Internally, we rename submitted classes and generated classes into statically defined namespaces. This is done for safety (even if somehow our whitelist check failed, the class name look-ups would fail) but also for identification (since we can statically determine what a class means based on this prefix).
The problem is that these names are gratuitously long, which will impose heavier restrictions on submitted class names, as described by our specification, since the underlying JVM has limits on these lengths, as well. Additionally, this makes any manual debugging more difficult since the names are longer, sometimes layered on top of each other, and differ only in small ways.
There are only a few of these so we should impose 1-character prefixes, instead of our longer ones.
Internally, we rename submitted classes and generated classes into statically defined namespaces. This is done for safety (even if somehow our whitelist check failed, the class name look-ups would fail) but also for identification (since we can statically determine what a class means based on this prefix).
The problem is that these names are gratuitously long, which will impose heavier restrictions on submitted class names, as described by our specification, since the underlying JVM has limits on these lengths, as well. Additionally, this makes any manual debugging more difficult since the names are longer, sometimes layered on top of each other, and differ only in small ways.
There are only a few of these so we should impose 1-character prefixes, instead of our longer ones.