I stumbled over a curious bug, I think:
I tried to read "512" as a number to base 6, and output it as base 16:
echo "ibase=6;obase=16;512" | bc
161
As you can see, the output is 161, but it should be bc
(sic!). I tried with base 10:
echo "ibase=6;obase=10;512" | bc
512
The value is unchanged. Curious! Default obase is 10. If I omit it:
echo "ibase=6;512" | bc
188
Well, that seems right. In a two step process, it works:
echo "obase=16;"$(echo "ibase=6;512" | bc) | bc
BC
So I made a script for different bases, but it keeps me puzzled:
for ib in {6,8,10,16}; do echo $ib; for ob in {10,16}; do echo -en $ib $ob" \t => " ; echo "ibase=$ib;obase=$ob;333" | bc ; done; done;
6
6 10 => 333
6 16 => 108
8
8 10 => 333
8 16 => 119
10
10 10 => 333
10 16 => 14D
16
16 10 => 333
16 16 => 01 15 05
Shall I file a bugreport, or do I miss the obvious? I can't really believe that such a basic tool is broken.
Not a bug.
As soon as
ibase=6
is interpreted, numbers are read in base 6. Soibase=6;obase=16
makes obase's value to be 16base 6 which is invalid, and is interpreted as 11decimal.From the man page:
So 16 is interpreted as 15base 6 which is 11decimal. And the conversion is correct.
Set
obase
beforeibase
, or make sure to specify your obase in baseibase
.See http://docstore.mik.ua/orelly/unix/upt/ch49_03.htm
When you set
ibase
orobase
, it's expressed with the current base ofibase
. So setobase
before you setibase
if you want to expressobase
in decimal.See also http://www.gnu.org/software/bc/manual/html_mono/bc.html#SEC9
So for
obase=16
inibase=6
, the6
becomes a5
, and it is equivalent to an output base of decimal6 * 1 + 1 * 5 == 11
, thus:Slightly modify the code (two ways) and your expected results appear:
execute in bash sandbox
results for both ways are: