Re: GTK2 GdkNativeWindow compilation errors

Okay, with the patch, 9.0.10 from git compiles fine on rocky 9.5, fedora 41=
 won't detect gtk2 properly so doesn't compile the gtk2 module, with or wit=
hout the patch.

Might be compiler, might be some other things.

I'm going to dig into the config.log for gtk2 module and compare.




On Friday, April 18, 2025 at 09:03:24 AM EDT, Lance Dillon <riffraff169@yah=
oo.com> wrote:=20





But only on fedora 41. Rocky 9 has gtk2 fine. hmm...






On Friday, April 18, 2025 at 08:50:06 AM EDT, Lance Dillon <riffraff169@yah=
oo.com> wrote:=20





Something weird. Pike 8.0.1916 compiles gtk2 fine, but 9.0.10 from git does=
n't, trying to figure that out.






On Thursday, April 17, 2025 at 04:51:36 PM EDT, H. William Welliver <willia=
[email protected]> wrote:=20





Sorry, I guess I never mentioned that. I=E2=80=99m testing against 9.0, but=
 I suspect the problem exists in 8.x as well.

> On Apr 17, 2025, at 4:50=E2=80=AFPM, Lance Dillon <[email protected]>=
 wrote:
>=20
> Also, is there a specific version you are patching against? 8.0, 8.1, or =
9.0?
>=20
>=20
>=20
>=20
>=20
>=20
>=20
> On Thursday, April 17, 2025 at 04:46:30 PM EDT, Lance Dillon <riffraff169=
@yahoo.com> wrote:=20
>=20
>=20
>=20
>=20
>=20
> Is this a compile time error, or a runtime error?=C2=A0 If runtime, is th=
ere example code you have that seems to trigger it?
>=20
>=20
>=20
>=20
>=20
>=20
> On Thursday, April 17, 2025 at 04:05:16 PM EDT, H. William Welliver <will=
[email protected]> wrote:=20
>=20
>=20
>=20
>=20
>=20
> Hi Lance-
>=20
> Thanks for the assistance. I=E2=80=99ve attached the patch to this messag=
e, it should detect the presence of GdkNativeWindow. Let me know how it wor=
ks for you (and the result of that test in the configure output).
>=20
>=20
>=20
>=20
>> On Apr 17, 2025, at 3:26=E2=80=AFPM, Lance Dillon <[email protected]=
> wrote:
>>=20
>> I could test
>>=20
>> Yahoo Mail: Search, Organize, Conquer
>>=20
>>=20
>>>=C2=A0=20
>>>=C2=A0=20
>>> On Thu, Apr 17, 2025 at 15:18, H. William Welliver
>>> <[email protected]> wrote:
>>>=20
>>>=20
>>>=C2=A0=20
>>> I compiled pike on a system that has a version of GTK2 that includes a =
typedef for GdkNativeWindow that the GTK2 code doesn=E2=80=99t specifically=
 address. The compiler (a recent clang) errors on these implicit casts to a=
nd from GdkNativeWindow (which I believe resolves to a platform dependent p=
ointer) and integers that the pike glue deals with. I was able to solve the=
 errors but wanted to get some feedback before I commit it, as I get the im=
pression that GdkNativeWindow was something added somewhere along the way, =
so there probably may need to be some configure glue added. Is anyone using=
 the GTK2 module that might be able to test these changes to see if they ca=
use problems elsewhere? Or should I commit a best-guess fix and wait to see=
 how pikefarm reacts?
>>>=20
>>> Bill
>>>=20
>>>=20
>=20