Hi, I think I've found a bug in a compiler. For the reproduction of this behaviour, use asrun sample, but modify the script in the script.as so the enum E looks similar to this one:

enum E
{
VALUE1 = 20somegarbage,
VALUE2 = 30moregarbage
}

The application will compile and run without any error. But it should throw some "bad suffix on number" error information.
I think that the problem is in as_tokenizer.cpp in method:

bool asCTokenizer::IsConstant(const char *source, size_t sourceLength, size_t &tokenLength, eTokenType &tokenType) const

which identifies "20somegarbage" and "30moregarbage" as ttIntConstant. In this loop:


size_t n;
for( n = 0; n < sourceLength; n++ )
{
if( source[n] < '0' || source[n] > '9' )
break;
}


it only goes to first unallowed char (in this case "s" for the VALUE1) and does nothing about the fact that 20somegarbage is not a proper constant identifier here.

For this test I'm using HEAD revision (currently 1301), but this error is present in earlier versions of the AngelScript library.

Thanks in advance!

Yes, you're correct. I'll have this fixed.

Thanks for the detailed analysis on the problem.

Thank you . You're welcome .

I've fixed this bug in revision 1303.

Thanks!