On 24 Jul 2015, at 14:33, Alejandro Zamora Fonseca <terefv@ltu.sld.cu> wrote:

Hi!

Just now playing a little with ABCL(1.3.2), I saw some strange behaviour:

CL-USER> '(2 . 5)
(2 . 5)
CL-USER> '(2 . 5 . 5)
(2 . 5)
CL-USER> (equal '(2 . 5) '(2 . 5 . 5))
T

while other implementations give me an error when i type '(2 . 5 . 5)
it's a bug or ANSI CL allows this?

Depends.

ANSI CL specifies the lisp reader in such a way that (2 . 5 . 5) is not a valid syntax.
However, ANSI CL specifies that implementation may extend the CL language, UNDER THE CONDITION that extensions BE DOCUMENTED.

Therefore, IF it is written in ABCL documentation that (2 . 5 . x) reads as (2 . 5) 
THEN it is not a bug, but an ABCL specific extension,
ELSE it is a conformity bug, it should signal a READER-ERROR (or perhaps just an ERROR).


Notice that since this is an extensions on the reader macro character #\(, it could be expected that the standard reader macro character for #\( still exist in the system and be obtained by (copy-readtable NIL) which shall return the standard readtable.
For example, in ccl the default *readtable* is poluted by a lot of extensions such as the dispatching #/ #_ #$ macro characters, but you can throw them away with (setf *readtable* (copy-readtable NIL)).  One could expect the same here.


http://www.lispworks.com/documentation/HyperSpec/Body/02_da.htm
http://www.lispworks.com/documentation/HyperSpec/Body/01_f.htm
http://www.lispworks.com/documentation/HyperSpec/Body/f_cp_rdt.htm
-- 
__Pascal J. Bourguignon__