Max-specpdl-size Bug in Emacs Lisp Interpreter?

Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Herbert Euler]

Hello,

These days I read source for 'catch' and 'throw' of emacs lisp.
Then I traced into unbind-to in eval.c, and some functions
operating 'specpdl'.  I created some hypothesises to understand
them.  Since 'let' invokes 'specbind' as well, I guessed I can
verify these hypothesises with let.  Here is what I did (I'm using
Emacs Unicode 2 from CVS, latest check Apr. 8, 2006):

(defmacro create-let (n)
  (let ((i 0)
	list)
    `(let ,(progn
	     (while (< i n)
	       (setq list (cons (intern (concat "v" (number-to-string i))) list))
	       (setq i (1+ i)))
	     list)
       (message "hello"))))
     => create-let

max-specpdl-size
     => 1000

(create-let 1000)
     => [error: Variable binding depth exceeds max-specpdl-size]

max-specpdl-size
     => 1000

(create-let 1000)
     => "hello"

(create-let 2000)
     => [error: Variable binding depth exceeds max-specpdl-size]

max-specpdl-size
     => 1101

(create-let 1200)
     => "hello"

max-specpdl-size
     => 1242

I just executed them sequentially.  The behavior of last several
evaluations may be different, but they are always not correct in
my Emacs (judged by 'max-specpdl-size' should not be modified
and 'create-let' should fail if its argument exceeds 'max-specpdl
-size').

Does this imply a bug in Emacs Lisp interpreter?

Regards,
Guanpeng Xu

_________________________________________________________________
Don't just search. Find. Check out the new MSN Search! 
http://search.msn.click-url.com/go/onm00200636ave/direct/01/

RE: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Herbert Euler]

Having read "(elisp)Local Variables", I know that it is a known feature.
But if so I think the name 'max-specpdl-size' should be altered, since
normally a maximum value needs explicitly change, i.e. there should be
a particular function to change a maximum value.  Such a way that
change it implicitly would makes people who don't know about it
confused about its "maximum" trait.

Regards,
Guanpeng Xu


>From: "Herbert Euler" <herberteuler@hotmail.com>
>To: emacs-devel@gnu.org
>Subject: Max-specpdl-size Bug in Emacs Lisp Interpreter?
>Date: Sat, 15 Apr 2006 16:24:56 +0800
>
>Hello,
>
>These days I read source for 'catch' and 'throw' of emacs lisp.
>Then I traced into unbind-to in eval.c, and some functions
>operating 'specpdl'.  I created some hypothesises to understand
>them.  Since 'let' invokes 'specbind' as well, I guessed I can
>verify these hypothesises with let.  Here is what I did (I'm using
>Emacs Unicode 2 from CVS, latest check Apr. 8, 2006):
>
>(defmacro create-let (n)
>  (let ((i 0)
>	list)
>    `(let ,(progn
>	     (while (< i n)
>	       (setq list (cons (intern (concat "v" (number-to-string i))) list))
>	       (setq i (1+ i)))
>	     list)
>       (message "hello"))))
>     => create-let
>
>max-specpdl-size
>     => 1000
>
>(create-let 1000)
>     => [error: Variable binding depth exceeds max-specpdl-size]
>
>max-specpdl-size
>     => 1000
>
>(create-let 1000)
>     => "hello"
>
>(create-let 2000)
>     => [error: Variable binding depth exceeds max-specpdl-size]
>
>max-specpdl-size
>     => 1101
>
>(create-let 1200)
>     => "hello"
>
>max-specpdl-size
>     => 1242
>
>I just executed them sequentially.  The behavior of last several
>evaluations may be different, but they are always not correct in
>my Emacs (judged by 'max-specpdl-size' should not be modified
>and 'create-let' should fail if its argument exceeds 'max-specpdl
>-size').
>
>Does this imply a bug in Emacs Lisp interpreter?
>
>Regards,
>Guanpeng Xu
>
>_________________________________________________________________
>Don't just search. Find. Check out the new MSN Search! 
>http://search.msn.click-url.com/go/onm00200636ave/direct/01/
>
>
>
>_______________________________________________
>Emacs-devel mailing list
>Emacs-devel@gnu.org
>http://lists.gnu.org/mailman/listinfo/emacs-devel

_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today it's FREE! 
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/

RE: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Herbert Euler]

Oh, no response.  Perhaps all deverlopers are too busy to
take a look at this.  Could somebody fix the following bug,
then?

    - Elisp manual says "The default value is 600" in
      lispref/variables.texi.  It is not true for Emacs in the
      CVS repository, in which the default value is 1000.

And I think it would be better if this information appears
not only in elisp manual, but also in docstring of 'max-specpdl
-size'.

Finally, a question on 'max-specpdl-size': the increment of
it cannot be avoided, can it?

Thanks.

Regards,
Guanpeng Xu

_________________________________________________________________
Don't just search. Find. Check out the new MSN Search! 
http://search.msn.click-url.com/go/onm00200636ave/direct/01/

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Romain Francoise]

"Herbert Euler" <herberteuler@hotmail.com> writes:

> Could somebody fix the following bug, then?

>    - Elisp manual says "The default value is 600" in
>      lispref/variables.texi.  It is not true for Emacs in the
>      CVS repository, in which the default value is 1000.

Thanks for the report, I fixed that.

-- 
Romain Francoise <romain@orebokech.com> | The sea! the sea! the open
it's a miracle -- http://orebokech.com/ | sea! The blue, the fresh, the
                                        | ever free! --Bryan W. Procter

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Richard Stallman]

    But if so I think the name 'max-specpdl-size' should be altered, since
    normally a maximum value needs explicitly change, i.e. there should be
    a particular function to change a maximum value.

What would you think of specpdl-size-limit as a name for this?

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Richard Stallman]

    Oh, no response.

Don't be so quick to say "no response"!
I responded yesterday.

You have to give TIME for people to respond.

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Herbert Euler]

>From: Richard Stallman <rms@gnu.org>
>Reply-To: rms@gnu.org
>To: "Herbert Euler" <herberteuler@hotmail.com>
>CC: emacs-devel@gnu.org
>Subject: Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?
>Date: Mon, 17 Apr 2006 13:56:06 -0400
>
>     But if so I think the name 'max-specpdl-size' should be altered, since
>     normally a maximum value needs explicitly change, i.e. there should be
>     a particular function to change a maximum value.
>
>What would you think of specpdl-size-limit as a name for this?

Yes it's better, since this name (in my opinion) is not confused.

Regards,
Guanpeng Xu

_________________________________________________________________
FREE pop-up blocking with the new MSN Toolbar - get it now! 
http://toolbar.msn.click-url.com/go/onm00200415ave/direct/01/

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Miles Bader]

"Herbert Euler" <herberteuler@hotmail.com> writes:
>>     But if so I think the name 'max-specpdl-size' should be altered, since
>>     normally a maximum value needs explicitly change, i.e. there should be
>>     a particular function to change a maximum value.
>>
>>What would you think of specpdl-size-limit as a name for this?
>
> Yes it's better, since this name (in my opinion) is not confused.

They sound exactly the same to me...

-Miles
-- 
A zen-buddhist walked into a pizza shop and
said, "Make me one with everything."

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Herbert Euler]

>From: Richard Stallman <rms@gnu.org>
>Reply-To: rms@gnu.org
>To: "Herbert Euler" <herberteuler@hotmail.com>
>CC: emacs-devel@gnu.org
>Subject: Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?
>Date: Mon, 17 Apr 2006 21:41:59 -0400
>
>     Oh, no response.
>
>Don't be so quick to say "no response"!
>I responded yesterday.
>
>You have to give TIME for people to respond.

Opps, I'm sorry.  But the mailing list is too slow, since I read the
first reply 3 days after my first post.  And I'm not used to this.
If one asks a simple problem or an important problem, one will get
response within an hour.  Compare these two different reacting
time, I conclude that

    1. Nobody is interesting at this topic.
    2. Nobody knows about it.
    3. Everybody is too busy to read, so they may want to return
      it later.
    4. The one knows about it thinks it's not a problem.
    5. Developers are working for it background.

Because I see that even thread posted after mine is replied, I can
disable case 3.  So it's only possible that the case is 1, 2, 4 or 5.  But
if a fresh man to Emacs (that's me) can understand the source code,
this problem isn't difficult, case 1 and 2 is apparently false.  So the
only true case is 4 or 5.  Three days is long enough for these two
cases, that's why I say "no response".

Besides, there are really lots of threads got no reply in this list (for
example, http://lists.gnu.org/archive/html/emacs-devel/2006-04/msg00458.html 
,
http://lists.gnu.org/archive/html/emacs-devel/2006-03/msg00879.html ,
and http://lists.gnu.org/archive/html/emacs-devel/2006-03/msg00397.html ).
I don't know if whoever will return to those topics, or if they've been
solved.  Perhaps Emacs developers know about it.

Regards,
Guanpeng Xu

_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today - it's FREE! 
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Luc Teirlinck]

Miles Bader wrote:

   "Herbert Euler" <herberteuler@hotmail.com> writes:
   >>     But if so I think the name 'max-specpdl-size' should be altered, since
   >>     normally a maximum value needs explicitly change, i.e. there should be
   >>     a particular function to change a maximum value.
   >>
   >>What would you think of specpdl-size-limit as a name for this?
   >
   > Yes it's better, since this name (in my opinion) is not confused.

   They sound exactly the same to me...

They sound exactly the same to me too.  I do not see any reason to
change the current name.

Sincerely,

Luc.

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[David Kastrup]

"Herbert Euler" <herberteuler@hotmail.com> writes:

>>From: Richard Stallman <rms@gnu.org>
>>Subject: Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?
>>Date: Mon, 17 Apr 2006 21:41:59 -0400
>>
>>     Oh, no response.
>>
>>Don't be so quick to say "no response"!
>>I responded yesterday.
>>
>>You have to give TIME for people to respond.
>
> Opps, I'm sorry.  But the mailing list is too slow, since I read the
> first reply 3 days after my first post.  And I'm not used to this.

The mailing list is not a single person, and you better get used to
it.  You can't expect everybody with an opinion on a question to read
the list hourly for your sake.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Juanma Barranquero]

On 4/18/06, Herbert Euler <herberteuler@hotmail.com> wrote:

> Compare these two different reacting
> time, I conclude that
>
>     1. Nobody is interesting at this topic.
>     2. Nobody knows about it.
>     3. Everybody is too busy to read, so they may want to return
>       it later.
>     4. The one knows about it thinks it's not a problem.
>     5. Developers are working for it background.

I think you're suffering from Warnock's Dilemma:

http://en.wikipedia.org/wiki/Warnock%27s_Dilemma

--
                    /L/e/k/t/u

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Herbert Euler]

>From: Luc Teirlinck <teirllm@dms.auburn.edu>
>To: miles@gnu.org
>CC: herberteuler@hotmail.com, rms@gnu.org, emacs-devel@gnu.org
>Subject: Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?
>Date: Tue, 18 Apr 2006 00:44:19 -0500 (CDT)
>
>Miles Bader wrote:
>
>    "Herbert Euler" <herberteuler@hotmail.com> writes:
>    >>     But if so I think the name 'max-specpdl-size' should be altered, 
>since
>    >>     normally a maximum value needs explicitly change, i.e. there 
>should be
>    >>     a particular function to change a maximum value.
>    >>
>    >>What would you think of specpdl-size-limit as a name for this?
>    >
>    > Yes it's better, since this name (in my opinion) is not confused.
>
>    They sound exactly the same to me...
>
>They sound exactly the same to me too.  I do not see any reason to
>change the current name.


Is "maximum" in English different from similar words in other languages?
Emacs is an "extensible, customizable" editor, leaving most options to
user.  Perhaps because of techniqucal reasons, this value cannot be
kept in poor conditions.  So it doesn't exactly confirm to what user
thinks, in my opinion, this is not the "maximum" value set by user.
So the two names are different to me.

Regards,
Guanpeng Xu

_________________________________________________________________
Don't just search. Find. Check out the new MSN Search! 
http://search.msn.com/

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Herbert Euler]

>From: David Kastrup <dak@gnu.org>
>To: "Herbert Euler" <herberteuler@hotmail.com>
>CC: rms@gnu.org,  emacs-devel@gnu.org
>Subject: Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?
>Date: Tue, 18 Apr 2006 08:36:01 +0200
>
>The mailing list is not a single person, and you better get used to
>it.  You can't expect everybody with an opinion on a question to read
>the list hourly for your sake.


So you minunderstood me.  Of course, I didn't mean I want a reply
within an hour.  I'm sorry, I didn't express it clearly.

When I posted this thread, I just wanted to confirm whether it's a bug.
Perhaps I should post it in bug-gnu-emacs, but since I was not sure about
it and most Emacs developers read threads in this list as well, I posted it
here.  Did you notice the last sentence?  It's "Does this imply a bug in
Emacs Lisp interpreter?"  Simply "no" is also an answer.  After that, most
developers had followed other threads, but didn't follow this one.  And
I then replied it myself, but again got no response after another day, in
which most developers had read this mailing list again.  And so on.  I'm
not an Emacs developer, and also not familar with Emacs internals.  I
believe Emacs developers know about it, so I subscribe this mailing list
and post.  Appearntly I'm too anxious to say "no response", for I'm
not sure does there someone notice that, and thread last more than
several days definitely will be forgot.

Sincerely,
Guanpeng Xu

_________________________________________________________________
FREE pop-up blocking with the new MSN Toolbar - get it now! 
http://toolbar.msn.click-url.com/go/onm00200415ave/direct/01/

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Richard Stallman]

It takes me at least a day to respond even to a trivial message.
You must give me time!

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Richard Stallman]

The name `max-specpdl-size' can be read as implying an inherent maximum
one that cannot be exceeded.  By contrast, the term `limit' seems to fit
an adjustable, arbitrary maximum--which is what this variable is.

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Miles Bader]

Richard Stallman <rms@gnu.org> writes:
> The name `max-specpdl-size' can be read as implying an inherent maximum
> one that cannot be exceeded.  By contrast, the term `limit' seems to fit
> an adjustable, arbitrary maximum--which is what this variable is.

One _could_ read those meanings into the words max and limit, but they
hardly seem inherent; one could just as well see "limit" as denoting a
fundamental limit which cannot be exceeded.  As far as I can tell it's a
personal language issue with the original poster.

Instead of changing the names of variables to suit the idiosyncrasies of
individuals, it seems much better to simply make sure the variable
documentation clearly states the meaning and addresses any common
misconceptions.

-Miles
-- 
x
y
Z!

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[David Kastrup]

Miles Bader <miles@gnu.org> writes:

> Richard Stallman <rms@gnu.org> writes:
>> The name `max-specpdl-size' can be read as implying an inherent maximum
>> one that cannot be exceeded.  By contrast, the term `limit' seems to fit
>> an adjustable, arbitrary maximum--which is what this variable is.
>
> One _could_ read those meanings into the words max and limit, but
> they hardly seem inherent; one could just as well see "limit" as
> denoting a fundamental limit which cannot be exceeded.  As far as I
> can tell it's a personal language issue with the original poster.
>
> Instead of changing the names of variables to suit the
> idiosyncrasies of individuals, it seems much better to simply make
> sure the variable documentation clearly states the meaning and
> addresses any common misconceptions.

And of course, consistency would do no harm.  I have not sampled other
limits, whether system-imposed or user-modifiable, so I can't vouch
whether one or the other name seems more in line with other usage.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Herbert Euler]

>From: Richard Stallman <rms@gnu.org>
>Reply-To: rms@gnu.org
>To: "Herbert Euler" <herberteuler@hotmail.com>
>CC: emacs-devel@gnu.org
>Subject: Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?
>Date: Tue, 18 Apr 2006 16:50:26 -0400
>
>It takes me at least a day to respond even to a trivial message.
>You must give me time!

I'm sorry I didn't know only you know about this topic.  I thought
others would know, and just left it because it's stupid.

Regards,
Guanpeng Xu

_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today it's FREE! 
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Herbert Euler]

>From: Miles Bader <miles@gnu.org>
>Reply-To: Miles Bader <miles@gnu.org>
>To: rms@gnu.org
>CC: herberteuler@hotmail.com,  emacs-devel@gnu.org
>Subject: Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?
>Date: Wed, 19 Apr 2006 08:34:53 +0900
>
>Richard Stallman <rms@gnu.org> writes:
> > The name `max-specpdl-size' can be read as implying an inherent maximum
> > one that cannot be exceeded.  By contrast, the term `limit' seems to fit
> > an adjustable, arbitrary maximum--which is what this variable is.
>
>One _could_ read those meanings into the words max and limit, but they
>hardly seem inherent; one could just as well see "limit" as denoting a
>fundamental limit which cannot be exceeded.  As far as I can tell it's a
>personal language issue with the original poster.
>
>Instead of changing the names of variables to suit the idiosyncrasies of
>individuals, it seems much better to simply make sure the variable
>documentation clearly states the meaning and addresses any common
>misconceptions.

I agree.  I think it would be better if this information appears not only
in elisp manual, but also in docstring of 'max-specpdl-size' as well, if the
increment cannot be avoided.

Regards,
Guanpeng Xu

_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today it's FREE! 
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Herbert Euler]

>From: David Kastrup <dak@gnu.org>
>To: Miles Bader <miles@gnu.org>
>CC: rms@gnu.org,  herberteuler@hotmail.com,  emacs-devel@gnu.org
>Subject: Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?
>Date: Wed, 19 Apr 2006 01:43:25 +0200
>
>And of course, consistency would do no harm.  I have not sampled other
>limits, whether system-imposed or user-modifiable, so I can't vouch
>whether one or the other name seems more in line with other usage.

Maybe it's stupid: Is it possible to let the Lisp debugger register some
variable so that once the Lisp interpreter see debugger has finished
and 'max-specpdl-size' is different from what user set it will set 'max-
specpdl-size' back, to keep the consistency of other applications of
the term 'max'?  (I think 'max' in other applications, at least in most
other applications, is inherent.)

Regards,
Guanpeng Xu

_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today it's FREE! 
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Luc Teirlinck]

Guanpeng Xu wrote:

   I agree.  I think it would be better if this information appears not only
   in elisp manual, but also in docstring of 'max-specpdl-size' as well, if the
   increment cannot be avoided.

It _already_ appears in the docstring of `max-specpdl-size':

  You can safely use a value considerably larger than the default value,
  if that proves inconveniently small.  However, if you increase it too far,
  Emacs could run out of memory trying to make the stack bigger.

  You can customize this variable.

Sincerely,

Luc.

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Luc Teirlinck]

Guanpeng Xu wrote:

   I agree.  I think it would be better if this information appears
   not only in elisp manual, but also in docstring of
   'max-specpdl-size' as well, if the increment cannot be avoided.

I now see that with "this information" you meant that entry to the
Lisp debugger increases the value.  It would do no harm to mention
this it the docstring.  However, I believe that the current docstring
already makes clear that `max-specpdl-size' is no hard limit.

Sincerely,

Luc.

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Herbert Euler]

>From: Luc Teirlinck <teirllm@dms.auburn.edu>
>To: herberteuler@hotmail.com
>CC: emacs-devel@gnu.org, miles@gnu.org
>Subject: Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?
>Date: Tue, 18 Apr 2006 23:03:25 -0500 (CDT)
>
>Guanpeng Xu wrote:
>
>    I agree.  I think it would be better if this information appears
>    not only in elisp manual, but also in docstring of
>    'max-specpdl-size' as well, if the increment cannot be avoided.
>
>I now see that with "this information" you meant that entry to the
>Lisp debugger increases the value.  It would do no harm to mention
>this it the docstring.  However, I believe that the current docstring
>already makes clear that `max-specpdl-size' is no hard limit.

If you mean "However, if you increase it too far, Emacs could run
out of memory trying to make the stack bigger." make "no hard
limit" clear, I don't agree.  Who will know what the connection
between 'max-specpdl-size' and stack is without reading source
and/or without reading elisp manual?  Perhaps even reading elisp
manual is insufficient.

Regards,
Guanpeng Xu

_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today it's FREE! 
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/

Re: Max-specpdl-size Bug in Emacs Lisp Interpreter?

[Richard Stallman]

    Maybe it's stupid: Is it possible to let the Lisp debugger register some
    variable so that once the Lisp interpreter see debugger has finished
    and 'max-specpdl-size' is different from what user set it will set 'max-
    specpdl-size' back, to keep the consistency of other applications of
    the term 'max'?

It would be a good idea to do this.