I have been running a little suite of shell scripts to do testing on my (glulx) WIP. For the most part, it works very well, but every now and then I get different behavior between the command-line interpreter (glulxe) and the IDE interpreter. For a while the only differences I saw were in randomly generated behavior, but just now I saw a completely different set of rules run after the same sequence of commands in each interpreter. What’s going on… is memory getting clobbered somewhere? How do I even troubleshoot this?
Okay, well I just noticed that there was a potential difference in the circumstances, as a result of a randomly chosen object earlier in the sequence of play. So do different interpreters have different pseudorandom number algorithms?
Conceivably. The spec leaves the PRNG choice up to the terp.
You should be able to set the IDE interpreter to glulxe, which may help, unless the two compilations linked in different PRNGs.