eval trouble

eval trouble

[Lennart Borgman (gmail)]

I expected the code below to give me a function named
mumamo-repl4-my-own9-mode. It does not. Can someone please explain what
I am doing wrong?

Interestingly if I do describe-variable on xx and then use the shown value t

  (setq xx THE-SHOWN-VALUE

then

   (eval xx)

does what I want.

Here is the code that does NOT work:

(defun mumamo-define-no-mode (mode-sym)
  (let ((mumamo-repl4 (make-symbol (format "mumamo-repl4-%s" mode-sym)))
        (lighter (format "No %s" mode-sym))
        (doc
         (format
	  "MuMaMo replacement for the mode %s which was not found."
          mode-sym)))
    (setq xx
      `(define-derived-mode ,mumamo-repl4 fundamental-mode ,lighter ,doc))
    (message "xx=%s" (pp-to-string xx))
    (eval xx)
    ))

(mumamo-define-no-mode 'my-own9-mode)

Re: eval trouble

[Lennart Borgman (gmail)]

Lennart Borgman (gmail) wrote:
> I expected the code below to give me a function named
> mumamo-repl4-my-own9-mode. It does not. Can someone please explain what
> I am doing wrong?
> 
> Interestingly if I do describe-variable on xx and then use the shown value t
> 
>   (setq xx THE-SHOWN-VALUE
> 
> then
> 
>    (eval xx)
> 
> does what I want.
> 
> Here is the code that does NOT work:
> 
> (defun mumamo-define-no-mode (mode-sym)
>   (let ((mumamo-repl4 (make-symbol (format "mumamo-repl4-%s" mode-sym)))

It works if I replace make-symbol with intern. But why does eval care
about that?

>         (lighter (format "No %s" mode-sym))
>         (doc
>          (format
> 	  "MuMaMo replacement for the mode %s which was not found."
>           mode-sym)))
>     (setq xx
>       `(define-derived-mode ,mumamo-repl4 fundamental-mode ,lighter ,doc))
>     (message "xx=%s" (pp-to-string xx))
>     (eval xx)
>     ))
> 
> (mumamo-define-no-mode 'my-own9-mode)
> 
> 
> 

Re: eval trouble

[Pascal J. Bourguignon]

"Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:
>> Here is the code that does NOT work:
>> 
>> (defun mumamo-define-no-mode (mode-sym)
>>   (let ((mumamo-repl4 (make-symbol (format "mumamo-repl4-%s" mode-sym)))
>
> It works if I replace make-symbol with intern. But why does eval care
> about that?

eval doesn't care, but how would you be able to refer again to the
defined thing if you don't have its name in some dictionary?

It's like putting your suitcase to the baggage checkroom, and losing
the key right away.  As soon as the checkroom realises there's no more
any key to unlock your suitcase, it'll recover the space and throw it
to the garbage.


-- 
__Pascal Bourguignon__                     http://www.informatimago.com/

THIS IS A 100% MATTER PRODUCT: In the unlikely event that this
merchandise should contact antimatter in any form, a catastrophic
explosion will result.

Re: eval trouble

[Lennart Borgman (gmail)]

Pascal J. Bourguignon wrote:
> "Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:
>>> Here is the code that does NOT work:
>>>
>>> (defun mumamo-define-no-mode (mode-sym)
>>>   (let ((mumamo-repl4 (make-symbol (format "mumamo-repl4-%s" mode-sym)))
>> It works if I replace make-symbol with intern. But why does eval care
>> about that?
> 
> eval doesn't care, but how would you be able to refer again to the
> defined thing if you don't have its name in some dictionary?


Thanks. I see. I thought eval interned the symbol, or perhaps rather
created an interned symbol. In this case it does not do that.

Or am I still misunderstanding something.

Re: eval trouble

[Lennart Borgman (gmail)]

Pascal J. Bourguignon wrote:
> Lennart Borgman (gmail) writes:
>> Pascal J. Bourguignon wrote:
>>> "Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:
>>>>> Here is the code that does NOT work:
>>>>>
>>>>> (defun mumamo-define-no-mode (mode-sym)
>>>>>   (let ((mumamo-repl4 (make-symbol (format "mumamo-repl4-%s" mode-sym)))
>>>> It works if I replace make-symbol with intern. But why does eval care
>>>> about that?
>>> eval doesn't care, but how would you be able to refer again to the
>>> defined thing if you don't have its name in some dictionary?
>>
>> Thanks. I see. I thought eval interned the symbol, or perhaps rather
>> created an interned symbol. In this case it does not do that.
>>
>> Or am I still misunderstanding something.
> 
> Only intern interns symbols.  intern is called automatically only by
> the lisp reader.  It's when a s-expression is read that symbols it
> contains are interned.  When you call eval, it's already done.

Yepp. I was still misunderstanding. Thanks, that was very clear.

Re: eval trouble

[Pascal J. Bourguignon]

Lennart Borgman (gmail) writes:
> Pascal J. Bourguignon wrote:
> > "Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:
> >>> Here is the code that does NOT work:
> >>>
> >>> (defun mumamo-define-no-mode (mode-sym)
> >>>   (let ((mumamo-repl4 (make-symbol (format "mumamo-repl4-%s" mode-sym)))
> >> It works if I replace make-symbol with intern. But why does eval care
> >> about that?
> > 
> > eval doesn't care, but how would you be able to refer again to the
> > defined thing if you don't have its name in some dictionary?
> 
> 
> Thanks. I see. I thought eval interned the symbol, or perhaps rather
> created an interned symbol. In this case it does not do that.
> 
> Or am I still misunderstanding something.

Only intern interns symbols.  intern is called automatically only by
the lisp reader.  It's when a s-expression is read that symbols it
contains are interned.  When you call eval, it's already done.


-- 
__Pascal Bourguignon__                     http://www.informatimago.com/

CAUTION: The mass of this product contains the energy equivalent of
85 million tons of TNT per net ounce of weight.

Re: eval trouble

[David Kastrup]

"Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:

> Lennart Borgman (gmail) wrote:
>> I expected the code below to give me a function named
>> mumamo-repl4-my-own9-mode. It does not.

It does.  It gives you a function named mumamo-repl4-my-own9-mode.  It
just does not make the name known to the world.

A symbol has four cells: a function cell, a name, a value cell and a
property list.  And you can intern a symbol into an OBARRAY, which is
sort of a hash references by name.

The Lisp reader identifies names with symbols by looking them up in the
global 'obarray.

IIRC, there are no functions to change a symbol's name (which is fixed
at creation time), and there is no way to enter a symbol into an obarray
at any index except its fixed name.  And you can only enter symbols into
a single obarray at most, at creation time when you specify the symbol's
name to "intern".  If you don't want to enter it into any obarray, use
make-symbol.  The given symbol has a name, but you can't reference it by
its name then.

>> Can someone please explain what I am doing wrong?
>> 
>> Interestingly if I do describe-variable on xx and then use the shown value t
>> 
>>   (setq xx THE-SHOWN-VALUE
>> 
>> then
>> 
>>    (eval xx)
>> 
>> does what I want.

Sure, because you then store the symbol itself and don't need its name
as a reference.

>> Here is the code that does NOT work:
>> 
>> (defun mumamo-define-no-mode (mode-sym)
>>   (let ((mumamo-repl4 (make-symbol (format "mumamo-repl4-%s" mode-sym)))
>
> It works if I replace make-symbol with intern. But why does eval care
> about that?

Because eval, when given the string "'mumamo-repl4-mode" will look up
mumamo-repl4-mode in the global obarray, and there is nothing with that
name in that obarray.

A symbol with that name exists, but not referenced through its name.

Each call of make-symbol creates a new symbol, like each call of list
creates a new list.

(eq (make-symbol "x") (make-symbol "x")) -> nil
(eq (intern "x") 'x) -> t
(make-symbol "x") -> x

Note that the last line just prints the _name_ of the created symbol.
It does not mean that you can use the name to get back the symbol, just
like

(eq 1.0 1.0) -> nil

creates two different objects which are not EQ.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: eval trouble

[David Kastrup]

"Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:

> Pascal J. Bourguignon wrote:
>> "Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:
>>>> Here is the code that does NOT work:
>>>>
>>>> (defun mumamo-define-no-mode (mode-sym)
>>>>   (let ((mumamo-repl4 (make-symbol (format "mumamo-repl4-%s" mode-sym)))
>>> It works if I replace make-symbol with intern. But why does eval care
>>> about that?
>> 
>> eval doesn't care, but how would you be able to refer again to the
>> defined thing if you don't have its name in some dictionary?
>
>
> Thanks. I see. I thought eval interned the symbol, or perhaps rather
> created an interned symbol. In this case it does not do that.

Eval _does_ create an interned symbol.  But it is not EQ to the unique
uninterned symbol that make-symbol creates.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: eval trouble

[Lennart Borgman (gmail)]

David Kastrup wrote:
> "Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:
> 
>> Pascal J. Bourguignon wrote:
>>> "Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:
>>>>> Here is the code that does NOT work:
>>>>>
>>>>> (defun mumamo-define-no-mode (mode-sym)
>>>>>   (let ((mumamo-repl4 (make-symbol (format "mumamo-repl4-%s" mode-sym)))
>>>> It works if I replace make-symbol with intern. But why does eval care
>>>> about that?
>>> eval doesn't care, but how would you be able to refer again to the
>>> defined thing if you don't have its name in some dictionary?
>>
>> Thanks. I see. I thought eval interned the symbol, or perhaps rather
>> created an interned symbol. In this case it does not do that.
> 
> Eval _does_ create an interned symbol.  But it is not EQ to the unique
> uninterned symbol that make-symbol creates.
> 

I am a bit confused, but I guess you mean if the symbol does not exist:

(let ((sym (make-symbol "xýzw")))
  (eval `(progn (setq ,sym 1)
                (put ',sym 'some 'value))))

(mapatoms (lambda (atom)
            (when (string= (symbol-name atom)
                           "xyzw")
              (error "Found %s" atom))))

Re: eval trouble

[David Kastrup]

"Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:

> David Kastrup wrote:
>> "Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:
>> 
>>> Pascal J. Bourguignon wrote:
>>>> "Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:
>>>>>> Here is the code that does NOT work:
>>>>>>
>>>>>> (defun mumamo-define-no-mode (mode-sym)
>>>>>>   (let ((mumamo-repl4 (make-symbol (format "mumamo-repl4-%s" mode-sym)))
>>>>> It works if I replace make-symbol with intern. But why does eval care
>>>>> about that?
>>>> eval doesn't care, but how would you be able to refer again to the
>>>> defined thing if you don't have its name in some dictionary?
>>>
>>> Thanks. I see. I thought eval interned the symbol, or perhaps rather
>>> created an interned symbol. In this case it does not do that.
>> 
>> Eval _does_ create an interned symbol.  But it is not EQ to the
>> unique uninterned symbol that make-symbol creates.
>
> I am a bit confused, but I guess you mean if the symbol does not exist:
>
> (let ((sym (make-symbol "xýzw")))
>   (eval `(progn (setq ,sym 1)
>                 (put ',sym 'some 'value))))

Ok, I actually talked nonsense right now: eval does _not_ create or
intern symbols.  It is the Lisp reader that creates/interns symbols when
you feed an identifier to it.  The Lisp reader will, when given the
above, maybe-create/intern/lookup all of let, sym, make-symbol, eval, \`
or something similar, progn, setq, \, or something similar, sym, put,
some and value.  And whatever else follows.

> (mapatoms (lambda (atom)
>             (when (string= (symbol-name atom)
>                            "xyzw")
>               (error "Found %s" atom))))

Well, you wrote xýzw rather than xyzw above.  But even if you didn't, as
long as nothing else interned 'xyzw in some manner, such a symbol would
not get found.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: eval trouble

[Lennart Borgman (gmail)]

David Kastrup wrote:
> Ok, I actually talked nonsense right now: eval does _not_ create or
> intern symbols.  It is the Lisp reader that creates/interns symbols when
> you feed an identifier to it.  The Lisp reader will, when given the
> above, maybe-create/intern/lookup all of let, sym, make-symbol, eval, \`
> or something similar, progn, setq, \, or something similar, sym, put,
> some and value.  And whatever else follows.

Thanks.

>> (mapatoms (lambda (atom)
>>             (when (string= (symbol-name atom)
>>                            "xyzw")
>>               (error "Found %s" atom))))
> 
> Well, you wrote xýzw rather than xyzw above.  But even if you didn't, as
> long as nothing else interned 'xyzw in some manner, such a symbol would
> not get found.

Shit. There is shit on my screen, I did not see that ý. I should keep
the screen more clean ;-)

Re: eval trouble

[David Combs]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1467 bytes --]

In article <mailman.17522.1219697977.18990.help-gnu-emacs@gnu.org>,
Lennart Borgman (gmail) <lennart.borgman@gmail.com> wrote:
>David Kastrup wrote:
>> "Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:
>> 
>>> Pascal J. Bourguignon wrote:
>>>> "Lennart Borgman (gmail)" <lennart.borgman@gmail.com> writes:
>>>>>> Here is the code that does NOT work:
>>>>>>
>>>>>> (defun mumamo-define-no-mode (mode-sym)
>>>>>>   (let ((mumamo-repl4 (make-symbol (format "mumamo-repl4-%s" mode-sym)))
>>>>> It works if I replace make-symbol with intern. But why does eval care
>>>>> about that?
>>>> eval doesn't care, but how would you be able to refer again to the
>>>> defined thing if you don't have its name in some dictionary?
>>>
>>> Thanks. I see. I thought eval interned the symbol, or perhaps rather
>>> created an interned symbol. In this case it does not do that.
>> 
>> Eval _does_ create an interned symbol.  But it is not EQ to the unique
>> uninterned symbol that make-symbol creates.
>> 
>
>I am a bit confused, but I guess you mean if the symbol does not exist:
>
>(let ((sym (make-symbol "xýzw")))
>  (eval `(progn (setq ,sym 1)
>                (put ',sym 'some 'value))))
>
>(mapatoms (lambda (atom)
>            (when (string= (symbol-name atom)
>                           "xyzw")
>              (error "Found %s" atom))))
>
>
>


Please, why the comma before sym -- quoted, no less:

>                (put ',sym 'some 'value))))

Thanks!

David

RE: eval trouble

[Drew Adams]

> Please, why the comma before sym -- quoted, no less:
> > (put ',sym 'some 'value))))

See the Emacs manual, node Backquote.

Inside a backquote (`):
 ,sym evaluates sym and uses that value.
 ' then wraps that in (quote ...)

So if the value of sym is hello,
then ,sym gives hello and ',sym gives 'hello.

Re: eval trouble

[Kevin Rodgers]

Drew Adams wrote:
>> Please, why the comma before sym -- quoted, no less:
>>> (put ',sym 'some 'value))))
> 
> See the Emacs manual, node Backquote.
> 
> Inside a backquote (`):
>  ,sym evaluates sym and uses that value.
>  ' then wraps that in (quote ...)
> 
> So if the value of sym is hello,
> then ,sym gives hello and ',sym gives 'hello.

As does (quote ,sym)

-- 
Kevin Rodgers
Denver, Colorado, USA

Re: eval trouble

[David Combs]

In article <mailman.19402.1221638109.18990.help-gnu-emacs@gnu.org>,
Kevin Rodgers  <kevin.d.rodgers@gmail.com> wrote:
>Drew Adams wrote:
>>> Please, why the comma before sym -- quoted, no less:
>>>> (put ',sym 'some 'value))))
>> 
>> See the Emacs manual, node Backquote.
>> 
>> Inside a backquote (`):
>>  ,sym evaluates sym and uses that value.
>>  ' then wraps that in (quote ...)
>> 
>> So if the value of sym is hello,
>> then ,sym gives hello and ',sym gives 'hello.
>
>As does (quote ,sym)
>
>-- 
>Kevin Rodgers
>Denver, Colorado, USA


I guess I need some more caffein (or sleep)!

THANKS!

David