From Newsgroup: comp.unix.shell

This was something that actually started elsewhere; first I noticed
the (token-)symmetric true predicate

((10#2==2#10))

i.e. that "2 cents (decimal) equals 10 cents (binary)". - Okay, that
started nice.


Then I wondered whether 2 and 10 are the only numbers that fulfill
the (informally written) equation

((n#m==m#n)) [*]

and wanted to automate such tests looping n, m through some numbers.

[*] Can that equation be proved or disproved for (m,n)-values other
than 2 and 10?


Quickly I noticed that 1#0 is not a legal number (in Kornshell), and
also 'bc' refuses that value

ibase=1
Runtime warning (func=(main), adr=3): ibase too small, set to 2

Of course math with only one digit 0 available makes not much sense,
but it didn't occur to me that base 1 is forbidden in the first place.
(I've certainly never pondered about that before.)


Continuing tests with 'bc'

ibase=17
Runtime warning (func=(main), adr=6): ibase too large, set to 16

And strange got this same error message for

ibase=14
Runtime warning (func=(main), adr=6): ibase too large, set to 16

I suppose no one cares about such details in exotic application cases.
(It's anyway a bug of only minor importance.)


BTW, Kornshell operates with bases up to 64 and I wonder what symbols
are used for that. Tests show that it depends on the base; for small
bases letters in caps are handled like lowercase letters, for larger
bases using caps matters, and for some bases we even get errors, e.g.

$ echo $((37#ff))
570
$ echo $((37#FF))
ksh: 37#FF: arithmetic syntax error

Looks like a bug to me.

Janis

--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: comp.unix.shell

On 18.02.2025 10:59, Janis Papanagnou wrote:

BTW, Kornshell operates with bases up to 64 and I wonder what symbols
are used for that. [...]

Experiments lead to: 0..9 a..z A..Z @ _
(and it's also documented - doh!)

Janis

--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: comp.unix.shell

On 2025-02-18, Janis Papanagnou <janis_papanagnou+ng@hotmail.com> wrote:

And strange got this same error message for

ibase=14
Runtime warning (func=(main), adr=6): ibase too large, set to 16

Did you already set ibase before and that "14" is no longer decimal
14?
--
Christian "naddy" Weisgerber naddy@mips.inka.de
--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: comp.unix.shell

On 18.02.2025 14:13, Christian Weisgerber wrote:

On 2025-02-18, Janis Papanagnou <janis_papanagnou+ng@hotmail.com> wrote:

And strange got this same error message for

ibase=14
Runtime warning (func=(main), adr=6): ibase too large, set to 16

Did you already set ibase before and that "14" is no longer decimal
14?

LOL - indeed. Thanks! - It didn't occur to me that setting the base
for calculations will also affect the base for setting parameters.
(Apart from the misleading error message it smells not like a good
design but that explains it.)

Janis

--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: comp.unix.shell

On 18.02.2025 10:59, Janis Papanagnou wrote:

[...]

BTW, Kornshell operates with bases up to 64 and I wonder what symbols
are used for that. Tests show that it depends on the base; for small
bases letters in caps are handled like lowercase letters, for larger
bases using caps matters, and for some bases we even get errors, e.g.

$ echo $((37#ff))
570
$ echo $((37#FF))
ksh: 37#FF: arithmetic syntax error

Looks like a bug to me.

...but it isn't! - As I've been pointed out base 37 just means that
0-9, a-z, and A are accepted as digits, which of course makes sense.

Janis

--- Synchronet 3.21b-Linux NewsLink 1.2