PRISM Language Changes

As a consequence of the new and improved language parser in new versions of PRISM, there are a few changes to the PRISM modelling language which may break your existing models. In particular:

• Updates in commands must be parenthesised, e.g. this is allowed:

  [] s=1 & x>0 -> (s'=2) & (x'=x+1);
  

  but this is not:

  [] s=1 & x>0 -> s'=2 & x'=x+1;
  

  The former has always been the correct syntax but the parser was previously more lenient.

• Ranges (e.g. x=1..3,5) are unfortunately no longer supported. These must now be specified explicitly, e.g.:

  (x>=1&x<=3)|x=5
  

  At a later date, we may introduce a function or construct to replace this functionality.

Why have these changed been made?

The old PRISM language parser had several flaws, including a grammar that was ambiguous in places. In order to develop a new parser, with greater efficiency and much improved error handling, we had to remove these ambiguities, resulting in the changes above.

Is there a quick way of converting my old model files?

Yes - on Linux/Mac/etc. use the script in etc/scripts/prism3to4. Usage is:

• etc/scripts/prism3to4 old_model.pm new_model.pm

or just:

• etc/scripts/prism3to4 model.pm model.pm

Other questions or problems?

Please post a message in the discussion group.