update libkdtree
update libkdtree to last version
How about using a submodule for this instead? That way:
- we can more easily update it in the future
- more secure updates, this update is so big that it is difficult to review that no malicious code has been added
@wwmayer what do you think about migrating to a submodule, as @hyarion suggests?
we can more easily update it in the future
Sure.
more secure updates, this update is so big that it is difficult to review that no malicious code has been added
It's not really big. Most of the PR is about to put the automake stuff back that we will never use. The only change actually is to re-enable the copy constructor of the iterator class.
The copy constructor once has been disabled with 3883ef3a30bb5e because it caused some compiler warnings in the mesh module. So, when we make this a sub-module is there a way to apply a patch to disable the copy constructor again?
So, when we make this a sub-module is there a way to apply a patch to disable the copy constructor again?
As far as I know, you can't apply local patches directly to submodules. Instead, the process would look like this:
- Fork the
libkdtreerepository to create aFreeCAD/libkdtreerepository. - Apply your local patch(es) in the
FreeCAD/libkdtreefork. - Use the
FreeCAD/libkdtreefork as the submodule in your FreeCAD repository.
When you want to pull in changes from the original libkdtree repository, the process is as follows:
- In the
FreeCAD/libkdtreefork, pull changes from the upstream repository and either merge them into or rebase your local patches onto the updated code. - Push the updated fork, including your patches, to the
FreeCAD/libkdtreerepository. - In the
FreeCADrepository, update the submodule to point to the latest commit in yourFreeCAD/libkdtreefork.
@mosfet80 ping for feedback
btw, git-subtree is an alternative to submodules that might also work in this case
@mosfet80 ping
bump
I agree with @hyarion that using git subtree is probably a better approach here (and likely elsewhere, but that is a discussion for another day). Then we can just make a commit on top of the subtree that disables the problematic copy constructor.
Any feedback here @mosfet80 ?
@mosfet80 FYI
@mosfet80 FYI
This week i switch libkdtree to a freecad submodule
@mosfet80 sorry for asking, but what was the reason to close this PR?
We decided on subtree, rather than submodule, didn't we?