In particular, it only parses fractions, e.g., 100221/19, rather than parsing decimals like 122.12.

The documentation, however, says this:

(** Converts a string to a rational.
    Plain decimals, and [/] separated decimal ratios (with optional sign) are
    understood.
    Additionally, the special [inf], [-inf], and [undef] are recognized
    (they can also be typeset respectively as [1/0], [-1/0], [0/0]).
 *)

The behavior is clear enough from looking at the implementation.

let of_string s =
  try
    let i  = String.index s '/' in
    make
      (Z.of_substring s ~pos:0 ~len:i)
      (Z.of_substring s ~pos:(i+1) ~len:(String.length s-i-1))
  with Not_found ->
    if s = "inf" || s = "+inf" then inf
    else if s = "-inf" then minus_inf
    else if s = "undef" then undef
    else of_bigint (Z.of_string s)

I do think it would be great to support "plain decimals", but I suppose the minimum fix is to update the comment.

z.mli does not mention what happens if the string given to Z.of_string does not correctly specify an integer. It would be nice to be able to catch some exception in that case. Please add error behaviour specification. (Maybe to other functions as well; Z.of_string just is the one that caught my attention.)

Zarith is used in nocrypto. It would be more secure and preferred to use the side-channel resistant mpz_powm_sec instead of mpz_powm.

Reference: Original ticket in nocrypto.

If two numbers are qual by = operator, will it be guaranteed that hash_param still return same results?

• Including the recent fixes on the Q.of_string and Z.of_string

We currently only test linux and macos

now that #79 and #94 are meged, it would be great to have a release of zarith :)

thanks a lot for your work.

q.mli describes the canonical representation of rational numbers, but the type is exposed, so one can construct non-canonical values. Perhaps the type should be made abstract or private to ensure proper behavior?

Right now it seems that they are not installed, which prevents merlin from accessing the documentation of functions

Hello

i have installed zarith with

opam depext zarith
opam install zarith

and with ocaml-top
i have to put this way

#load "/Users/uio/.opam/system/lib/zarith/zarith.cma";;

because #load "zarith.cma";; doesn't work (cannot find file zarith.cma)

with function example test from zarith source

let rec fib_z n =
if n < 2 then Z.one else Z.add (fib_z(n-1)) (fib_z(n-2))

i get

...Z.one Error : Unbound module Z

Here my PATH
echo $PATH
/Users/uio/.opam/system/bin:/Library/Frameworks/Python.framework/Versions/3.6/bin:/usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin:/opt/X11/bin

Thanks

Ubuntu 18.04, opam 2.0, switch is 4.07.0+32bit:

./configure 
binary ocaml: found in /home/egallego/.opam/4.07.0+32bit/bin
binary ocamlc: found in /home/egallego/.opam/4.07.0+32bit/bin
binary ocamldep: found in /home/egallego/.opam/4.07.0+32bit/bin
binary ocamlmklib: found in /home/egallego/.opam/4.07.0+32bit/bin
binary ocamldoc: found in /home/egallego/.opam/4.07.0+32bit/bin
binary ar: found in /usr/bin
binary perl: found in /usr/bin
binary gcc: found in /usr/bin
binary ocamlopt: found in /home/egallego/.opam/4.07.0+32bit/bin
checking compilation with gcc -O3 -Wall -Wextra : working
include caml/mlvalues.h: found
library dynlink.cmxa: found
binary ocamlfind: found in /home/egallego/.opam/4.07.0+32bit/bin
OCaml's word size is 32
binary uname: found in /bin
include gmp.h: found
library gmp: found
OCaml extended comparison supported
OCaml new hash functions available
OCaml supports -bin-annot to produce documentation

detected configuration:

  native-code:          yes
  dynamic linking:      yes
  asm path:             x86_64
  defines:              -DZ_OCAML_HASH -DZ_OCAML_COMPARE_EXT -DHAS_GMP -DZ_ELF -DZ_DOT_LABEL_PREFIX 
  libraries:             -lgmp
  C options:            -O3 -Wall -Wextra 
  asm options           
  installation path:    /home/egallego/.opam/4.07.0+32bit/lib
  installation method   findlib

$ make
project.mak:151: depend: No existe el archivo o el directorio
./z_pp.pl x86_64
Name "main::extra" used only once: possible typo at ./z_pp.pl line 29.
found assembly file caml_z_x86_64.S
  found abs
  found add
  found div
  found divexact
  found logand
  found lognot
  found logor
  found logxor
  found mul
  found neg
  found pred
  found rem
  found shift_left
  found shift_right
  found sub
  found succ
ocamldep -native  z.ml q.ml big_int_Z.ml z.mli q.mli big_int_Z.mli > depend
ocamlc -I +compiler-libs -bin-annot  -c z.mli
ocamlc -I +compiler-libs -bin-annot  -c z.ml
ocamlc -I +compiler-libs -bin-annot  -c q.mli
ocamlc -I +compiler-libs -bin-annot  -c q.ml
ocamlc -I +compiler-libs -bin-annot  -c big_int_Z.mli
ocamlc -I +compiler-libs -bin-annot  -c big_int_Z.ml
ocamlmklib -failsafe -o zarith z.cmo q.cmo big_int_Z.cmo -lgmp
gcc -DZ_OCAML_HASH -DZ_OCAML_COMPARE_EXT -DHAS_GMP -DZ_ELF -DZ_DOT_LABEL_PREFIX     -c -o caml_z_x86_64.o caml_z_x86_64.S
ocamlc -ccopt "-I/home/egallego/.opam/4.07.0+32bit/lib/ocaml  -DZ_OCAML_HASH -DZ_OCAML_COMPARE_EXT -DHAS_GMP -DZ_ELF -DZ_DOT_LABEL_PREFIX  -O3 -Wall -Wextra " -c caml_z.c
ocamlmklib -failsafe -o zarith caml_z_x86_64.o caml_z.o -lgmp
/usr/bin/ld: la arquitectura i386:x86-64 del fichero de entrada `caml_z_x86_64.o' es incompatible con la salida i386
collect2: error: ld returned 1 exit status
ocamlc -I +compiler-libs -bin-annot  -c zarith_top.ml
ocamlc -o zarith_top.cma -a zarith_top.cmo
ocamlopt -I +compiler-libs  -c z.ml
ocamlopt -I +compiler-libs  -c q.ml
ocamlopt -I +compiler-libs  -c big_int_Z.ml
ocamlmklib -failsafe -o zarith z.cmx q.cmx big_int_Z.cmx -lgmp
ocamlopt -shared -o zarith.cmxs -I . zarith.cmxa -linkall
/usr/bin/ld: se salta el ./libzarith.a incompatible mientras se busca -lzarith
/usr/bin/ld: no se puede encontrar -lzarith
collect2: error: ld returned 1 exit status
File "caml_startup", line 1:
Error: Error during linking
project.mak:79: recipe for target 'zarith.cmxs' failed
make: *** [zarith.cmxs] Error 2

I can't make a PR because gh-pages is an orphan branch, but if anyone with admin rights wants to pull from my fork:

• branch gh-pages here: https://github.com/c-cube/Zarith/tree/gh-pages (with generated doc)
• points to https://c-cube.github.io/Zarith/ with the online docs for people who desperately search it online (here it'd be https://ocaml.github.io/Zarith/)

Consider min_int, the smallest value of type int.

When converted to Z.t using Z.of_int, it is represented as a tagged integer:

# let a = Z.of_int min_int;;
val a : Z.t = <abstr>
# Obj.tag (Obj.repr a);;
- : int = 1000

But when min_int is produced as the result of a big int computation, it is represented as a heap-allocated custom block:

# let b = Z.succ (Z.pred a);;
val b : Z.t = <abstr>
# Obj.tag (Obj.repr b);;
- : int = 255

That's because ml_z_reduce in caml_z.c decides to return a tagged integer if abs(result) <= max_int, not if min_int <= result <= max_int.

At first I thought this double representation is harmless, since Zarith functions seem to do the right thing, even equality:

# a = b;;
- : bool = true

However, hashing is confused: (see also #81)

# Hashtbl.hash a;;
- : int = 69827754
# Hashtbl.hash b;;
- : int = 694698810

Also, if it was not for this double representation of min_int, we could optimize the fast path of certain operations. For example, equality Z.equal x y could be

   if is_long(x) || is_long(y) then x == y else (* slow path *)

since a small Z.t and a big Z.t would never be equal, and == returns false when comparing their value representations. Z.compare can also be improved along the same lines.

Opinions?

Maybe once #47 is addressed

And close #50

Coq 8.13 now has a hard dependency on Zarith and can no longer use Num, so Homebrew needs a formula for ocaml-zarith in analogy to ocaml-num.

I did what I could to write that formula at Homebrew/homebrew-core#68494.

Z.root on a negative number raises the sigfpe signal instead of an ocaml exception.

$ utop 
# Z.(root (of_string "-2") 2);;
Floating point exception (core dumped)
$

When given a ridiculously large number, Big_int_Z.power_int_positive_int aborts with the message:

gmp: overflow in mpz type

In some cases, it would be better to be able to capture the error or to terminate in a less abrupt way.

For instance, the following program triggers this behavior:

let () =
  let _ = Big_int_Z.power_int_positive_int 5 999999999999999 in
  ()

It seems the issue is in function ml_z_pow, in caml_z.c, when calling mpz_pow_ui(mbase, mbase, e);

The real issue seems to be upstream in gmp (e.g. https://stackoverflow.com/questions/3558684, https://stackoverflow.com/questions/13328409), but I wonder if something could be done in zarith to raise an OCaml exception instead.

Q is missing a to_float : t -> float function that converts the rational q to the floating point number to_float q that best approximates q.

Hello,
I've been trying to get a Windows build compatible with MSVC ABI (to load a .dll in a MSVC compiled binary).
Has anyone managed to compile gmp/zarith in a setup like this ?

Hi there,

I've been trying to get compilation with Zarith to work, but I'm having some trouble.

I have installed Zarith with the "opam install zarith" command. I also have everything from the Requirements section installed. I'm on Linux Mint 18.

Opam version is 1.2.2 and Ocaml version is 4.02.3.

The commands "opam list" and "ocamlfind list" do have zarith listed as an installed package.

I've tried to compile the code in different ways to no avail. Let's say I have a single line "let x = Z.of_int 4" in a test.ml file.

If I do compilation with "ocamlopt -I +zarith test.ml" I get:
Error: Unbound module Z

With "ocamlfind ocamlopt -package zarith test.ml":
test.o: In function camlTest__entry': test.ml:(.text+0xc): undefined reference to ml_z_of_int'
collect2: error: ld returned 1 exit status
File "caml_startup", line 1:
Error: Error during linking

I also tried "ocamlopt zarith.cmxa test.ml", which gives me:
Error: Cannot find file zarith.cmxa

However, in the OCaml TopLevel, if I call:
#require "zarith";;
#load "zarith.cma";;

I can use the module just fine, inside the TopLevel.

How could I fix this problem with the compilation? Thank you in advance.

Building ZArith with assembly instructions for the fast path on Cygwin64/mingw-w64 fails
due to the instructions for making the stack non-executable that have been introduced by 96a8242.

The GCC version used is x86_64-w64-mingw32-gcc (GCC) 6.4.0, the error message reads:

caml_z_x86_64_mingw64.S: Assembler messages:
caml_z_x86_64_mingw64.S:24: Error: junk at end of line, first unrecognized character is `-'

Reverting the commit solves the issue (but leaves the stack executable).
Just replacing @progbits by %progbits (as proposed in #14) does not solve the issue on Cygwin/mingw-w64.

Given the other bug report #14 that concerns 96a8242, the change appears to be rather non-portable.
Is there a better solution available?

When numbers are small, the result of a call to gcd is positive when one parameter is zero, and the other parameter is negative.

However, when the negative value is "large", it returns a negative number.

At first glance, it appears that the "normalization" in the small-case happens here:
https://github.com/ocaml/Zarith/blob/master/caml_z.c#L1743

  let gcd_helper a b = gcd (Z.of_string a) (Z.of_string b) |> Z.print

  let%expect_test "" =
    gcd_helper "0" "-350854590609916240021621339209265605473999635689";
    [%expect "-350854590609916240021621339209265605473999635689"]
  ;;

  let%expect_test "" =
    gcd_helper "0" "-55";
    [%expect "55"]
  ;;

Consider the following example submitted by @mbarbin in 5463dbc#commitcomment-27324931:

(* q = 7.56181796669062E-309 *)
let q = Q.make (Z.of_string "756181796669062") (Z.pow (Z.of_int 10) (309+14));;

Printf.printf "%h\n" (Q.to_float q);;
(* 0x0.570020d1942p-1022 *)

Printf.printf "%h\n" 7.56181796669062E-309;;
(* 0x0.570020d1941ffp-1022 *)

The result of Q.to_float is not the FP number closest to the exact value.

@pascal-cuoq notes that the "round to odd" technique used in Q.to_float is incorrectly used in this particular case involving denormal FP numbers as the result.

Hi all,

I noticed that Q.compare has the following semantics for comparing with NaN:

let nan = Q.(zero / zero);;

Q.compare nan Q.(neg inf);;
(* -1 *)

Q.compare nan nan;;
(* 0 *)

This was a little surprising to me, as these comparisons imply that NaN is "smaller" than all other bignums, including negative infinity. More specifically, I was expecting a behaviour similar to IEEE 754 floating point, where all inequalities involving NaN return false, and NaN <> any value.

Looking at the code, it seems to me like this behaviour was deliberately chosen:
https://github.com/ocaml/Zarith/blob/cda0bb6/q.ml#L146
Could someone explain the rationale behind having NaN compare in this way?

Thanks in advance!
Cheng

Request extracted from #10

Hi,

For « fun » I decided to try to build Prose (the collaborative text editor) on Windows/Ubuntu through its recent Linux subsystem. Well, for fun and to find out whether this subsystem could handle non-trivial workloads which include builds and packaging.

Build worked but ocsigenserver failed at runtime with the following error message:

Error: Error on dynamically loaded library: /home/prose/.opam/4.04.2/lib/stublibs/dllzarith.so: /home/prose/.opam/4.04.2/lib/stublibs/dllzarith.so: cannot enable executable stack as shared object requires: Invalid argument\n

I don't know if the requirement on an executable stack comes from Zarith or from GMP so maybe I'm filling this issue report in the wrong project.

I tried implementing finite field arithmetic for GF2^n (here and here) but it performs horribly for large integers (e.g. several thousand bits long) due to the need to copy these around rather than mutating the values directly.

It would be nice if there were a mutable API for this purpose, e.g. inside a child module Z.Mutable or something.

Hello everyone,

I was trying to install zarith on my opensuse server however, I'm running into issues wrt gmp

I've also reported this issue here ocaml/opam-repository#12926

Hello!

It would really nice if you could cut a new release.

@antoinemine recently added this commit 96a8242 which fixes an important security issue that caused various memory mappings in programs using zarith to be both executable and writable at the same time, which makes it much easier for hackers to exploit vulnerabilities. Thanks to @antoinemine for spotting and fixing this, appreciate it a lot!

Since Zarith is used in security-critical libraries like https://github.com/mirleft/ocaml-nocrypto that are used to protect network-facing services, the inherited protection flags transitionally weakens the security of every ocaml library and program that use the current zarith releases. I ran into this problem while trying to figure out why my own library suddenly had executable stack.

I tested the patch by pinning the master branch of zarith, and confirmed that it correctly fixes the protection flags. A new release would enable nocrypto to update its dependency on zarith and mitigate the problem.

Is there a fundamental impediment to supporting OCaml's polymorphic comparison operators over Q.t values? Could it be done similarly to how it's done for Z.t values (e.g., via something like ml_z_custom_compare)?

If there's no fundamental issue making this not possible, we could perhaps try to do this.

Many thanks.

This simple program prints res1=1 and res2=0, which is wrong. The second result should be 1 as well. Or am I missing something?

open Q

let by_function a b =
  if Q.lt a b then
    0
  else
    1

let by_infix a b =
  if a < b then
    0
  else
    1

let main () =
  let a = Q.of_float 100.0 in
  let b = of_float 0.001 in
  let res1 = by_function a b in
  let res2 = by_infix a b in
  let _ = Printf.printf "res1=%d\n" res1 in
  let _ = Printf.printf "res2=%d\n" res2 in
  0

let _ = main ()

The ocaml version is 4.02.3
The zarith library is 1.7, installed by opam:

$ opam show zarith
             package: zarith
             version: 1.7
          repository: default
        upstream-url: https://github.com/ocaml/Zarith/archive/release-1.7.tar.gz
       upstream-kind: http
   upstream-checksum: 80944e2755ebb848451a77dc2ad0651b
            homepage: https://github.com/ocaml/Zarith
         bug-reports: https://github.com/ocaml/Zarith/issues
            dev-repo: https://github.com/ocaml/Zarith.git
              author: Antoine Miné, Xavier Leroy, Pascal Cuoq
             depends: ocamlfind & conf-gmp & conf-perl
   installed-version: 1.7 [system]
  available-versions: 1.1, 1.2, 1.3, 1.4, 1.4.1, 1.5, 1.6, 1.7
         description: Implements arithmetic and logical operations over arbitrary-precision integers
The Zarith library implements arithmetic and logical operations over
arbitrary-precision integers. It uses GMP to efficiently implement
arithmetic over big integers. Small integers are represented as Caml
unboxed integers, for speed and space economy.

Decimal floating-point (DFP) arithmetic refers to both a representation and operations on decimal floating-point numbers. Working directly with decimal (base-10) fractions can avoid the rounding errors that otherwise typically occur when converting between decimal fractions (common in human-entered data, such as measurements or financial information) and binary (base-2) fractions.

The advantage of decimal floating-point representation over decimal fixed-point and integer representation is that it supports a much wider range of values. For example, while a fixed-point representation that allocates 8 decimal digits and 2 decimal places can represent the numbers 123456.78, 8765.43, 123.00, and so on, a floating-point representation with 8 decimal digits could also represent 1.2345678, 1234567.8, 0.000012345678, 12345678000000000, and so on. This wider range can dramatically slow the accumulation of rounding errors during successive calculations; for example, the Kahan summation algorithm can be used in floating point to add many numbers with no asymptotic accumulation of rounding error.
Fixed-length decimal point numbers are popular in various business applications, the most notorious example is COBOL written applications.

The Rust community for example, has three crates for dealing with decimal numbers: decimal, rust-decimal, and bigdecimal.

There seems some demand for this type of arithmetic in OCaml world too:

• https://discuss.ocaml.org/t/decimal-floating-point-arithmetic/1447
• https://discuss.ocaml.org/t/exact-rational-numbers/4240

Aside "classic" decimal numbers there are also so called posits (unums) in OCaml - see examples in other languages. There are no benefits of using them while there is no hardware available, nevertheless it might be interesting to play with the software implementation for evaluation reasons.

See also Decimal Point Algorithms for Computers and Software Implementation of DecimalFloating-Point papers.

ml_z_extract have a fast path if its argument is a natint but it could also use the fact that its result will be a natint if len argument is lower than Z_BASE2_LENGTH_OP -- saving the cost of allocating a limb just before reducing it to a natint.

The latest release is quite old and trigger an OCaml warning on recent compilers because .cmx files are not shipped with the library.

unfortunately since this code update I'm not able anymore to install Zarith on Raspberry Debian Jessie :(

this is the error I get, after the command 'make':

pi@raspberrypi:~/.opam/system/build/zarith.1.6 $ make
gcc -DZ_OCAML_HASH -DZ_OCAML_COMPARE_EXT -DHAS_GMP     -c -o caml_z_arm.o caml_z_arm.S
caml_z_arm.S: Assembler messages:
caml_z_arm.S:25: Error: junk at end of line, first unrecognized character is `,'
<eingebaut>: die Regel für Ziel „caml_z_arm.o“ scheiterte
make: *** [caml_z_arm.o] Fehler 1

with earlier versions installation has been working without issues - and after commenting-out this line of code the problem vanishes as well...

do we really need this addition - or can we fix it in order to not break the installation on Debian Jessie?

Thanks for any hints / updates!
best regards.

[edited 11.10.2017 15:55]:
is related to #13

There is appearantly no concept of DESTDIR. One can use --installdir to concat DESTDIR+$prefix, but ocamlfind does not cope with it:

[    4s] + make install DESTDIR=/home/abuild/rpmbuild/BUILDROOT/ocaml-zarith-1.9.1-190925162731.0.x86_64
[    4s] ocamlfind install -destdir "/home/abuild/rpmbuild/BUILDROOT/ocaml-zarith-1.9.1-190925162731.0.x86_64/usr/lib64/ocaml" zarith META zarith.h zarith.cma libzarith.a z.mli q.mli big_int_Z.mli z.cmi q.cmi big_int_Z.cmi zarith_top.cma zarith.a zarith.cmxa z.cmx q.cmx big_int_Z.cmx zarith.cmxs z.cmti q.cmti big_int_Z.cmti -optional dllzarith.so
[    4s] ocamlfind: Bad configuration: Cannot mkdir /home/abuild/rpmbuild/BUILDROOT/ocaml-zarith-1.9.1-190925162731.0.x86_64/usr/lib64/ocaml/zarith because a path component does not exist or is not a directory

Apparently the install -d command must also be used the findlib case.

Hi guys,

When I tried to build and install Zarith, after configure it, it showed me below error.
I'm using windows 10, cygwin + OPAM (which provided by OCaml 64bit windows installation).

Could you give me some suggestions on it? Thank you very much!

The error:

Steve@DESKTOP-HI9GPS4 /cygdrive/d/zarith
$ make
make all-recursive
make[1]: Entering directory '/cygdrive/d/zarith'
Making all in .
make[2]: Entering directory '/cygdrive/d/zarith'
make[2]: Nothing to be done for 'all-am'.
make[2]: Leaving directory '/cygdrive/d/zarith'
Making all in examples
make[2]: Entering directory '/cygdrive/d/zarith/examples'
make[2]: Nothing to be done for 'all'.
make[2]: Leaving directory '/cygdrive/d/zarith/examples'
Making all in lib
make[2]: Entering directory '/cygdrive/d/zarith/lib'
make all-am
make[3]: Entering directory '/cygdrive/d/zarith/lib'
make[3]: Leaving directory '/cygdrive/d/zarith/lib'
make[2]: Leaving directory '/cygdrive/d/zarith/lib'
Making all in src
make[2]: Entering directory '/cygdrive/d/zarith/src'
x86_64-w64-mingw32-gcc.exe -g -O2 -o m4.exe m4.o builtin.o debug.o eval.o format.o freeze.o input.o macro.o output.o path.o symtab.o ../lib/libm4.a
../lib/libm4.a(wait-process.o): In function wait_subprocess': /cygdrive/d/m4/lib/wait-process.c:298: undefined reference to waitpid'
collect2: error: ld returned 1 exit status
make[2]: *** [Makefile:1491: m4.exe] Error 1
make[2]: Leaving directory '/cygdrive/d/zarith/src'
make[1]: *** [Makefile:1506: all-recursive] Error 1
make[1]: Leaving directory '/cygdrive/d/zarith'
make: *** [Makefile:1461：all] Error 2

Cheers,
Steve

https://github.com/janestreet/zarith_stubs_js just got released. How do you feel about adding a section in the readme about it ?

I recently added an OCaml implementation of the Miler-Rabin primality test on Rosetta Code, but I find it kind of unfortunate that I have to depend on Jane Street's Bignum library for the sole purpose of generating a random bignum.

Is this something that could be added to Zarith in the future?

While working on #23, it occurred to me that ZArith's own test suite consistently reports errors on Windows 64 bit builds. I attached the outputs of the zq. exe test binary here:
zarith_test_failures.tar.gz

All tests have been conducted with OCaml 4.06.0 and the latest ZArith on github.

It seems that some floating-point precision is lost on both Windows ports.
Even more disturbing, the results also differ between the Cygwin and the native built.
Platform-specific assembly code does not seem to influence the outcomes.

On ubuntu-20.04 (x86_64) with the 4.11.1+musl+static+flambda zarith fails to build: caml_z.c:32:10: fatal error: gmp.h: No such file or directory. (libgmp-dev is installed and vanilla 4.11.1 works fine)

#=== ERROR while compiling zarith.1.10 ========================================#
# context              2.0.7 | linux/x86_64 | ocaml-variants.4.11.1+musl+static+flambda | git+https://github.com/ocaml/opam-repository.git
# path                 ~/.opam/4.11.1+musl+static+flambda/.opam-switch/build/zarith.1.10
# command              ~/.opam/opam-init/hooks/sandbox.sh build make
# exit-code            2
# env-file             ~/.opam/log/zarith-14525-ab7083.env
# output-file          ~/.opam/log/zarith-14525-ab7083.out
### output ###
# [...]
# ocamlc -I +compiler-libs -bin-annot  -c q.mli
# ocamlc -I +compiler-libs -bin-annot  -c q.ml
# ocamlc -I +compiler-libs -bin-annot  -c big_int_Z.mli
# ocamlc -I +compiler-libs -bin-annot  -c big_int_Z.ml
# ocamlmklib -failsafe -o zarith z.cmo q.cmo big_int_Z.cmo -lgmp
# gcc -DZ_OCAML_HASH -DZ_OCAML_COMPARE_EXT -DHAS_GMP -DZ_ELF -DZ_DOT_LABEL_PREFIX     -c -o caml_z_x86_64.o caml_z_x86_64.S
# ocamlc -ccopt "-I/home/runner/.opam/4.11.1+musl+static+flambda/lib/ocaml  -DZ_OCAML_HASH -DZ_OCAML_COMPARE_EXT -DHAS_GMP -DZ_ELF -DZ_DOT_LABEL_PREFIX  -O3 -Wall -Wextra " -c caml_z.c
# caml_z.c:32:10: fatal error: gmp.h: No such file or directory
#    32 | #include <gmp.h>
#       |          ^~~~~~~
# compilation terminated.
# make: *** [project.mak:142: caml_z.o] Error 2

I'm trying to convert between Z.t and the mpz_ptr C type reified in OCaml by Ctypes, following following my own feature request. My plan was to use zarith's C interface zarith.h since, at the ocaml level, Z.t is abstract.
I'm failing to link

+ ocamlfind ocamlopt -linkpkg -g -package ctypes -package ctypes.foreign -package yices2_bindings -package sexplib -package containers -package ppx_deriving.std -cclib -lyices -ccopt -L/Users/e29546/.opam/4.10.0+flambda/lib/zarith/ -ccopt -I/Users/e29546/.opam/4.10.0+flambda/lib/zarith/ -cclib -lunix -cclib -lasmrun -cclib -lcamlrun -I src_tests src_tests/context_test.cmx src_tests/error_test.cmx src_tests/yices_runtime.cmx -o src_tests/yices_runtime.native
Undefined symbols for architecture x86_64:
  "_ml_z_from_mpz_ml", referenced from:
      _camlYices_runtime__entry in yices_runtime.o
      _camlYices2_high__x_26 in yices2_bindings.a(yices2_high.o)
      _camlZbindings__anon_fn$5bzbindings$2eml$3a9$2c0$2d$2d66$5d_24 in yices2_bindings.a(zbindings.o)
      _camlZbindings in yices2_bindings.a(zbindings.o)
  "_ml_z_mpz_init_set_z_ml", referenced from:
      _camlYices_runtime__entry in yices_runtime.o
      _camlYices2_high__x_26 in yices2_bindings.a(yices2_high.o)
      _camlZbindings__anon_fn$5bzbindings$2eml$3a6$2c0$2d$2d85$5d_17 in yices2_bindings.a(zbindings.o)
      _camlZbindings in yices2_bindings.a(zbindings.o)
  "_ml_z_mpz_set_z_ml", referenced from:
      _camlZbindings__anon_fn$5bzbindings$2eml$3a3$2c0$2d$2d75$5d_10 in yices2_bindings.a(zbindings.o)
      _camlZbindings in yices2_bindings.a(zbindings.o)

My understanding is that the implementation of the three functions offered in zarith.h is not found and, when I see the content of findlib's zarith directory, indeed I see zarith.h and only OCaml's compiled files, so I'm not sure how to link C code with them. As you can see in the command, I've tried many combinations of the C options I gathered from here and there.

Is this related to issue #8? I do not have a dllzarith.* file anywhere...

A small module Zarit_top to be loaded in the toplevel REPL (automatically with #require) would be nice.

Hello,

I see that to_bits and of_bits return or read from little endian ordering. Would it be possible to also add functions for big-endian ordering?

I see that to_bits and of_bits are implemented in the C functions ml_z_to_bits and ml_z_of_bits. I assume adding corresponding big endian functions would entail implementing corresponding C functions?

Could there be another way (without much C) of getting the big endian representation? It seems like GMP uses a big endian representation internally: https://gmplib.org/manual/Raw-Output-Internals.

I am interested in big endian representation for implementing CBOR (RFC8949) in OCaml. In particular to support Bignums.

Thanks!

Trying to build zarith on OS X 10.11.6

$ clang --version
Apple LLVM version 8.0.0 (clang-800.0.42.1)
Target: x86_64-apple-darwin15.6.0
Thread model: posix
InstalledDir: /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin

$ ./configure
binary ocaml: found in /Users/davidlaxer/.opam/default/bin
binary ocamlc: found in /opt/local/bin
binary ocamldep: found in /opt/local/bin
binary ocamlmklib: found in /opt/local/bin
binary ocamldoc: found in /opt/local/bin
binary ar: found in /usr/bin
binary perl: found in /usr/bin
binary gcc: found in /usr/bin
binary ocamlopt: found in /opt/local/bin
checking compilation with gcc -O3 -Wall -Wextra : working
include caml/mlvalues.h: found
library dynlink.cmxa: found
binary ocamlfind: found in /Users/davidlaxer/.opam/default/bin
OCaml's word size is 64
binary uname: found in /usr/bin
checking compilation with gcc -arch x86_64 -O3 -Wall -Wextra : working
include gmp.h: found
library gmp: found
OCaml extended comparison supported
OCaml new hash functions available
OCaml supports -bin-annot to produce documentation

detected configuration:

  native-code:          yes
  dynamic linking:      yes
  asm path:             x86_64
  defines:              -DZ_OCAML_HASH -DZ_OCAML_COMPARE_EXT -DHAS_GMP -DZ_UNDERSCORE_PREFIX -DZ_MACOS 
  libraries:             -lgmp
  C options:            -arch x86_64 -O3 -Wall -Wextra 
  asm options           -arch x86_64 
  installation path:    /Users/davidlaxer/.opam/default/lib
  installation method   findlib

configuration successful!
now type "make" to build
then type "make install" or "sudo make install" to install
(ai) MacBook-Pro:zarith davidlaxer$ make
project.mak:151: depend: No such file or directory
./z_pp.pl x86_64
Name "main::extra" used only once: possible typo at ./z_pp.pl line 29.
found assembly file caml_z_x86_64.S
  found abs
  found add
  found div
  found divexact
  found logand
  found lognot
  found logor
  found logxor
  found mul
  found neg
  found pred
  found rem
  found shift_left
  found shift_right
  found sub
  found succ
ocamldep -native  z.ml q.ml big_int_Z.ml z.mli q.mli big_int_Z.mli > depend
ocamlc -I +compiler-libs -bin-annot  -c z.mli
ocamlc -I +compiler-libs -bin-annot  -c z.ml
ocamlc -I +compiler-libs -bin-annot  -c q.mli
ocamlc -I +compiler-libs -bin-annot  -c q.ml
ocamlc -I +compiler-libs -bin-annot  -c big_int_Z.mli
ocamlc -I +compiler-libs -bin-annot  -c big_int_Z.ml
ocamlmklib -failsafe -o zarith z.cmo q.cmo big_int_Z.cmo -lgmp
gcc -DZ_OCAML_HASH -DZ_OCAML_COMPARE_EXT -DHAS_GMP -DZ_UNDERSCORE_PREFIX -DZ_MACOS  -arch x86_64    -c -o caml_z_x86_64.o caml_z_x86_64.S
ocamlc -ccopt "-I/opt/local/lib/ocaml  -DZ_OCAML_HASH -DZ_OCAML_COMPARE_EXT -DHAS_GMP -DZ_UNDERSCORE_PREFIX -DZ_MACOS  -arch x86_64 -O3 -Wall -Wextra " -c caml_z.c
caml_z.c:2987:3: warning: implicit declaration of function 'mpz_2fac_ui' is
      invalid in C99 [-Wimplicit-function-declaration]
  mpz_2fac_ui(mr, ma);
  ^
caml_z.c:3002:3: warning: implicit declaration of function 'mpz_mfac_uiui' is
      invalid in C99 [-Wimplicit-function-declaration]
  mpz_mfac_uiui(mr, ma, mb);
  ^
caml_z.c:3017:3: warning: implicit declaration of function 'mpz_primorial_ui' is
      invalid in C99 [-Wimplicit-function-declaration]
  mpz_primorial_ui(mr, ma);
  ^
3 warnings generated.
ocamlmklib -failsafe -o zarith caml_z_x86_64.o caml_z.o -lgmp
ld: illegal text-relocation to '___gmp_binvert_limb_table' in /usr/local/lib/libgmp.a(mp_minv_tab.o) from '___gmpn_divexact_1' in /usr/local/lib/libgmp.a(dive_1.o) for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
ocamlc -I +compiler-libs -bin-annot  -c zarith_top.ml
ocamlc -o zarith_top.cma -a zarith_top.cmo
ocamlopt -I +compiler-libs  -c z.ml
ocamlopt -I +compiler-libs  -c q.ml
ocamlopt -I +compiler-libs  -c big_int_Z.ml
ocamlmklib -failsafe -o zarith z.cmx q.cmx big_int_Z.cmx -lgmp
ocamlopt -shared -o zarith.cmxs -I . zarith.cmxa -linkall
ld: illegal text-relocation to '___gmp_binvert_limb_table' in /usr/local/lib/libgmp.a(mp_minv_tab.o) from '___gmpn_divexact_1' in /usr/local/lib/libgmp.a(dive_1.o) for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
File "caml_startup", line 1:
Error: Error during linking
make: *** [zarith.cmxs] Error 2
(ai) MacBook-Pro:zarith davidlaxer$