binary compatibility: occt + patch?

[looo]

There is a request to add a patch[1] to occt [2] for a reported issue [3]. Now the question is if applying this patch will result in a binary-compatible build. If not we have to rebuild netgen, smesh, freecad with a new occt (A lot of work) So my question is more general:
Which diff of the source lead to binary incompatible builds and which diff result in a binary compatible build?

[1] http://git.dev.opencascade.org/gitweb/? ... 636c2cf130
[2] https://github.com/conda-forge/occt-feedstock/pull/30
[3] https://tracker.dev.opencascade.org/view.php?id=30760

[wmayer]

Now the question is if applying this patch will result in a binary-compatible build.

This won't break binary compatibility.

Which diff of the source lead to binary incompatible builds and which diff result in a binary compatible build?

In general changes to source files (.cpp, .cxx, ..) don't break binary compatibility while changes in header files can break it. This happens when changing the signature of an existing function, removing a function, changing the order of class members or adding new members.

See also
https://stackoverflow.com/questions/371 ... patibility
https://www.acodersjourney.com/20-abi-breaking-changes/

[looo]

Thanks!

[looo]

I have another question regarding binary compatibility:

boost depends on icu which recently was updated. Is it necessary to rebuild any boost dependent package because of the newer icu version?

[kkremitzki]

Speaking of this, I'm about to upload OCCT 7.3.0p3 aka 7.3.3 to Debian...

[Kunda1]

Speaking of this, I'm about to upload OCCT 7.3.0p3 aka 7.3.3 to Debian...

Nice!