CoCoA » CoCoALib

As far as I can see, the CoCoA-5 interpreter needs to be modified so that it catches both CoCoA::ErrorInfo and CoCoA::InterruptReceived objects. At the moment it seems to ignore the latter... to be honest I am not even sure how it achieves the current behaviour.

There are 11 places where Interpreter.C catches ErrorInfo -- why so many???
Do I need to add another catch command (for InterruptReceived) after each one?

The other problem is what to do once I have caught an InterruptReceived. With an ErrorInfo the existing code simply creates a RuntimeException object (incl. information about where in the CoCoA-5 source the interpreter had reached).

In amongst the CoCoA-5 code, I see that there is also InterruptException; this is presumably the correct one to throw, but in its current form the sole constructor is very limiting. I had better add some more ctors... sigh!

I have made some changes to Interpreter.C (mostly adding catch commands for InterruptReceived).

I have not (yet) changed the way Anna used CheckForInterrupt in TmpGReductor.C; while its use in RingBase::mySequentialPower is as I had intended. The following transcript show the slightly different behaviour which CoCoA-5 exhibits when CoCoALib is interrupted:

>>> f := x+y+z;
>>> g := f^300;  //  START THIS THEN INTERRUPT IT!
  C-c C-c
-------------------------
>>> CoCoA interrupted <<<
-------------------------

--> ERROR: 

*** Interrupted ***

--> g := f^300;  //  START THIS THEN INTE ...
-->       ^

>>> S := support(f^30);
>>> Sshifted := subst(S,[[x,x-2],[y,y+3],[z,z-5]]);
>>> I := ideal(Sshifted); // strangely slow!
>>> GB := GBasis(I); //  START THIS THEN INTERRUPT IT!
  C-c C-c
-------------------------
>>> CoCoA interrupted <<<
-------------------------

--> ERROR: InterruptReceived
--> GB := GBasis(I); //  START THIS THEN INTERRUP ...
-->       ^^^^^^^^^

>>> 

Comments? Opinions? Preferences?

I have now changed the call to CheckForInterrupt in TmpGReductor to the simple call that I had expected to see, and it works as I hoped/intended/expected/wanted/etc.

Since it is now just a matter of inserting calls to CheckForInterrupt("Fn name"); I am hoping that soon several new calls will be judiciously inserted so that lengthy CoCoALib computations can be interrupted with only a reasonable wait for recognition of the interruption.