Moved Sfm classes to their own files

[dellaert]

• Moved SfmData and SfmTrack into sfm
• deprecated incorrectly named functions
• moved wrapping

[johnwlambert]

Hi Frank, thanks for the PR. Was curious how long we plan to keep this flag set to true by default (GTSAM_ALLOW_DEPRECATED_SINCE_V42).

We do use the number_measurements() flag and number_tracks() methods all over GTSFM, so we would need to refactor quite a few files. Was just curious the rationale for the renaming.

[dellaert]

Flag should actually be OFF by default.
Yes, I'm aware that if we build a new wheel (4.2a5) you'll have to amend GTSAM.

[johnwlambert]

Looks like CI is failing:

[ 24%] Building CXX object gtsam/CMakeFiles/gtsam.dir/sfm/SfmData.cpp.o
[ 24%] Building CXX object gtsam/CMakeFiles/gtsam.dir/sfm/SfmTrack.cpp.o
/home/runner/work/gtsam/gtsam/gtsam/sfm/SfmTrack.cpp:30:1: error: stray ‘\303’ in program
 bool SfmTrack::equals(const SfmTrack& sfmTrack, double tolÏ) const {
 ^
/home/runner/work/gtsam/gtsam/gtsam/sfm/SfmTrack.cpp:30:1: error: stray ‘\217’ in program
make[3]: *** [gtsam/CMakeFiles/gtsam.dir/sfm/SfmTrack.cpp.o] Error 1
gtsam/CMakeFiles/gtsam.dir/build.make:1797: recipe for target 'gtsam/CMakeFiles/gtsam.dir/sfm/SfmTrack.cpp.o' failed
make[3]: *** Waiting for unfinished jobs....
CMakeFiles/Makefile2:3093: recipe for target 'gtsam/CMakeFiles/gtsam.dir/all' failed
make[2]: *** [gtsam/CMakeFiles/gtsam.dir/all] Error 2
CMakeFiles/Makefile2:2267: recipe for target 'CMakeFiles/check.dir/rule' failed
make[1]: *** [CMakeFiles/check.dir/rule] Error 2
Makefile:199: recipe for target 'check' failed

[dellaert]

Looks like CI is failing:

I know, I'll fix it, along with addressing any other comments you might have. If you have no comments, please approve. If you do, still approve and trust me I will execute on them or talk to you if not :-)

[johnwlambert]

Sounds good @dellaert. The PR looks nice and I'll approve on my end. I like the decomposition of the BAL stuff into sfmData.cpp, since it never belonged in slamDataset.cpp so much.

I understand the snake case -> camel case change, to be consistent with our C++ guidelines. But I personally would find SfmData.numberTracks() to be more readable than .nrTracks(). This is a minor nit though, and if you feel like the latter is more consistent with GTSAM conventions, that's fine with me.

[dellaert]

But I personally would find SfmData.numberTracks() to be more readable than .nrTracks(). This is a minor nit though, and if you feel like the latter is more consistent with GTSAM conventions, that's fine with me.

yeah, I could change that...