Re: [coreutils] tr: case mapping anomaly

[Eric Blake]

On 09/29/2010 06:40 AM, Pádraig Brady wrote:
> > > +    # Ensure the size of the case classes are accounted
> > > +    # for as a unit.
> > > +    echo 'ABCDEFGHIJKLMNOPQRSTUVWXYZ' |
> > > +    tr '[:upper:]A-B' '[:lower:]0'>out ||  _fail=1
> > > +    echo '00cdefghijklmnopqrstuvwxyz'>  exp
> >
> > Huh?  A and B are both in [:upper:]; when a character is listed more
> > than once in string1, it is only transliterated according to the first
> > listing.  I think this should be 'abc...' not '00c...' for the expected
> > results.
>
> Does POSIX specify that?

Hmm - POSIX appears to be silent for both tr and m4's translit (but 
whereas 'tr long short' is unspecified and we extend the last byte of 
short to match, m4's translit(data,long,short) is explicitly documented 
as deleting bytes from long with no match in short).

> That's not what we do, nor what I would expect.
>
> $ echo 'A' | LANG=C tr 'AA' '01'
> 1

And Solaris' tr does this as well.

But m4 behaves in the way I specified:

$ echo 'translit(a,aa,01)' | m4
0
$ echo 'translit(a,aa,01)' | /usr/ccs/bin/m4
0

Time for me to ask for clarification from the Austin Group, I suppose. 
But given existing practice, I guess the argument should be that tr is 
explicitly different than m4.

--
Eric Blake   address@hidden    +1-801-349-2682
Libvirt virtualization library http://libvirt.org