(n)sublis doesn't look inside backquoted expression

[ds26gte]

(sublis '((a . b) (c . d)) '`(,a ,b ,c ,d))

                   ;where the second arg is a backquoted list preceded by a regular quote

should produce (the equivalent of)

`(,B ,B ,D ,D)

Instead it produces

`(,A ,B ,C ,D)

Similar bug for nsublis. Reference correct implementations include: Clozure CL, ECL, MKCL, ECL, CLISP.

I'm using

ds26gte@longitude:~/tmp$ sbcl --version
SBCL 1.3.14.88-60d0ebeed
ds26gte@longitude:~/tmp$ uname -a
Linux longitude 4.8.0-34-generic #36-Ubuntu SMP Wed Dec 21 17:24:18 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux

* *features*

(:QUICKLISP :SB-BSD-SOCKETS-ADDRINFO :ASDF-PACKAGE-SYSTEM :ASDF3.1 :ASDF3
 :ASDF2 :ASDF :OS-UNIX :NON-BASE-CHARS-EXIST-P :ASDF-UNICODE :64-BIT
 :64-BIT-REGISTERS :ALIEN-CALLBACKS :ANSI-CL :ASH-RIGHT-VOPS
 :C-STACK-IS-CONTROL-STACK :COMMON-LISP :COMPACT-INSTANCE-HEADER
 :COMPARE-AND-SWAP-VOPS :COMPLEX-FLOAT-VOPS :CYCLE-COUNTER :ELF :FLOAT-EQL-VOPS
 :FP-AND-PC-STANDARD-SAVE :GENCGC :IEEE-FLOATING-POINT :IMMOBILE-CODE
 :IMMOBILE-SPACE :INLINE-CONSTANTS :INTEGER-EQL-VOP :LARGEFILE :LINKAGE-TABLE
 :LINUX :LITTLE-ENDIAN :MEMORY-BARRIER-VOPS :MULTIPLY-HIGH-VOPS
 :OS-PROVIDES-BLKSIZE-T :OS-PROVIDES-DLADDR :OS-PROVIDES-DLOPEN
 :OS-PROVIDES-GETPROTOBY-R :OS-PROVIDES-POLL :OS-PROVIDES-PUTWC
 :OS-PROVIDES-SUSECONDS-T :PACKAGE-LOCAL-NICKNAMES :PRECISE-ARG-COUNT-ERROR
 :RAW-INSTANCE-INIT-VOPS :RAW-SIGNED-WORD :SB-DOC :SB-EVAL :SB-FUTEX :SB-LDB
 :SB-PACKAGE-LOCKS :SB-SIMD-PACK :SB-SOURCE-LOCATIONS :SB-TEST :SB-THREAD
 :SB-UNICODE :SBCL :STACK-ALLOCATABLE-CLOSURES :STACK-ALLOCATABLE-FIXED-OBJECTS
 :STACK-ALLOCATABLE-LISTS :STACK-ALLOCATABLE-VECTORS
 :STACK-GROWS-DOWNWARD-NOT-UPWARD :SYMBOL-INFO-VOPS :UNBIND-N-VOP :UNIX
 :UNWIND-TO-FRAME-AND-CALL-VOP :X86-64)

Thanks,

--d

[Stas Boukarev (stassats)]

Where does the standard state that this should happen?

[Attila Lendvai (attila-lendvai)]

isn't it undefined by the standard what the backquote reader reads into? you can only rely on its evaluation rules.

[ds26gte (ds26gte)]

While different Lisps do expand backquote in their own way, it remains that in all implementations the resulting expression is a tree, and in my example, the tree from the RHS does contain the symbols A, B, C, D as leaves.

The standard says: "sublis looks at all subtrees and leaves of _tree_; if a subtree or leaf appears as a key in _alist_ ... it is replaced ..."

[Stas Boukarev (stassats)]

SBCL doesn't use conses for representing ",", so, no trees for them.

[Brett van de Sande (bvds)]

Warning, this behavior has changed in SBCL itself ...
In SBCL version 1.0.29:

* (sublis '((a . 1) (b . 2)) '(a `(,b 3)))

(1 `(,2 3))

In SBCL version 1.2.11:

* (sublis '((a . 1) (b . 2)) '(a `(,b 3)))

(1 `(,B 3))

I was unable to find any documentation associated with this change in behavior.
Quietly changing a behavior like this, with no warning messages, is not a good thing.

[Brett van de Sande (bvds)]

Here is the change in SBCL:

http://www.sbcl.org/all-news.html#1.2.2

With some further comments here:

http://christophe.rhodes.io/notes/blog/posts/2014/backquote_and_pretty_printing/

[Attila Lendvai (attila-lendvai)]

if you need to rely on the internal format of the quasi quote reader, then you can use this library instead of the implementation provided one:

http://quickdocs.org/fare-quasiquote/