Original ticket http://projects.scipy.org/scipy/ticket/713 on 2008-08-04 by trac user yosefm, assigned to unknown.
The attached patch allows the use of scipy.optimize.fsolve and scipy.optimize.leastsq in concurrent threads.
Extra data required to the callback functions (that are called from the Fortran code) is passed in by appending a pointer to it to the end of the buffers used by the callbacks - instead of the current method of using globals.
Attachment added by trac user yosefm on 2008-08-04: nuke_globals-v3.patch
Milestone changed to 0.8.0 by @cournape on 2009-05-02
@dlax wrote on 2012-09-18
If you are still interested, it'd be great if you could turn your patch into a pull request on github. Otherwise, I'm afraid this might stay there for a while... Thanks.