bug#6457: 24.0.50; Inappropriate error msg for set-variable

bug#6457: 24.0.50; Inappropriate error msg for set-variable

[Drew Adams]

emacs -Q
M-x set-variable RET scroll-conservatively 
Set scroll-conservatively globally to value (0): 99999999 RET
 
No problem. Now do it again, adding one more `9':
M-x set-variable RET scroll-conservatively
Set scroll-conservatively globally to value (99999999): 999999999 RET
 
The error message is this:
 
Value `999999999.0' does not match type integer of
scroll-conservatively.
 
1. There should be no `.0' added to the echoed input. The actual input
should be echoed.  Emacs shouldn't seem to be complaining about
something the user didn't enter.
 
2. The cause of the error is an arithmetic overflow.  The message should
indicate this, if possible.  If not, it's OK to say that an integer
wasn't entered (since 999999999 is too big to correspond to an integer),
but such a message is not ideal and will be confusing to some users.
 
But the main problem is #1.  Confusion from a message saying that you
didn't input an integer is even greate if what is echoed is a decimal
number (999999999.0).
 
In GNU Emacs 24.0.50.1 (i386-mingw-nt5.1.2600)
 of 2010-06-14 on 3249CTO
Windowing system distributor `Microsoft Corp.', version 5.1.2600
configured using `configure --with-gcc (4.4) --no-opt --cflags -Ic:/xpm/include'
 

bug#6457: 24.0.50; Inappropriate error msg for set-variable

[Juri Linkov]

> emacs -Q
> M-x set-variable RET scroll-conservatively
> Set scroll-conservatively globally to value (0): 99999999 RET
>
> No problem. Now do it again, adding one more `9':
> M-x set-variable RET scroll-conservatively
> Set scroll-conservatively globally to value (99999999): 999999999 RET

What is a number in parens?  I don't have these parens.
Do you use a modified source?

> The error message is this:
>
> Value `999999999.0' does not match type integer of
> scroll-conservatively.

I can't reproduce this bug when I set to 999999999.

When I tried a larger value "9999999999999999999",
I get the following error:

  Arithmetic overflow error: "9999999999999999999"

-- 
Juri Linkov
http://www.jurta.org/emacs/

bug#6457: 24.0.50; Inappropriate error msg for set-variable

[Drew Adams]

> > emacs -Q
> > M-x set-variable RET scroll-conservatively
> > Set scroll-conservatively globally to value (0): 99999999 RET
> >
> > No problem. Now do it again, adding one more `9':
> > M-x set-variable RET scroll-conservatively
> > Set scroll-conservatively globally to value (99999999): 
> 999999999 RET
> 
> What is a number in parens?  I don't have these parens.
> Do you use a modified source?

Forget the number in parens.
No, it does not appear with emacs -Q.

> > The error message is this:
> >
> > Value `999999999.0' does not match type integer of
> > scroll-conservatively.

That is still the (inappropriate) error message.

> I can't reproduce this bug when I set to 999999999.

Perhaps it is platform-specific.

> When I tried a larger value "9999999999999999999",
> I get the following error:
> 
>   Arithmetic overflow error: "9999999999999999999"

Perhaps the inappropriate error message is platform-specific.

I definitely see that message with emacs -Q.

bug#6457: 24.0.50; Inappropriate error msg for set-variable

[Lars Ingebrigtsen]

"Drew Adams" <drew.adams@oracle.com> writes:

>> > The error message is this:
>> >
>> > Value `999999999.0' does not match type integer of
>> > scroll-conservatively.
>
> That is still the (inappropriate) error message.
>
>> I can't reproduce this bug when I set to 999999999.
>
> Perhaps it is platform-specific.
>
>> When I tried a larger value "9999999999999999999",
>> I get the following error:
>> 
>>   Arithmetic overflow error: "9999999999999999999"
>
> Perhaps the inappropriate error message is platform-specific.
>
> I definitely see that message with emacs -Q.

I get:

set-variable: Value `1e+19' does not match type integer of scroll-conservatively

So I can reproduce this bug.

(This is with Emacs 23.3.50 on Fedora.)

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/

bug#6457: 24.0.50; Inappropriate error msg for set-variable

[Glenn Morris]

Lars Ingebrigtsen wrote:

> set-variable: Value `1e+19' does not match type integer of scroll-conservatively
>

I guess this is due to the feature whereby values too large for integers
are read as floats, eg

http://lists.gnu.org/archive/html/emacs-devel/2010-03/msg00181.html

If so, I guess it is impossible for lisp code to know whether the user
originally typed an integer or a float. Therefore it is impossible to do
anything about this. It frankly doesn't seem worth bothering about
anyway.

bug#6457: 24.0.50; Inappropriate error msg for set-variable

[Eli Zaretskii]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Date: Wed, 05 Feb 2014 16:28:23 -0800
> Cc: 6457@debbugs.gnu.org
> 
> "Drew Adams" <drew.adams@oracle.com> writes:
> 
> >> > The error message is this:
> >> >
> >> > Value `999999999.0' does not match type integer of
> >> > scroll-conservatively.
> >
> > That is still the (inappropriate) error message.
> >
> >> I can't reproduce this bug when I set to 999999999.
> >
> > Perhaps it is platform-specific.
> >
> >> When I tried a larger value "9999999999999999999",
> >> I get the following error:
> >> 
> >>   Arithmetic overflow error: "9999999999999999999"
> >
> > Perhaps the inappropriate error message is platform-specific.
> >
> > I definitely see that message with emacs -Q.
> 
> I get:
> 
> set-variable: Value `1e+19' does not match type integer of scroll-conservatively
> 
> So I can reproduce this bug.

Yes, you can, just in a slightly different way: the user didn't enter
1e+19, either.

bug#6457: 24.0.50; Inappropriate error msg for set-variable

[Eli Zaretskii]

> From: Glenn Morris <rgm@gnu.org>
> Date: Wed, 05 Feb 2014 20:41:29 -0500
> 
> Lars Ingebrigtsen wrote:
> 
> > set-variable: Value `1e+19' does not match type integer of scroll-conservatively
> >
> 
> I guess this is due to the feature whereby values too large for integers
> are read as floats, eg
> 
> http://lists.gnu.org/archive/html/emacs-devel/2010-03/msg00181.html
> 
> If so, I guess it is impossible for lisp code to know whether the user
> originally typed an integer or a float. Therefore it is impossible to do
> anything about this. It frankly doesn't seem worth bothering about
> anyway.

Then "wontfix" it.

bug#6457: 24.0.50; Inappropriate error msg for set-variable

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

>> set-variable: Value `1e+19' does not match type integer of
>> scroll-conservatively
>> 
>> So I can reproduce this bug.
>
> Yes, you can, just in a slightly different way: the user didn't enter
> 1e+19, either.

Yeah, I said "can", not "can't".  >"?

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/

bug#6457: 24.0.50; Inappropriate error msg for set-variable

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

>> If so, I guess it is impossible for lisp code to know whether the user
>> originally typed an integer or a float. Therefore it is impossible to do
>> anything about this. It frankly doesn't seem worth bothering about
>> anyway.
>
> Then "wontfix" it.

I agree that it's a minor thing, but it's an "er what?" moment for the
user.  So if anybody has an idea for how to fix this in a manner that
works across everything that asks for an integer, that would be nice...

Or is this problem only present for `set-variable'?  Then it might not
be worth fixing.

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/

bug#6457: 24.0.50; Inappropriate error msg for set-variable

[Drew Adams]

> Or is this problem only present for `set-variable'?  Then it might
> not be worth fixing.

No, it is a general problem.

It's a problem here because `read-from-minibuffer' reads an
integer numeral as input, and the value read is effectively a
floating-point number.

You can see the same thing happen for (format "%S" 999999999).
That returns the number 999999999.0.

IOW, it is the Lisp reader that reads an integer numeral and
returns a non-integer number: (not (integerp 999999999)).

bug#6457: 24.0.50; Inappropriate error msg for set-variable

[Glenn Morris]

Lars Ingebrigtsen wrote:

> I agree that it's a minor thing, but it's an "er what?" moment for the
> user.

Realistically, how often do you expect people to try to customize
integer variables to values larger than most-positive-fixnum?
"I wonder what will happen if I set this to an absurdly large value"
does not count.

bug#6457: 24.0.50; Inappropriate error msg for set-variable

[Eli Zaretskii]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Cc: drew.adams@oracle.com,  6457@debbugs.gnu.org
> Date: Wed, 05 Feb 2014 22:28:02 -0800
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> >> set-variable: Value `1e+19' does not match type integer of
> >> scroll-conservatively
> >> 
> >> So I can reproduce this bug.
> >
> > Yes, you can, just in a slightly different way: the user didn't enter
> > 1e+19, either.
> 
> Yeah, I said "can", not "can't".  >"?

Sorry, I didn't have enough caffeine in my blood when I wrote that.

bug#6457: 24.0.50; Inappropriate error msg for set-variable

[Lars Ingebrigtsen]

Drew Adams <drew.adams@oracle.com> writes:

>> Or is this problem only present for `set-variable'?  Then it might
>> not be worth fixing.
>
> No, it is a general problem.
>
> It's a problem here because `read-from-minibuffer' reads an
> integer numeral as input, and the value read is effectively a
> floating-point number.
>
> You can see the same thing happen for (format "%S" 999999999).
> That returns the number 999999999.0.
>
> IOW, it is the Lisp reader that reads an integer numeral and
> returns a non-integer number: (not (integerp 999999999)).

I think this has been fixed once and for all in Emacs 27 with the new
bignum support, so I'm closing this bug report.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no