It is good practice to initialize pointers (to NULL, if nothing else). There seems to be a code path where this pointer could be returned to the user with the intention to have it deallocated, though I admit I do not follow all the code's logic well enough to know if this would happen in practice.
In examining the code, this line seemed like a candidate to solve the problem in #83 - and it seems to have, at least in my application.
I am aware that memory issues are hard to diagnose, and simply recompiling with this change could have moved the issue elsewhere (where my application does not tickle it). So I am not certain this is 100% a fix.
It is good practice to initialize pointers (to NULL, if nothing else). There seems to be a code path where this pointer could be returned to the user with the intention to have it deallocated, though I admit I do not follow all the code's logic well enough to know if this would happen in practice.
In examining the code, this line seemed like a candidate to solve the problem in #83 - and it seems to have, at least in my application.
I am aware that memory issues are hard to diagnose, and simply recompiling with this change could have moved the issue elsewhere (where my application does not tickle it). So I am not certain this is 100% a fix.