Ok, I'm knee-deep in packfile.c looking at alignment issues, and have a
two questions about storing FLOATVALs in the CONSTANTS section of the
bytecode: [First, thanks to Gregor Purdy for the self-documenting
packfile.c and the nice companion docs/parrotbyte.pod.]
In docs/parrotbyte.pod, under CONSTANTS, it says
For number constants (S is constant, and is equal to C<sizeof(FLOATVAL)>):
+----------+----------+----------+----------+
| |
| S' bytes of Data |
| |
+----------+----------+----------+----------+
where
S' = S + (S % 4) ? (4 - (S % 4)) : 0
If S' E<gt> S, then the extra bytes are filled with zeros.
First, may I assume that '4' is supposed to be sizeof(opcode_t), not a plain
4?.
Second, this claims that FLOATVALS are supposed to be padded out to an
opcode_t boundary, but currently, in packfile.c, they are not. One place
where this is an issue is gcc/x86, where sizeof(long long) = 8, but
sizeof(long double) = 12, so that FLOATVALS aren't automatically padded
out to an opcode_t boundary.
Are they supposed to be padded?
--
Andy Dougherty [EMAIL PROTECTED]
