Build failure in nsterm.m

Build failure in nsterm.m

[martin rudalics]

Building master configured as

"CFLAGS='-O0 -g3 -no-pie -std=gnu99' ../configure --with-gnutls=no --without-pop --without-harfbuzz --with-ns"

currently fails here with

../../src/nsterm.m: In function ‘ns_relocate’:
../../src/nsterm.m:514:54: error: stray ‘@’ in program
    NSString *fixedPath = [NSString pathWithComponents:@[root, original]];
                                                       ^
../../src/nsterm.m:514:60: warning: left-hand operand of comma expression has no effect [-Wunused-value]
    NSString *fixedPath = [NSString pathWithComponents:@[root, original]];
                                                             ^
../../src/nsterm.m:514:70: error: expected ‘:’ before ‘]’ token
    NSString *fixedPath = [NSString pathWithComponents:@[root, original]];
                                                                       ^
                                                                       :

Help appreciated, martin

Re: Build failure in nsterm.m

[Alan Third]

On Tue, Jun 29, 2021 at 10:28:42AM +0200, martin rudalics wrote:
> Building master configured as
> 
> "CFLAGS='-O0 -g3 -no-pie -std=gnu99' ../configure --with-gnutls=no --without-pop --without-harfbuzz --with-ns"
> 
> currently fails here with
> 
> ../../src/nsterm.m: In function ‘ns_relocate’:
> ../../src/nsterm.m:514:54: error: stray ‘@’ in program
>    NSString *fixedPath = [NSString pathWithComponents:@[root, original]];
>                                                       ^
> ../../src/nsterm.m:514:60: warning: left-hand operand of comma expression has no effect [-Wunused-value]
>    NSString *fixedPath = [NSString pathWithComponents:@[root, original]];
>                                                             ^
> ../../src/nsterm.m:514:70: error: expected ‘:’ before ‘]’ token
>    NSString *fixedPath = [NSString pathWithComponents:@[root, original]];
>                                                                       ^
>                                                                       :
> 
> Help appreciated, martin

I've pushed a fix. I think this is a shorthand way of writing an ObjC
array that GCC doesn't understand.

I'm using a version of GNUstep built from git and I can no longer
build Emacs with GCC at all, only clang, so these sorts of things slip
through quite easily.

I'm not sure why GCC doesn't work, I'm sure it's not the Emacs code
that's at fault, so I have to assume that the GNUstep devs have given
up waiting for GCC to catch up.
-- 
Alan Third

Re: Build failure in nsterm.m

[martin rudalics]

 > I've pushed a fix. I think this is a shorthand way of writing an ObjC
 > array that GCC doesn't understand.

It's fixed now.

 > I'm using a version of GNUstep built from git and I can no longer
 > build Emacs with GCC at all,

In which sense does it fail?

 > only clang, so these sorts of things slip
 > through quite easily.
 >
 > I'm not sure why GCC doesn't work, I'm sure it's not the Emacs code
 > that's at fault, so I have to assume that the GNUstep devs have given
 > up waiting for GCC to catch up.

Maybe we should drop a note about this somewhere.

Thanks, martin

Re: Build failure in nsterm.m

[Alan Third]

On Wed, Jun 30, 2021 at 09:38:16AM +0200, martin rudalics wrote:
> > I've pushed a fix. I think this is a shorthand way of writing an ObjC
> > array that GCC doesn't understand.
> 
> It's fixed now.

Thanks!

> > I'm using a version of GNUstep built from git and I can no longer
> > build Emacs with GCC at all,
> 
> In which sense does it fail?

configure fails to find GNUstep requirements. There's a script to
source which sets up the environment for GNUstep builds, but it makes
no difference. The only thing that allows it to continue is to use
"CC=clang".

I imagine it's my build environment, but I followed the GNUstep
instructions from their wiki and I can't work out how to get it to
work.

I should probably ask them.

> > only clang, so these sorts of things slip
> > through quite easily.
> >
> > I'm not sure why GCC doesn't work, I'm sure it's not the Emacs code
> > that's at fault, so I have to assume that the GNUstep devs have given
> > up waiting for GCC to catch up.
> 
> Maybe we should drop a note about this somewhere.

I'd rather be sure it's not just me being useless before making any
official notice.
-- 
Alan Third