Bryce told me to direct this to Mental or Bulia:
Since we are clearing source changes now, I'll mention these two things that need tweaking. Nothing major, but they can break builds:
1. In satisfied_guide_cns.cpp, the prototype for
void satisfied_guide_cns(SPDesktop const &desktop, std::vectorNR::Point &snappoints, std::vector<SPGuideConstraint> &cns)
...needs to match the .h file. This can cause a linking problem in some situations, such as dynamic instead of static, or link order.
2. This one is my fault. I recoded prefix.cpp to use gthreads; it works nicely, but this file is not our responsibility. The original binreloc version needs to be put back. (OOPS. sorry)
I'll not touch any sources until after the release :)
Bob
On Tue, 13 Jul 2004 11:26:19 -0500, Bob Jamison wrote:
- This one is my fault. I recoded prefix.cpp to use
gthreads; it works nicely, but this file is not our responsibility. The original binreloc version needs to be put back. (OOPS. sorry)
Oh, that's OK, nothing wrong with it using gthreads if you like. You only need to renamespace the functions if you change their semantics and/or prototypes, this change doesn't.
thanks -mike
On Tue, 2004-07-13 at 12:26, Bob Jamison wrote:
- This one is my fault. I recoded prefix.cpp to use
gthreads; it works nicely, but this file is not our responsibility. The original binreloc version needs to be put back. (OOPS. sorry)
I'm inclined to leave it like that for now. IMO, at least for our purposes, using pthreads directly is a bug.
-mental
On Tue, Jul 13, 2004 at 11:26:19AM -0500, Bob Jamison wrote:
- In satisfied_guide_cns.cpp, the prototype for
void satisfied_guide_cns(SPDesktop const &desktop, std::vectorNR::Point &snappoints, std::vector<SPGuideConstraint> &cns)
...needs to match the .h file.
Both should have
std::vectorNR::Point const &snappoints
as second argument.
pjrm.
participants (4)
-
Bob Jamison -
MenTaLguY -
Mike Hearn -
Peter Moulder