bug#14978: 24.3.50; `savehist-printable' implementation question for strings

bug#14978: 24.3.50; `savehist-printable' implementation question for strings

[Drew Adams]

The code does this:

;; String without properties
((and (stringp value)
      (equal-including-properties value (substring-no-properties value))))

1. Doesn't this do the same thing as that `equal-including-properties'
   test:

   (and (null (text-properties-at 0 value))
        (= 0 (next-property-change 0 value)))

   If it does do the same thing, is it more efficient or less?

   (The reason I ask whether it does the same thing is for use in an
   older Emacs release where the functions in the original test are
   not available.)

2. Is it really necessary to exclude all text properties from a printed
   string in order for it to be Lisp-readable?  If not, can we improve
   this code so it is not unnecessarily restrictive?

I'm guessing that for #2 the answer is yes, it is necessary, because a
text property can have any Lisp value, including a circular list value,
but it's not clear to me what the problems are.


In GNU Emacs 24.3.50.1 (i686-pc-mingw32)
 of 2013-07-21 on ODIEONE
Bzr revision: 113485 lekktu@gmail.com-20130722012547-e3b7qxn1dba5vf20
Windowing system distributor `Microsoft Corp.', version 6.1.7601
Configured using:
 `configure --prefix=/c/Devel/emacs/binary --enable-checking=yes,glyphs
 CFLAGS=-O0 -g3 LDFLAGS=-Lc:/Devel/emacs/lib
 CPPFLAGS=-Ic:/Devel/emacs/include'

bug#14978: 24.3.50; `savehist-printable' implementation question for strings

[Stefan Monnier]

> 1. Doesn't this do the same thing as that `equal-including-properties'
>    test:

>    (and (null (text-properties-at 0 value))
>         (= 0 (next-property-change 0 value)))

I think so, yes.

>    If it does do the same thing, is it more efficient or less?

Yours might be more efficient because it avoids memory allocation.
But they probably each are more efficient in some circumstances.

> 2. Is it really necessary to exclude all text properties from a printed
>    string in order for it to be Lisp-readable?  If not, can we improve
>    this code so it is not unnecessarily restrictive?

You'd have to make sure all the properties are themselves readable.

> I'm guessing that for #2 the answer is yes, it is necessary, because a
> text property can have any Lisp value,

That's right.

> including a circular list value,

Actually circular values should be handled fine; problems come when you
bump into objects like markers, buffers, windows, processes, ...


        Stefan

bug#14978: 24.3.50; `savehist-printable' implementation question for strings

[Drew Adams]

> > including a circular list value,
> 
> Actually circular values should be handled fine

Yes, but only if you bind (or set) `read-circle' to non-nil when
reading, no?

Anyway, thanks for confirming my guesses.  I have closed the bug.

bug#14978: 24.3.50; `savehist-printable' implementation question for strings

[Stefan Monnier]

>> > including a circular list value,
>> Actually circular values should be handled fine
> Yes, but only if you bind (or set) `read-circle' to non-nil when
> reading, no?

That's right, except that it's non-nil by default.  So it works unless
you bind it to nil.


        Stefan