedikt Meurer wrote:
> 
>> Update of /var/cvs/xfce/xfce4/xfdesktop/src
>> In directory espresso.foo-projects.org:/tmp/cvs-serv10094/src
>>
>> Modified Files:
>>      Tag: xfce_4_2
>>     xfce-desktop.c Log Message:
>> Remove all occurences of G_INLINE_FUNC, as it causes too much trouble on
>> Solaris, which offers quite a lot of different compiler/linker combos.
>>
> Do you know how it behaves if you just change it to 'static inline'?  I 
> recently learned that G_INLINE_FUNC actually does some weird stuff 
> depending on how you compile the app.  In fact, before your commit, 
> xfdesktop probably failed to compile at all with --enable-debug and glib 
> 2.6.1 and above, because the glib devs decided to change something.
> 
> Though it appears that the inline keyword was only standardised in C99.  
> *sigh*  Though the gcc guide notes that 'static inline' is the portable 
> subset of using the inline keyword, so it might still be ok.
> 
> What's the usual method for checking what version of the C standard a 
> compiler supports?  __STDC__?
> 
> I think it's pretty sad that we can't rely on a standard finalised 6 
> years ago.

Forget about the inline keyword, its not important at all today. Modern 
compilers do better to decide which function to inline and which not, so 
its not worth the effort.

And this causes really a lot of trouble (esp. with Solaris, but also 
with Linux and older binutils), so I decided to ditch all occurences of 
G_INLINE_FUNC from xfce_4_2. In HEAD, xfdesktop is the only component to 
use G_INLINE_FUNC currently; I didn't change it there, because I wasn't 
sure if we want to maintain the 4.2 menu code for a while. Brian? If 
yes, we need to forward-port that fix.

>    -b

greets,
Benedikt