Re: bignum branch

[Andy Moreton]

On Mon 06 Aug 2018, Tom Tromey wrote:

>>>>>> "Andy" == Andy Moreton <address@hidden> writes:
>
> Andy> purecopy also needs updating tosupport bignums
>
> Did you have a use for this?  It seems like there must not be any
> bignums being dumped currently, since if there were, surely something
> would fail.
>
> Anyway, if you do have a use, could you try the appended?

I don't have a use for this yet - more a case of thing that should be
done before this branch is merged to master.

> Andy> and also the macros in .gdbinit.
>
> This doesn't seem like a must-have to me; but if it is, I think it would
> be best to start by writing pretty-printers to submit to GMP.

GMP already has a gmp_*printf family of functions which should be
suitable for formatted output. Fixing up .gdbinit is nice to have, but
not essential before getting the bignum branch merged into master.

> diff --git a/src/alloc.c b/src/alloc.c
> index 367bb73fc1..dba90e7eb2 100644
> --- a/src/alloc.c
> +++ b/src/alloc.c
> @@ -5535,6 +5535,28 @@ make_pure_float (double num)
>    return new;
>  }
>  
> +/* Value is a bignum object with value VALUE allocated from pure
> +   space.  */
> +
> +static Lisp_Object
> +make_pure_bignum (struct Lisp_Bignum *value)
> +{
> +  Lisp_Object new;
> +  size_t nbytes = value->value[0]._mp_alloc * sizeof (mp_limb_t);

This should use mpz_size.
> +
> +  struct Lisp_Bignum *b = pure_alloc (sizeof (struct Lisp_Bignum), 
> Lisp_Misc);
> +  b->type = Lisp_Misc_Bignum;
> +
> +  /* An mpz_t is an array of one element, so this is the correct way
> +     to copy the contents.  */
> +  b->value[0] = value->value[0];
> +
> +  b->value[0]._mp_d = pure_alloc (nbytes, -1);
> +  memcpy (b->value[0]._mp_d, value->value[0]._mp_d, nbytes);

Use a loop with mpz_getlimbn or mpz_limbs_read to do this with the API.
mpz_roinit_n may be useful instead, using the low level APIs.
See (info "(gmp) Integer Special Functions").

> +
> +  XSETMISC (new, b);
> +  return new;
> +}
>  
>  /* Return a vector with room for LEN Lisp_Objects allocated from
>     pure space.  */
> @@ -5676,6 +5698,8 @@ purecopy (Lisp_Object obj)
>        /* Don't hash-cons it.  */
>        return obj;
>      }
> +  else if (BIGNUMP (obj))
> +    obj = make_pure_bignum (XBIGNUM (obj));
>    else
>      {
>        AUTO_STRING (fmt, "Don't know how to purify: %S");