Using '-1' in a method named '*'

Using '-1' in a method named '*'

[Alejandro Sanchez]

Hello,

I have been trying to write a Guile library for 3D motion and I wrote a ‘vector3’ class and added the method ‘*’ for scaling a vector. This works fine for any scalar, except for ‘-1’. Here is a minimal code example, I am using Guile 2.0.14 on OS X 10.11.6:

(use-modules (oop goops))

(define-class <vector3> ()
  ;; Only one slot, for simlicity
  (x #:init-value 0 #:getter get-x #:init-keyword #:x))

(define-method (*  (n <real>) (v <vector3>))
  ;; Return anything, doesn't matter
  3)


(define v (make <vector3> #:x 1))

(* -1 v)  ; Does not work
(* -2 v)  ; Works fine


The error message is:

<unnamed port>:14:0: In procedure #<procedure 104d7b8e0 at <current input>:14:0 ()>:
<unnamed port>:14:0: In procedure -: Wrong type argument in position 1: #<<vector3> 104ccdce0>

The backtrace is:

     16:0  0 (#<procedure 104db6240 at <current input>:16:0 ()>)

This issue only happens when the argument is -1, it doesn’t happen for other negative integers and it doesn’t happen for ‘-1.0’ either. And it’s not just the literal ‘-1’, if instead I use an expression that evaluates to ‘-1’ like ‘(- 2 3)’ the same happens. However, changing the order of arguments in the definition of the method does work.

The reason for this seems that the expression ‘(* -1 x)’ for any x is re-written by Guile to ‘(- x)’, because by adding the following method it gets working:

(define-method (- (v <vector3>))
  2)

However, since '(* -1 v)’ gets re-written to ‘(- v)’ the result is ‘2' instead of ‘3’. Granted, this is a contrived example and in any scenario I can think of '(* -1 v)’ and ‘(- v)’ would yield the same result, but is this intended behaviour? It was something that got me hung up over the last few days quite a lot (I was already in the process of writing this email when it occurred to me to define the ‘-‘ method).

Re: Using '-1' in a method named '*'

[Andy Wingo]

Hi,

On Mon 27 Feb 2017 11:06, Alejandro Sanchez <hiphish@openmailbox.org> writes:

> (define v (make <vector3> #:x 1))
>
> (* -1 v)  ; Does not work
> (* -2 v)  ; Works fine

I believe that Guile is doing strength reduction, transforming (* -1 v)
to (- 0 v).

It could be that this is totally the wrong thing.  Is (* x 2) -> (+ x x)
a valid transformation if you don't know the type of x?  I don't know.

I think there's currently an assumption that if you extend *, that you
will do so in a mathy way, and that you implement - + and similar.  But
in this case it's not the clear right thing to do.

WDYT?  We could remove this transformation, or only apply it when type
inference has run.

Andy

Re: Using '-1' in a method named '*'

[Alejandro Sanchez]

> It could be that this is totally the wrong thing.  Is (* x 2) -> (+ x x)
> a valid transformation if you don't know the type of x?  I don't know.
Only if both addition and multiplication are defined for the type. In a group-like structure there is only one operation, while ring-like structures have two (usually called addition and multiplication). All the numeric types included in Guile are ring-like, so that works fine. Outside of that all bets are off. What would transforming (* -1 v) to (- 0 v) mean? You cannot add or subtract scalars and vectors, unless you were to embed both into a field where they are both defined, like the quaternion field. Or you would have to use the zero-vector as the value of 0.

> WDYT?  We could remove this transformation, or only apply it when type
> inference has run.
I don’t know. I guess there must be a good reason why Guile does these transformations, all I know is that they are not guaranteed to make sense outside Guile’s numeric types. Maybe it was my mistake for trying to make methods called * or + instead of making up my own names. The reason why I used those names (aside from simplicity) is that in those case it does make sense to mix multiplication of different types. I can write (* 2 3 q 4 5) where q is a quaternion and it will apply the correct operation to all operands, just as if I had written it on paper. It also allows me to use any type that implements * as the component of a dual number and have it work.

If you can stop Guile from re-writing code when it is not certain that the rewritten code makes sense it would be great. Pretty much every test that I have written over the past few days was broken even though my math was correct.

Re: Using '-1' in a method named '*'

[tomas]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Mon, Feb 27, 2017 at 08:21:03PM +0100, Andy Wingo wrote:
> Hi,
> 
> On Mon 27 Feb 2017 11:06, Alejandro Sanchez <hiphish@openmailbox.org> writes:
> 
> > (define v (make <vector3> #:x 1))
> >
> > (* -1 v)  ; Does not work
> > (* -2 v)  ; Works fine
> 
> I believe that Guile is doing strength reduction, transforming (* -1 v)
> to (- 0 v).

Woah. This is somehow... exquisite.

> It could be that this is totally the wrong thing.  Is (* x 2) -> (+ x x)
> a valid transformation if you don't know the type of x?  I don't know.

What we really want to know is the meaning of '*' and '+' (is that akin
to constant folding?), but yes, for that we need some info about the type
of x, I guess.

> I think there's currently an assumption that if you extend *, that you
> will do so in a mathy way, and that you implement - + and similar.  But
> in this case it's not the clear right thing to do.
> 
> WDYT?  We could remove this transformation, or only apply it when type
> inference has run.

If overloading is allowed then yes, it seems type inference has to go
in first?

regards
- -- tomás
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)

iEYEARECAAYFAli1Lh4ACgkQBcgs9XrR2kYpEwCdGS4EaOgUy/tBL46i2Isiz1MZ
sNwAnj+JM+oX7RM6LTjC8+h6r0qnIiBO
=Cc5b
-----END PGP SIGNATURE-----

Re: Using '-1' in a method named '*'

[Andy Wingo]

On Mon 27 Feb 2017 20:21, Andy Wingo <wingo@pobox.com> writes:

> Hi,
>
> On Mon 27 Feb 2017 11:06, Alejandro Sanchez <hiphish@openmailbox.org> writes:
>
>> (define v (make <vector3> #:x 1))
>>
>> (* -1 v)  ; Does not work
>> (* -2 v)  ; Works fine
>
> I believe that Guile is doing strength reduction, transforming (* -1 v)
> to (- 0 v).

I spoke too soon.  Guile 2.1.x does indeed do this at compile-time, but
only if "v" is an instance of a built-in number type.  Guile 2.0.x
doesn't do this strength reduction at all at compile-time.

It turns out there is an easier way to reproduce this bug without GOOPS:

   (* -1 "")

scm_product was handling this case specially because of weird
implementation reasons (making sure that negating the most negative
fixnum was a bignum; see b5c40589).

In summary I think Guile's strength reduction is fine as it relies on
type inference to prove when it is safe.  This is "just" a bug, albeit
an annoying one that we need to fix!

Andy

Re: Using '-1' in a method named '*'

[Andy Wingo]

On Tue 28 Feb 2017 09:54, Andy Wingo <wingo@pobox.com> writes:

> In summary I think Guile's strength reduction is fine as it relies on
> type inference to prove when it is safe.  This is "just" a bug, albeit
> an annoying one that we need to fix!

Fixed in 2.0 and master.  Thanks for the report :)

Andy

Re: Using '-1' in a method named '*'

[Alejandro Sanchez]

> On 28 Feb 2017, at 10:19, Andy Wingo <wingo@pobox.com> wrote:
> 
> On Tue 28 Feb 2017 09:54, Andy Wingo <wingo@pobox.com> writes:
> 
>> In summary I think Guile's strength reduction is fine as it relies on
>> type inference to prove when it is safe.  This is "just" a bug, albeit
>> an annoying one that we need to fix!
> 
> Fixed in 2.0 and master.  Thanks for the report :)
> 
> Andy

Great, than you. Do you have any approximation on when a release with that fix will happen, or am I better off getting the source and building myself? I would rather not mess around with building, but I’ll bit the bullet and build myself if that’s the way to go. I am currently on 2.0.14 because my package manager only offers the latest stable release.

Re: Using '-1' in a method named '*'

[Amirouche]

Le 28/02/2017 à 15:24, Alejandro Sanchez a écrit :
>> On 28 Feb 2017, at 10:19, Andy Wingo <wingo@pobox.com> wrote:
>>
>> On Tue 28 Feb 2017 09:54, Andy Wingo <wingo@pobox.com> writes:
>>
>>> In summary I think Guile's strength reduction is fine as it relies on
>>> type inference to prove when it is safe.  This is "just" a bug, albeit
>>> an annoying one that we need to fix!
>> Fixed in 2.0 and master.  Thanks for the report :)
>>
>> Andy
> Great, than you. Do you have any approximation on when a release with that fix will happen,

>   or am I better off getting the source and building myself? I would rather not mess around with building, but I’ll bit the bullet and build myself if that’s the way to go. I am currently on 2.0.14 because my package manager only offers the latest stable release.
I think git checkout and compiling is worth it.

git clone http://git.sv.gnu.org/r/guile.git

Don't forget to ./configure --prefix=/usr.

Regards,

Amirouche ~ amz3

Re: Using '-1' in a method named '*'

[Ralf Mattes]

On Tue, Feb 28, 2017 at 06:53:33PM +0100, Amirouche wrote:
> 
> 
> Don't forget to ./configure --prefix=/usr.

Unless you run this on a self-compiled syytem that seems like a
very bad idea. You will happily overwrite a distribution provided guile
and during the next system update your package manager will happily
erase your custom guile and install the distribution version.

Binaries installed into /usr/local (the default chosen by autoconf) will
be the first binary found in the path in all distibutions I have seen
so far.

Cheers, Ralf Mattes

> Regards,
> 
> Amirouche ~ amz3
> 

Re: Using '-1' in a method named '*'

[Andy Wingo]

On Tue 28 Feb 2017 15:24, Alejandro Sanchez <hiphish@openmailbox.org> writes:

>> On 28 Feb 2017, at 10:19, Andy Wingo <wingo@pobox.com> wrote:
>> 
>> On Tue 28 Feb 2017 09:54, Andy Wingo <wingo@pobox.com> writes:
>> 
>>> In summary I think Guile's strength reduction is fine as it relies on
>>> type inference to prove when it is safe.  This is "just" a bug, albeit
>>> an annoying one that we need to fix!
>> 
>> Fixed in 2.0 and master.  Thanks for the report :)
>> 
>> Andy
>
> Great, than you. Do you have any approximation on when a release with
> that fix will happen, or am I better off getting the source and
> building myself? I would rather not mess around with building, but
> I’ll bit the bullet and build myself if that’s the way to go. I am
> currently on 2.0.14 because my package manager only offers the latest
> stable release.

2.2.0 should come soon (a couple weeks hopefully), but I don't know
about 2.0.14.  In the meantime if defining an implementation for "-"
works for you to handle negation, that's an OK workaround.

Andy

Re: Using '-1' in a method named '*'

[Alejandro Sanchez]

> On 01 Mar 2017, at 18:25, Andy Wingo <wingo@pobox.com> wrote:
> 
> 2.2.0 should come soon (a couple weeks hopefully), but I don't know
> about 2.0.14.  In the meantime if defining an implementation for "-"
> works for you to handle negation, that's an OK workaround.
> 
> Andy
This may be a stupid question, but how do I build guile? I ran

git clone http://git.sv.gnu.org/r/guile.git

but there is no “configure” file. There is an “automake.sh”, but when I run it it fails with an error

configure.ac:943: error: possibly undefined macro: AM_GNU_GETTEXT
      If this token and others are legitimate, please use m4_pattern_allow.
      See the Autoconf documentation.
autoreconf: /usr/local/Cellar/autoconf/2.69/bin/autoconf failed with exit status: 1

I guess it’s something about my setup. What are those tarballs I can download that have a configure file? Are those like half-built where someone has already run automake for me?

Re: Using '-1' in a method named '*'

[Thomas Morley]

2017-03-02 0:46 GMT+01:00 Alejandro Sanchez <hiphish@openmailbox.org>:
>
>> On 01 Mar 2017, at 18:25, Andy Wingo <wingo@pobox.com> wrote:
>>
>> 2.2.0 should come soon (a couple weeks hopefully), but I don't know
>> about 2.0.14.  In the meantime if defining an implementation for "-"
>> works for you to handle negation, that's an OK workaround.
>>
>> Andy
> This may be a stupid question, but how do I build guile? I ran
>
> git clone http://git.sv.gnu.org/r/guile.git
>
> but there is no “configure” file. There is an “automake.sh”, but when I run it it fails with an error

I did after git clone:

sh autogen.sh
./configure
make -j5
sudo make install

More info in HACKING

HTH,
  Harm

>
> configure.ac:943: error: possibly undefined macro: AM_GNU_GETTEXT
>       If this token and others are legitimate, please use m4_pattern_allow.
>       See the Autoconf documentation.
> autoreconf: /usr/local/Cellar/autoconf/2.69/bin/autoconf failed with exit status: 1
>
> I guess it’s something about my setup. What are those tarballs I can download that have a configure file? Are those like half-built where someone has already run automake for me?

Re: Using '-1' in a method named '*'

[Sandic, Aleksandar]

Thank you, everyone. It is working now, I built from HEAD and negating a vector works as expected. This will bridge me over until 2.2 is released.