Marco Baringer wrote:
speaking of compiling directly to pbc, parrotbyte.pod says that number constants should be encoded as FLOATVALs, which is a system dependant careteristic, shouldn't there be a specified format for floats? ie either ieee single/double float or, preferably, the number segment specifies how bytes are exponent and how many bytes are mantissa?
This is currently not finished. packfile does read native doubles (and on i386 long doubles) that's all, but only if written with same config.
And yes, we need some notation, how packed doubles look like:
- ieee
- machine X / Y ...
- proabably textual like string constants as last effort
sorry if this has already been discussed, i only looked at the
assembler and noticed it uses pack("d").
May be discussed but currently unsolved (s. KNOWN_ISSUES) leo
