[fset-devel] Rereading reader bug for collections with non-NIL defaults.
Scott, Evaluation of this form fails with an error: (with-standard-io-syntax (let ((*readtable* (copy-readtable *readtable*)) (v (with-default (map) (set)))) (fset::fset-setup-rereading-readtable *readtable*) (with-input-from-string (s (with-output-to-string (s) (write (list v) :stream s))) (read s)))) ..whereas this succeeds: (with-standard-io-syntax (let ((*readtable* (copy-readtable *readtable*)) (v (map))) (fset::fset-setup-rereading-readtable *readtable*) (with-input-from-string (s (with-output-to-string (s) (write (list v) :stream s))) (read s)))) ...which probably suggests that the reader hasn't been conditioned to take care of the trailing "/foo" syntax. -- regards, Samium Gromoff -- "Actually I made up the term 'object-oriented', and I can tell you I did not have C++ in mind." - Alan Kay (OOPSLA 1997 Keynote)
On Thu, Jun 30, 2011 at 1:41 AM, Samium Gromoff <_deepfire@feelingofgreen.ru> wrote:
Scott,
Evaluation of this form fails with an error: [snip] ...which probably suggests that the reader hasn't been conditioned to take care of the trailing "/foo" syntax.
Thanks for the bug report! I think it's the first that's ever been sent to this list :-) So you're actually using the reader macros? I wasn't sure anyone would. (I haven't even been using them myself.) Do you need a fix quickly? -- Scott
On Thu, 30 Jun 2011 15:04:24 -0700, "Scott L. Burson" <Scott@sympoiesis.com> wrote:
Thanks for the bug report! I think it's the first that's ever been sent to this list :-)
I think I know the feeling : -) FSET more than deserves more visibility and public attention..
So you're actually using the reader macros? I wasn't sure anyone would. (I haven't even been using them myself.)
Yes, the kind of simple, observable and robust serialisation/deserialisation provided by PRINT/READ fits nicely for my current task.
Do you need a fix quickly?
Not really -- I've opted for a low-intrusivity workaround and the fix surely isn't critical in any sense. What I'm struggling now with is "sharing abbreviation", as mentioned in CLHS[1]: CL-USER> (read-from-string (write-to-string (let ((c (cons nil nil))) (cons c c)))) (#1=(NIL) . #1#) I'm trying to convert PRINT-* functions in fset.lisp to use PPRINT-LOGICAL-BLOCK, but am hitting some unexpected problems. Note that PPRINT-LOGICAL-BLOCK deals with *PRINT-LENGTH* and *PRINT-DEPTH* (which FSET was dealing with on its own), but requires the "logical contents" to be arranged as a list of conses -- which I think is a small price to pay. -- 1. http://cs.hbg.psu.edu/LOCAL/HyperSpec/Body/mac_pprint-logical-block.html -- regards, Samium Gromoff -- "Actually I made up the term 'object-oriented', and I can tell you I did not have C++ in mind." - Alan Kay (OOPSLA 1997 Keynote)
participants (2)
-
Samium Gromoff -
Scott L. Burson