Smallest possible size

[kengruven]

The CBE spec says:

"Preservation of the original numeric data type information is not considered important by default. Encoders SHOULD [i.e., but are not required to] use the smallest encoding that stores a value without data loss."

It looks like enctool is not yet clever enough to pick the smallest numeric type for a given value. It just looks at the CTE notation to decide which CBE encoding to use. Sometimes it actually picks the biggest one:

$ echo 'c1 32768.0' | ../enctool/enctool convert -df cbe|hexdump
0000000 03 00 65 00 80 80 02                           
0000007
$ echo 'c1 0x8000.0' | ../enctool/enctool convert -df cbe|hexdump
0000000 03 00 70 00 47                                 
0000005
$ echo 'c1 32768' | ../enctool/enctool convert -df cbe|hexdump
0000000 03 00 6a 00 80                                 
0000005

[kstenerud]

Yup, the current reference implementation doesn't try very hard yet.