Re: [Bug-apl] ⎕TC presents as Unix, not APL

[David B. Lamkins]

That's fine. I'm comfortable with the Unix/ASCII conventions. I was
mainly wondering whether dependence on ⎕TC might break my program on a
different APL.

On Sun, 2014-06-01 at 12:02 +0200, Juergen Sauermann wrote:
> Hi David,
> 
> I believe that the IBM specified behavior is only achievable (if at
> all)
> with platform specific (actually terminal specific) APL variants. The
> header
> on page 335 says:
> 
> ⎕TC Contains a three-item character vector of terminal control
> characters:
>  ⎕TC[1]—backspace
>  ⎕TC[2]—new line (return)
>  ⎕TC[3]—line feed
> 
> This is what GNU APL has implemented, taking ASCII verbatim:
> 
>       ⎕UCS ⎕TC
> 8 13 10
> 
> These numbers come from ASCII rather than from Unix. Then the purpose
> of ⎕TC is explained:
> 
> Use ⎕TC rather than ⎕AV to avoid system dependencies (because the
> order of ⎕AV is different in different APL implementations).
> 
> This assumes that the same character - take newline as an example -
> may be located at different positions
> of the ⎕AV but behaves identically otherwise. But then the display vs.
> typewriter example shows that this is
> not the case. I would add the Windows vs. Unix example where newline
> may or may not imply a carriage
> return.
> 
> If we would use curses as Elias suggested earlier that we could easily
> violate the 'three-item character vector'
> spec in the header.
> 
> Taking this all together tells me that ⎕TC, which was mainly provided
> for IBM code compatibility and not for real
> use in new programs, is a stone-age left-over (I remember having it
> seen in classical APL but never used) should
> remain as is.
> 
> Thank you nevertheless for pointing this out!
> /// Jürgen
> 
> 
> 
> On 05/31/2014 08:50 PM, David B. Lamkins wrote:
> 
> > See IBM Reference, page 335.
> > 
> > ⎕tc[2] and ⎕tc[3] differ from the IBM-specified behavior, presenting as
> > Unix format effectors rather than as APL format effectors.
> > 
> > In particular, APL specifies that ⎕tc[2] should advance the print
> > position to the start of the next line and ⎕tc[3] should advance the
> > print position downward one line while retaining the current column
> > position.
> > 
> > The current behavior of ⎕tc[1] is acceptable according to the IBM
> > Reference.
> > 
> > 
> > 
> > 
>