SANS mantid crash

[Gesner Passos]

Introduced in Septempter 27th.

Run the ticket #8245 twice and mantid crashes.

[Gesner Passos]

This is a quite strange error. I wander how to explain it. The crash was first seen in the commit ​https://github.com/mantidproject/mantid/commit/212816a712e70926fedeb2fdfebf2047fb041d6a.

The connection on how that code could allow the system to crash is obscure.

[Gesner Passos]

How to reproduce the error:

• Open SANS interface
• select SANS2D
• Load Mask (​​http://trac.mantidproject.org/mantid/attachment/ticket/8245/MASKSANS2D_133R_12m_8mm_XPress_changer.txt)
• Change to batch-mode
• Load ​​http://trac.mantidproject.org/mantid/attachment/ticket/8245/test_trans_inp.csv
• Run 1D reduction.
• Close SANS interface
• Open SANS interface again
• Load ​​http://trac.mantidproject.org/mantid/attachment/ticket/8245/test_trans_inp.csv

Run 1D reduction.

Mantid crashes.

[Gesner Passos]

It was found that the problem is related to the m_delete_observer, a Poco::NObserver<SANSRunWindow, Mantid::API::WorkspacePostDeleteNotification>. It was added to added to the notificationCenter twice but removed just once. Due to this, it might be that after closing the interface, the notificationCenter may try to get an invalid callback.

[Gesner Passos]

re #8294: SANS window remove double addition of delete observer

This cause the Mantid to crash when you close and reopen the SANS interface and perform some reduction.

Changeset: db98ba3b3d9c766828d0bab641a05b37d2a16f68

[Karl Palmen]

There is no crash when I reproduce the error as described above, but the 1D reduction does cause a python error.

[Karl Palmen]

Merge remote-tracking branch 'origin/bugfix/8294_sans_crash_deleting_wks'

Full changeset: ​0310fa2efe256093a7ca9482c4dae35cfd53eae9

[Stuart Campbell]

This ticket has been transferred to github issue ​9139