On 2026/07/02 14:14, Theo Buehler wrote:
> On Wed, Jul 01, 2026 at 04:38:25PM +0000, Johannes Thyssen Tishman wrote:
> > 2026-06-20T13:54:47+0000 Johannes Thyssen Tishman <[email protected]>:
> > > Please find below an update for graphics/coin to version 4.0.9.
> > >
> > > Changelog: https://github.com/coin3d/coin/releases/tag/v4.0.9
> > >
> > > Notes about this update:
> > > 1. Enable tests: 100% tests passed, 0 tests failed out of 1
> > > 2. No shared library bump needed as there were no dynamic export
> > > changes.
>
> The value of an enum changed (math is hard). This often requires
> a major bump (see below):
good catch!
> diff -pu -r /usr/local/include/Inventor/C/base/string.h
> /usr/ports/pobj/coin-4.0.9/fake-amd64/usr/local/include/Inventor/C/base/string.h
> --- /usr/local/include/Inventor/C/base/string.h Tue Mar 24 09:14:35 2026
> +++
> /usr/ports/pobj/coin-4.0.9/fake-amd64/usr/local/include/Inventor/C/base/string.h
> Tue Jun 16 23:00:52 2026
> @@ -44,7 +44,7 @@ extern "C" {
> /* ********************************************************************** */
>
> enum cc_string_constants {
> - CC_STRING_MIN_SIZE = 128 - sizeof(char *) + sizeof(size_t),
> + CC_STRING_MIN_SIZE = 128 - sizeof(char *) - sizeof(size_t),
> CC_STRING_RESIZE = 128
> };
>
> This change shrinks the buffer in the public struct cc_string:
>
> struct cc_string {
> char * pointer;
> size_t bufsize;
> char buffer[CC_STRING_MIN_SIZE];
> };
>
> On all our platforms CC_STRING_MIN_SIZE would previously work out to 128
> and now it is 112 on LP64 and 120 on ILP32. This is an ABI break
> requiring a major bump, as applications expecting to be able to write
> CC_STRING_MIN_SIZE to buffer need to be recompiled to avoid writing past
> the end.
>
> i don't think this is actually used this way anywhere. We should still
> bump the libCoin major.
>
> > > 3. Build tested consumers graphics/py-pivy and cad/freecad (which I'll
> > > bump after committing this update)
>
> Not against it, but there is no static lib and it's not a header-based
> library as far as I can see, so I see no need for a bump of the consumers.
> The shlib major bump alone should do the trick.
>
> ok tb
>
