changing SPECTUB global variables to class members

[danieldeidda]

No description provided.

[KrisThielemans]

remove comment in

STIR/src/include/stir/recon_buildblock/SPECTUB_Weight3d.h

Line 27 in 94efb22

\todo Variables wm, wmh and Rrad are currently global variables. This means that this code would be very dangerous

remove extern declarations at

STIR/src/include/stir/recon_buildblock/SPECTUB_Weight3d.h

Lines 38 to 40 in 94efb22

	extern wm_da_type wm; //! weight (or probability) matrix
	extern wmh_type wmh; //! information to construct wm
	extern float * Rrad; //! radii per view

Will need to disable the copy-constructor which is something like

// disable copy-constructor as currently unsafe to copy due to bare pointers
ProjMatrixSPECTUB(const &ProjMatrixSPECTUB) = delete;

Ideally (in next step) make Rrad and std::vector such that we don't need to deallocate etc. Can then be passed to SPECTUB functions as &Rrad[0], which is of course a bit ugly... So we can leave that for later.

[danieldeidda]

Will need to disable the copy-constructor which is something like

// disable copy-constructor as currently unsafe to copy due to bare pointers
ProjMatrixSPECTUB(const &ProjMatrixSPECTUB) = delete;

Where should this go I am guessing in the ProjMatrixByBinSPECTUB.h?

[KrisThielemans]

notice a syntax error. that'll likely have to be

ProjMatrixSPECTUB(const ProjMatrixSPECTUB&) = delete;

[KrisThielemans]

let's try to make float * Rrad a const float * Rrad where possible

[KrisThielemans]

Linking errors

SPECTUB_Tools.cxx:(.text+0x3b0a): undefined reference to `SPECTUB::wmh

so some functions are still expecting the global varaibles

[KrisThielemans]

Great! It is working fine.

Can you clean-up by removing comments, and adding something to release_5.2.htm? (along the lines of "removed global variables in the SPECTUB code, such that we can now have multiple SPECTUB projectors (PR #1169)"

[KrisThielemans]

@samdporter want to try this out? it should work now.

[KrisThielemans]

@mastergari we'll have to do the same for the pinhole code presumably. can do that later.

[samdporter]

@samdporter want to try this out? it should work now.

Looks good with the tests in here.

I'll do something a bit more difficult and let you know

Sam

[KrisThielemans]

great. @samdporter let us know when you're happy.

Ok to squash-merge?

[samdporter]

great. @samdporter let us know when you're happy.

Ok to squash-merge?

Looks good to me

see updated repo