Edward Lam
Edward Lam
Sorry, I've been pulled away and haven't found time to really explore this again. So this should theoretically work with `RTLD_DEEPBIND` because we're injecting the hook after `dlopen()` has loaded...
Sorry, I'm late to the game but I'm going to voice my dissent here anyways. First change I see is in `openvdb/Exceptions.h` which got `` modified to `"openvdb/version.h"` instead. But...
> Is this the only issue you see or are you disagreeing with this PR more generally? This is easily fixable by tweaking the CMake include paths (which are needed...
> We've reviewed what these changes intend... right? We just missed this one. `version.h` is special because, at build time, it has to be pulled from the binary directory. But...
If we can't have upwards relative include paths, then I don't see why we should make any of these changes at all since it defeats the purpose of changing to...
> However, if a path is given with -I then **this causes the path to be searched after the current directory and AHEAD of the standard system directories.** But this...
PS. I can't seem to edit my comment above but you'll need to rename the decompressed `inctest.tar.gz` to `/tmp/test` to use `run.sh`, or else you'll need to modify the paths...
> It seems that we've ultimately come to the conclusion that this change won't change anything for clients. As Jeff noted above, the only potential advantage of using quote includes...
@Idclip wrote: > I wasn't able to confirm whether or not Houdini on Windows does automatically `scan C:\Users\Me\houdini18.0\dso` ? Houdini on Windows scans `HOUDINI_DSO_PATH` in the same way as all...
@Idclip wrote: > Sorry, bit confused here - do you mean to say that `C:\Users\\houdini\dso` is automatically scanned on windows irrespective of the value of `HOUDINI_DSO_PATH`? As I'm unable to...