rpav / cl-autowrap Goto Github PK

View Code? Open in Web Editor NEW

210.0 210.0 41.0 311 KB

(c-include "file.h") => complete FFI wrapper

License: BSD 2-Clause "Simplified" License

Common Lisp 21.95% C 0.20% Python 77.85%

cl-autowrap's People

Contributors

Stargazers

Watchers

cl-autowrap's Issues

The name of `C-LET` is confusing

I find that the default of c-let is confusing, semantically.

Having agonised over the choice of better names for the :free T/NIL and :ptr variants, I've finally come to this:

The real LET -- equivalent to WITH-FOREIGN-OBJECTS -- C-LET
The 'no release' (:free nil) variant -- C-LET-RETAIN
The pointer variant -- C-LET-PTR.

What do you think?

The current need to specify the :options feels like a kludge..

need something like :include-sources

Often it seems easier to specify a pattern for the sources
to include rather then to exclude. So, either we need
a :include-sources or a way to invert the pattern match for
patterns in :exclude-sources - maybe:

:exclude-sources ("foo.h"
("my_target" :invert))

TIA Ralf Mattes

(setf (c-aref x 1 :int) 10) not working?

With SBCL 1.1.16, following code doesn't work.
(autowrap:with-alloc (i :int)
(setf (autowrap:c-aref i 0 :int)
10)
)

Any hint will be appreciated.

Calling/invoking a function pointer

This might be just a documentation request:

I'm wrapping a structure that contains function pointers

struct my_struct {
...
  char* (*get_name)(my_struct* self);
  int (*get_id)(my_struct* self);
...
}

Give a (wrapped) pointer to such a struct, how would I call the callback functions, preferably without
having to generate accessors.

Creating foreign objects

So, c2ffi looks like the best thing since Quicklisp. I used c2ffi-cffi for a YAML parser I'm building, and decided to try cl-autowrap after I heard it made accessors easy.

The only problem is, I can't create anything of any type with CFFI:WITH-FOREIGN-OBJECT, and libyaml has no equivalent of the GET-FOO function in the example. Does the SFFI abstraction layer have anything that works like it? I found some unexported functions, MAKE-[type name], but they all seem to return null pointers.

An alternative would be to create my own C file with functions, but then this is exactly what I was doing before with c2ffi-cffi to make accessors simpler and I'd prefer to eliminate a dependency on a C compiler for a simple parser.

ffi funcall expands into non-hygienic code

the ffi funcall expansion contains variable bindings with the name that is used in the original ffi function definition. the with this problem is that their value is expressions from user code, so these variables can have surprising interference with user code, unless their name is gensym'ed.

array of array

If I try to c-include a structure with an array of an array like:

#define MAX_ENUM_STRING_SIZE 26
#define MAX_ENUM_STATES 16
struct test_struct{
    char strs[MAX_ENUM_STATES][MAX_ENUM_STRING_SIZE];
};

autowrap::make-array-accessor ends up in an infinite recursion and the stack gets exhausted eventually.

Structure type miswrapping

Actors:

clutter/clutter-types.h, which defines ClutterColor as:
typedef struct _ClutterColor ClutterColor;
clutter/clutter-color.h, which includes clutter/clutter-types.h, and defines
struct _ClutterColor, which has actual content -- four per-component guint8's
the spec file, which misrepresents struct _ClutterColor as having zero fields

Setting:

clutter/clutter-types.h

typedef struct _ClutterColor                    ClutterColor;

clutter/clutter-color.h

#include <clutter/clutter-types.h>
....
struct _ClutterColor
{
  /*< public >*/
  guint8 red;
  guint8 green;
  guint8 blue;

  guint8 alpha;
};

clutter-raw.x86_64-pc-linux-gnu.spec

{ "tag": "struct", "name": "_ClutterColor", "id": 0, "location": "/usr/include/clutter-1.0/clutter/clutter-types.h:80:16", "bit-size": 0, "bit-alignment": 0, "fields": [] },
{ "tag": "typedef", "name": "ClutterColor", "location": "/usr/include/clutter-1.0/clutter/clutter-types.h:80:49", "type": { "tag": ":struct", "name": "_ClutterColor", "id": 0 } },

Class is not yet defined or was undefined; T names a defined constant.

Pulled c2ffi at Feb 18 05:24 (SHA1 value: 9a69c865193e6182cd9fce59cf1772ef883526a7), installed cl-autowrap-20160208-git with quicklisp, using SBCL 1.3.1 and trying to wrap libdrm. Seems like no problem with JSON generation. But. Here is code I run:

(asdf:oos 'asdf:load-op :cffi)
(asdf:oos 'asdf:load-op :cl-autowrap)
(defpackage :libdrm-ffi
  (:use #:common-lisp
    #:cffi)
  (:documentation ""))

(in-package #:libdrm-ffi)
(define-foreign-library libdrm
    (t (:default "libdrm"))) ; /usr/lib/x86_64-linux-gnu/libdrm.so
(use-foreign-library libdrm)
(setf autowrap:*c2ffi-program* #P"/home/y/stand/c2ffi/build/src/c2ffi")
(autowrap:c-include #P"/usr/include/xf86drm.h"
            :sysincludes '("/usr/include/x86_64-linux-gnu/"
                   "/usr/include/libdrm/")
            :spec-path #P"/home/y/dev/drm/autowrap/spec/")

Here is resulting codition:

Execution of a form compiled with errors.
Form:
  (DEFSTRUCT
    (DRM-STATS-T (:CONSTRUCTOR MAKE-DRM-STATS-T) (:CONC-NAME "DRM-STATS-T-")
     (:INCLUDE _DRM_STATS)))
Compile-time error:
  (during macroexpansion of (DEFSTRUCT (DRM-STATS-T # ...)))
Class is not yet defined or was undefined: _DRM_STATS
   [Condition of type SB-INT:COMPILED-PROGRAM-ERROR]

There is no _drm_stats objects of any kind neither in JSON, nor in C sources. Only drm_stats (without prefixed underscore).

And when I evaluate the same autowrap:c-include form again without any changes applied to any code, I suddenly got a lot of errors like this:

  error: 
    T names a defined constant, and cannot be used as a local variable.
    --> PROGN EVAL-WHEN AUTOWRAP:DEFINE-CFUN DEFMACRO PROGN EVAL-WHEN 
    --> SB-C::%DEFMACRO SB-C::%DEFMACRO SB-INT:NAMED-LAMBDA FUNCTION 
    --> SB-C::NAMED-DS-BIND SB-INT:BINDING* 
    ==>
      (LET* ((#:G32226 (SB-C::CHECK-DS-LIST (CDR #:EXPR) 1 1 '(# T)))
             (T (POP #:G32226)))
        (BLOCK LIBDRM-FFI:DRM-HASH-DESTROY
          (LET ((#:!FUN32223 #))
            (WITH-SLOTS (#)
                #:!FUN32223
              (AUTOWRAP::FOREIGN-TO-FFI # '# # #:!FIELDS32224 #)))))

i686-pc-win32 triple no longer valid

i686-pc-win32 should now be i686-pc-windows-msvc
http://llvm.org/viewvc/llvm-project?view=revision&revision=204977

Changing in c2ffi.lisp a pair of occurrences of "-win32" to "-windows-msvc" solves the issue (at least it does for me).

Compilation of cl-sdl2 fails in autowrap.lisp on ARM (Raspberry Pi2 + Linux) with CCL 1.10-r16196

When trying to compile SDL2 I ran into following error:

Invoking restart: Retry compiling #<CL-SOURCE-FILE "sdl2" "autowrap">.
> Error: #<CODE-VECTOR #x15A8D3FE> can't be destructured against the lambda list (AUTOWRAP::NAME AUTOWRAP::C-SYMBOL &KEY AUTOWRAP::VARIADIC-P &ALLOW-OTHER-KEYS), because it does not contain at least 2 elements.
> While executing: CCL::PREPARE-TO-DESTRUCTURE, in process listener(1).
> Type :GO to continue, :POP to abort, :R for a list of available restarts.
> If continued: Skip loading "crawler.lisp"
> Type :? for other options.
1 >

With SBCL it will compile nicely, however starting any cl-sdl2 up on ARM will fail with SBCL because of lack of thread support.
If I can provide more infos or test something else, please let me know.

Make C2FFI optional

I would like cl-autowrap to generate bindings using a pre-existing spec file. So that I can run my software on systems without c2ffi. (I can't compile it on a Fedora 19)

Problem with anonymous union

I use c-include on the following structure with an unnamed union. Every call of c-include produces a
different accessor name in Common Lisp, e.g. (usbdevfs-urb.field-803.stream-id u) or (usbdevfs-urb.field-1187.stream-id u). This is quite inconvenient, because I have to change the Lisp code, in particular when I copy the spec file to a computer without c2ffi.

struct usbdevfs_urb {
        unsigned char type;
        unsigned char endpoint;
        int status;
        unsigned int flags;
        void *buffer;
        int buffer_length;
        int actual_length;
        int start_frame;
        union {
                int number_of_packets;  /* Only used for isoc urbs */
                unsigned int stream_id; /* Only used with bulk streams */
        };
        int error_count;
        unsigned int signr;     /* signal to be sent on completion,
                                  or 0 if none should be sent. */
        void *usercontext;
        struct usbdevfs_iso_packet_desc iso_frame_desc[0];
};

(AUTOWRAP:DEFINE-FOREIGN-RECORD 'USBDEVFS-URB :STRUCT 448 64
                                    '((:TYPE :UNSIGNED-CHAR :BIT-SIZE 8
                                       :BIT-OFFSET 0 :BIT-ALIGNMENT 8)
                                      (:ENDPOINT :UNSIGNED-CHAR :BIT-SIZE 8
                                       :BIT-OFFSET 8 :BIT-ALIGNMENT 8)
                                      (:STATUS :INT :BIT-SIZE 32 :BIT-OFFSET 32
                                       :BIT-ALIGNMENT 32)
                                      (:FLAGS :UNSIGNED-INT :BIT-SIZE 32
                                       :BIT-OFFSET 64 :BIT-ALIGNMENT 32)
                                      (:BUFFER (:POINTER :VOID) :BIT-SIZE 64
                                       :BIT-OFFSET 128 :BIT-ALIGNMENT 64)
                                      (:BUFFER-LENGTH :INT :BIT-SIZE 32
                                       :BIT-OFFSET 192 :BIT-ALIGNMENT 32)
                                      (:ACTUAL-LENGTH :INT :BIT-SIZE 32
                                       :BIT-OFFSET 224 :BIT-ALIGNMENT 32)
                                      (:START-FRAME :INT :BIT-SIZE 32
                                       :BIT-OFFSET 256 :BIT-ALIGNMENT 32)
                                      (#:FIELD-1187
                                       (UNION
                                        (NIL :ID 22 :BIT-SIZE 32 :BIT-ALIGNMENT
                                         32)
                                        (:NUMBER-OF-PACKETS :INT :BIT-SIZE 32
                                         :BIT-OFFSET 0 :BIT-ALIGNMENT 32)
                                        (:STREAM-ID :UNSIGNED-INT :BIT-SIZE 32
                                         :BIT-OFFSET 0 :BIT-ALIGNMENT 32))
                                       :BIT-SIZE 32 :BIT-OFFSET 288
                                       :BIT-ALIGNMENT 32)
                                      (:ERROR-COUNT :INT :BIT-SIZE 32
                                       :BIT-OFFSET 320 :BIT-ALIGNMENT 32)
                                      (:SIGNR :UNSIGNED-INT :BIT-SIZE 32
                                       :BIT-OFFSET 352 :BIT-ALIGNMENT 32)
                                      (:USERCONTEXT (:POINTER :VOID) :BIT-SIZE
                                       64 :BIT-OFFSET 384 :BIT-ALIGNMENT 64)
                                      (:ISO-FRAME-DESC
                                       (:ARRAY
                                        (:STRUCT (USBDEVFS-ISO-PACKET-DESC)) 0)
                                       :BIT-SIZE 0 :BIT-OFFSET 448
                                       :BIT-ALIGNMENT 32)))

Include a way to specify language to c2ffi

I'm writing a wrapper to OpenCV (as all the others are out of date & don't work on windows) and nothing was working properly. I'm simply trying to load up an image with OpenCV's imread function, and that function was missing. running c2ffi on a simple test file that included all the sources gave me a spec file that had everything in it, but when generating with cl-autowrap, it never did.

Finally I tried getting the output, and looking at the temp files it generates, and that is where i found my problem! First, you generate a tmpBLAH.tmp file that has a bunch of c2ffi defs & such, then you generate a tmpBLAH2.tmp that #includes the given file, along with the previous tmpBLAH.tmp file. now when c2ffi runs this, it seems to assume it's plain C, and so if my header is a C++ header instead of a C one (as is the case for OpenCV stuff) you get the following:

In file included from /tmp/tmp6F83SDCV.tmp:1:
In file included from /media/sf_local-projects/opencv-lisp/src/spec/opencv.hpp:46:
/usr/include/opencv2/core.hpp:49:4: error: core.hpp header must be compiled as C++
#  error core.hpp header must be compiled as C++
   ^

etc...

The solution is very simple, take the aforementioned tmpBLAH2.tmp files, rename them to tempBLAH2.hpp and run those through c2ffi by hand w/ the proper arguments! Though it would be nice if the "run those through c2ffi by hand" part would disappear. Maybe it's as simple as changing it so the *.tmp files are named *.<extension-of-the-given-file> instead?

Accessing an element at an index from an array of structs through c-ref

From the c-plus examples:

(c-ref object foo-t :a)              ;; => object.a
(c-ref object foo-t :c 2)            ;; => object.c[2]
(c-ref object foo-t :x 2 :b0)        ;; => object.x[2].b0
(c-ref object foo-t :x :b0)          ;; => object.x[0].b0
(c-ref object foo-t :c *)            ;; => *(object.c)
(c-ref object foo-t :x 1 :s :x &)    ;; => &(object.x[1].s.x)

Is there a simpler way to access an element at an index from an array of structs?

(...)              ;; => object[1].a

Currently, I get a pointer first using c-aptr:

(c-with ((iovec (:struct (c:iovec)) :count 2))
      (c-ref (c-aptr iovec 0 '(:struct (c:iovec))) ;; pointer to struct at index 0
             (:struct (c:iovec)) ;; pointer type
             :iov-len)) ;; member

A cleaner way would be something like:

(c-with ((iovec (:struct (c:iovec)) :count 2))
      (iovec :index 0 :iov-len))

Having some trouble running functions...

I'm in the process of wrapping nuklear and I thought i was good to go but after writing a small test function it barfs at me saying that it can't run functions.

function:

(defun test ()
  (c-with ((ctx nk-ffi:nk-context)
         (lay nk-ffi:nk-panel))
    (format t "Pre: Context ~A; Panel ~A~%" ctx lay)
    (format t "Addresses: ctx ~A lay ~A~%" (ctx &) (lay &))
    (finish-output)
    (nk-begin (ctx &) (lay &) "Demo"
                            (nk-rect 50 50 210 250)
                            (logior nk-ffi:+nk-window-border+
                                    nk-ffi:+nk-window-movable+
                                    nk-ffi:+nk-window-scalable+
                                    nk-ffi:+nk-window-minimizable+
                                    nk-ffi:+nk-window-title+))
    (format t "Post: Context ~A; Panel ~A~%" ctx lay)
    (nk-end ctx)))

error:

$ cl.ccl
Welcome to Clozure Common Lisp Version 1.11-r16635  (LinuxX8664)!

CCL is developed and maintained by Clozure Associates. For more information
about CCL visit http://ccl.clozure.com.  To enquire about Clozure's Common Lisp
consulting services e-mail [email protected] or visit http://www.clozure.com.

? (ql:quickload :nuklear)
To load "nuklear":
  Load 1 ASDF system:
    nuklear
; Loading "nuklear"
Value: 0
Pointer: #<A Foreign Pointer #x7FC6EC0244A0>
Value: 42
; Total of 0 load-time skipped definitions
; Total of 0 load-time missing entities

(:NUKLEAR)
? (nuklear:test)
Pre: Context #<NK-CONTEXT {#X7FC6EC0249D0}>; Panel #<NK-PANEL {#X7FC6EC0239C0}>
Addresses: ctx #<A Foreign Pointer #x7FC6EC0249D0> lay #<A Foreign Pointer #x7FC6EC0239C0>
> Error: Call-by-value not implemented yet for NK-FFI::NK-BEGIN
> While executing: NUKLEAR:TEST, in process listener(1).
> Type :POP to abort, :R for a list of available restarts.
> Type :? for other options.
1 >

And whats weird is that nk-begin shouldn't be in the nk-ffi package but nk-ffi.fns. Everything else seems to work fine and I'm basing my code off of the other autowrapped packages you've built, so I'm confused.

The full code is here: https://github.com/serialhex/cl-nuklear and it's barebones as I'm building it up, but I wanted to get it working before I get it finished...

Thanks in advance!

libffi shared library name

When loading cl-autowrap/libffi, I get an error about not being able to open the shared library "libffi"

Latest libffi by default will build out as libffi-6

If you see CFFI's libffi loading, their form looks like this:

(define-foreign-library (libffi)
  (:darwin (:or "libffi.dylib" "libffi32.dylib" "/usr/lib/libffi.dylib"))
  (:solaris (:or "/usr/lib/amd64/libffi.so" "/usr/lib/libffi.so"))
  (:openbsd "libffi.so")
  (:unix (:or "libffi.so.6" "libffi32.so.6" "libffi.so.5" "libffi32.so.5"))
  (:windows (:or "libffi-6.dll" "libffi-5.dll" "libffi.dll"))
  (t (:default "libffi")))

(see here)

Autowrap's looks like this:

(cffi:define-foreign-library libffi
  (:darwin (:or (:framework "libffi") (:default "libffi")))
  (:unix "libffi.so")
  (:windows "libffi.dll")
  (t (:default "libffi")))

(cffi:use-foreign-library libffi)

I think autowrap should use the same form CFFI does, for completeness

c-ref expansion contains literal CLOS objects

form:

(c-ref hci-filter hci-filter :event-mask)

expansion:

(LET ((#:WRAPPER1448 HCI-FILTER))
  (CFFI-SYS:%MEM-REF (CFFI-SYS:INC-POINTER (PTR #:WRAPPER1448) 4) #<AUTOWRAP:FOREIGN-ALIAS UINT32-T {10075361E3}>)

hci-filter FFI def:

(AUTOWRAP:DEFINE-FOREIGN-RECORD 'HCI-FILTER :STRUCT 128 32
                                    '((:TYPE-MASK UINT32-T :BIT-SIZE 32 :BIT-OFFSET 0 :BIT-ALIGNMENT 32)
                                      (:EVENT-MASK (:ARRAY UINT32-T 2) :BIT-SIZE 64 :BIT-OFFSET 32 :BIT-ALIGNMENT 32)
                                      (:OPCODE UINT16-T :BIT-SIZE 16 :BIT-OFFSET 96 :BIT-ALIGNMENT 16)))

trace:

  0: (PLUS-C::BUILD-REF :EVENT-MASK #<AUTOWRAP:FOREIGN-ALIAS HCI-FILTER {100753CA03}> (PTR #:WRAPPER1448) NIL)
    1: (PLUS-C::BUILD-REF :EVENT-MASK #<AUTOWRAP:FOREIGN-RECORD HCI-FILTER {1007ACFF83}> (PTR #:WRAPPER1448) NIL)
      2: (PLUS-C::BUILD-REF NIL #<AUTOWRAP:FOREIGN-ARRAY TO-TYPE:UINT32-T {1008E8B663}> (CFFI-SYS:INC-POINTER (PTR #:WRAPPER1448) 4) NIL)
      2: PLUS-C::BUILD-REF returned (CFFI-SYS:%MEM-REF (CFFI-SYS:INC-POINTER (PTR #:WRAPPER1448) 4) #<AUTOWRAP:FOREIGN-ALIAS UINT32-T {10075361E3}>)
    1: PLUS-C::BUILD-REF returned (CFFI-SYS:%MEM-REF (CFFI-SYS:INC-POINTER (PTR #:WRAPPER1448) 4) #<AUTOWRAP:FOREIGN-ALIAS UINT32-T {10075361E3}>)
  0: PLUS-C::BUILD-REF returned (CFFI-SYS:%MEM-REF (CFFI-SYS:INC-POINTER (PTR #:WRAPPER1448) 4) #<AUTOWRAP:FOREIGN-ALIAS UINT32-T {10075361E3}>)

Wrapping my head around autoharp

I know I am missing something...

So: I installed c2ffi and it appears to be working as advertised. Next I am trying to see how c-autowrap works.

Here is my .h file (straight from the README):

// -*- Mode: C++ -*-

/* test.h - abbreviated from example */

typedef struct foo {
  int a, b;
  char c[3];

  struct {
    unsigned int b0 : 2, b1 : 3;

    struct {
      char x, y;
    } s;
  } x[2];
} foo_t;

foo_t* get_foo();
// foo_t* get_foo(int, int);
void free_foo(foo_t *foo);
int* get_int();

Here is my CL file:

;;;; -*- Mode: Lisp -*-

(in-package "CL-AUTOWRAP-TESTS")

(eval-when (:load-toplevel :compile-toplevel :execute)
  (setf autowrap:*c2ffi-program*
        "/Users/marcoxa/Projects/Software/c2ffi/c2ffi/build/bin/c2ffi")
  )

(autowrap:c-include "cl-autowrap-test.h")

If I load this file, I obtain a bunch of .spec files, which contain only [ ]. Nothing else appears to be generated. This is the transcript:

CL-USER 8 > (load "cl-autowrap-test.lisp")
; Loading text file /Users/marcoxa/Projects/Lang/CL/tests/cl-autowrap/cl-autowrap-test.lisp
; Total of 0 compile-time skipped definitions
; Total of 0 compile-time missing entities
; Total of 0 load-time skipped definitions
; Total of 0 load-time missing entities
#P"/Users/marcoxa/Projects/Lang/CL/tests/cl-autowrap/cl-autowrap-test.lisp"

CL-USER 9 > (apropos "GET-FOO")

CL-USER 10 >

I am obviously missing something, but what?

Help please

Thank you

Marco

Autowrap/CFFI: the split world

Sometimes the impedance mismatch between CFFI and Autowrap abstractions comes to bite.

As an example, the following does not work, so it appears as if one has to resort to CFFI:WITH-FOREIGN-OBJECTS:

(c-dxlet ((argc :int)
                (argv (:pointer (:string))))
        (setf argc 0
              argv 0)
        (clutter-init (argc &) (argv &)))

Another instance is checking for the NULL pointer -- CFFI:NULL-POINTER-P does not work, as one has to unwrap the objects with AUTOWRAP:PTR.

autowrap:define-bitmask fails under a 32-bit lisp

Something is amiss when using autowrap:define-bitmask in a 32-bit environment. Somehow (<-2 nil nil) is getting evaluated instead of what would normally be expected.

Tried under sbcl 1.0.58 and Clozure 1.9.

Name conversion heuristics: behaves like one : -)

The following names are what I get when wrapping Glib:

;   G-BOOKMARK-FILE-SET-VISITED G-CCLOSURE-MARSHAL-BOOLEAN--BOXED-BOXE-DV
;   G-CCLOSURE-MARSHAL-BOOLEAN--FLAG-SV
;   G-CCLOSURE-MARSHAL-STRING--OBJECT-POINTE-RV
;   G-CCLOSURE-MARSHAL-VOID--BOOLEA-NV G-CCLOSURE-MARSHAL-VOID--BOXE-DV
;   G-CCLOSURE-MARSHAL-VOID--CHA-RV G-CCLOSURE-MARSHAL-VOID--DOUBL-EV
;   G-CCLOSURE-MARSHAL-VOID--ENU-MV G-CCLOSURE-MARSHAL-VOID--FLAG-SV
;   G-CCLOSURE-MARSHAL-VOID--FLOA-TV G-CCLOSURE-MARSHAL-VOID--IN-TV
;   G-CCLOSURE-MARSHAL-VOID--LON-GV G-CCLOSURE-MARSHAL-VOID--OBJEC-TV
;   G-CCLOSURE-MARSHAL-VOID--PARA-MV G-CCLOSURE-MARSHAL-VOID--POINTE-RV
;   G-CCLOSURE-MARSHAL-VOID--STRIN-GV G-CCLOSURE-MARSHAL-VOID--UCHA-RV
;   G-CCLOSURE-MARSHAL-VOID--UIN-TV G-CCLOSURE-MARSHAL-VOID--UINT-POINTE-RV
;   G-CCLOSURE-MARSHAL-VOID--ULON-GV G-CCLOSURE-MARSHAL-VOID--VARIAN-TV
;   G-CCLOSURE-MARSHAL-VOID--VOI-DV

Erroneous incomplete type "struct uiInitOptions" when wrapping libui

This doesn't reproduce with a minimal testcase of:

typedef struct foo foo;

struct foo {
    int x;
};

But does happen with the header here: https://github.com/andlabs/libui/blob/80b8fddbea397a7970532b5ac44c552d90cb3c81/ui.h

The .spec file generated seems to think uiInitOptions an incomplete type: https://gist.github.com/anonymous/bb9764df433cc262635c24cb7fedd663 I don't know if this is a c2ffi issue or a cl-autowrap issue as I haven't delved into the details of the implementation.

FOREIGN-TYPE-SIZE seems not wrapped

Hey,

(autowrap:c-include "/usr/include/lzma/LzmaDec.h") ;; this works
(foreign-type-size '(:struct c-lzma-enc-props))

This complains: Unknown CFFI type (:STRUCT C-LZMA-ENC-PROPS)

The structure is wrapped correctly and the accessor works, but it looks like the defctype or defcstruct was not called. Because, if it was called, then FOREIGN-TYPE-SIZE would not complain, correct?

Does not build on SBCL - There is no class named AUTOWRAP:WRAPPER.

I get this error:

There is no class named AUTOWRAP:WRAPPER.

It's from compiling sffi.lisp.

Swank autodoc problems

I could've sworn it worked before! I am not getting any Emacs autodoc function parameter help with an autowrapped library. Watching slime-events shows that autodoc responds with 'not available'. Is there a trick to get it working, or is it just not supported?
Note that symbol completions work fine.

DEFINE-FOREIGN-TYPE callers followed by DEFINE-ACCESSOR leads to failure

Good day Ryan!

It appears that there is a load/compile time issue wrt. ordering
of funcalled type definers vs. the macroexpansion of DEFINE-ACCESSORS.
For example, in expasion of C-INCLUDE:

    (AUTOWRAP::DEFINE-FOREIGN-RECORD 'SIGACTION :STRUCT 1216 64
                                     '((__SIGACTION_HANDLER
                                        (UNION
                                         (NIL :ID 27 :BIT-SIZE 64
                                          :BIT-ALIGNMENT 64)
                                         (SA-HANDLER __SIGHANDLER_T :BIT-SIZE
                                          64 :BIT-OFFSET 0 :BIT-ALIGNMENT 64)
                                         (SA-SIGACTION (:POINTER (:VOID))
                                          :BIT-SIZE 64 :BIT-OFFSET 0
                                          :BIT-ALIGNMENT 64))
                                        :BIT-SIZE 64 :BIT-OFFSET 0
                                        :BIT-ALIGNMENT 64)
                                       (SA-MASK __SIGSET_T :BIT-SIZE 1024
                                        :BIT-OFFSET 64 :BIT-ALIGNMENT 64)
                                       (SA-FLAGS :INT :BIT-SIZE 32 :BIT-OFFSET
                                        1088 :BIT-ALIGNMENT 32)
                                       (SA-RESTORER (:POINTER (:VOID))
                                        :BIT-SIZE 64 :BIT-OFFSET 1152
                                        :BIT-ALIGNMENT 64)))
 ...
 (AUTOWRAP:DEFINE-ACCESSORS (:STRUCT (SIGACTION))
                            #<PACKAGE "COMMON-LISP-USER">)

Passing and returning structs by value

I'm trying to autowrap Chipmunk2D, a C library for 2D physics simulation: https://github.com/slembcke/Chipmunk2D

Many functions return and receive by value a structure that represents a 2D vector (cpVect if you are interested).

In the FAQs you mention to let you know if passing structs by value is necessary so you can prioritize this issue :D

What do you think? Is it doable?

Thank you!

Recent commit breaks constant wrapping

SHA: 85c687c breaks constant wrapping,
as is described in the inline comments in the commit.

call-by-value not yet implemented...

I wrapped a C library, and trying to call a function results in
; Evaluation aborted on #<SIMPLE-ERROR "Call-by-value not implemented yet for ~S" {...}>.

Any suggestions?

P.S. Even simple calls to functions that take no arguments result in this, so I have trouble believing that it is somehow related to struct-by-value problems in an earlier issue...

c-including /usr/include/lzma/LzmaEnc.h fails

(autowrap:c-include "/usr/include/lzma/LzmaEnc.h") fails with error code 134 for me.

This is from freshly compiled branch clang-4.0.0.

https://pastebin.com/0TmJKZxz <- logs from running the c2ffi command manually. Last line:

c2ffi: /usr/lib/llvm-4.0/include/clang/AST/Expr.h:1532: llvm::StringRef clang::StringLiteral::getString() const: Assertion `CharByteWidth==1 && "This function is used in places that assume strings use char"' failed.

Problem with calling ioctl

I think the code for the variadic argument of an ioctl call isn't properly passed through the macros.

My goal is to replace libusb with pure lisp. I managed to parse the ioctl constant definitions with the following code:

(ql:quickload :cl-autowrap)
(eval-when (:compile-toplevel :execute :load-toplevel)
  (defparameter *spec-path* (merge-pathnames "stage/sb-look-ma-no-libusb/"
                         (user-homedir-pathname))))

(progn
  (with-open-file (s "/tmp/usb0.h"
             :direction :output
             :if-does-not-exist :create
             :if-exists :supersede)
    (format s "#include \"/usr/include/linux/usbdevice_fs.h\"~%"))
  (autowrap::run-check autowrap::*c2ffi-program*
               (autowrap::list "/tmp/usb0.h"
                       "-D" "null"
                       "-M" "/tmp/usb_macros.h"
                       "-A" "x86_64-pc-linux-gnu"
                       ))


  (with-open-file (s "/tmp/usb1.h"
             :direction :output
             :if-does-not-exist :create
             :if-exists :supersede)

    (format s "#include <sys/types.h>~%")
    (format s "#include <sys/stat.h>~%")
    (format s "#include <sys/ioctl.h>~%")
    (format s "#include \"/tmp/usb0.h\"~%")
    (format s "#include \"/tmp/usb_macros.h\"~%")))

(autowrap:c-include "/tmp/usb1.h"
            :spec-path *spec-path*
            :exclude-arch ("arm-pc-linux-gnu"
                   "i386-unknown-freebsd"
                   "i686-apple-darwin9"
                   "i686-pc-linux-gnu"
                   "i686-pc-windows-msvc"
                   "x86_64-apple-darwin9"
                    ;"x86_64-pc-linux-gnu"
                   "x86_64-pc-windows-msvc"
                   "x86_64-unknown-freebsd")
            :exclude-sources ("/usr/include/linux/types.h"
                      "/usr/include/linux/magic.h")
            :include-sources ("/usr/include/linux/ioctl.h")
            :trace-c2ffi t)

(defun usb-control-msg (fd requesttype request value index buffer &key (timeout-ms 1000))
  (declare (type (unsigned-byte 8) requesttype request)
       (type (unsigned-byte 16) value index)
       (type (unsigned-byte 32) timeout-ms))
  (let ((n (length buffer)))
    (declare (type (unsigned-byte 16) n))
    (sb-sys:with-pinned-objects (buffer)
      (autowrap:with-alloc (c '(:struct (usbdevfs-ctrltransfer)))
       (setf (usbdevfs-ctrltransfer.b-request-type c) requesttype
         (usbdevfs-ctrltransfer.b-request c) request
         (usbdevfs-ctrltransfer.w-value c) value
         (usbdevfs-ctrltransfer.w-index c) index
         (usbdevfs-ctrltransfer.w-length c) n
         (usbdevfs-ctrltransfer.data c) (sb-sys:vector-sap buffer)
         (usbdevfs-ctrltransfer.timeout c) timeout-ms)

       (assert (= 0  (CFFI-SYS:%FOREIGN-FUNCALL "ioctl"
                        (:INT fd :UNSIGNED-LONG +USBDEVFS-CONTROL+
                              :POINTER (AUTOWRAP:PTR C) :INT)
                        :CONVENTION :CDECL)))))
    buffer))

I would expect that the following line should work instead of my call to cffi-sys:%foreign-funcall:

(ioctl fd +usbdevfs-control+ (autowrap:ptr c)

However this macro-expands as shown below, i.e. the type for the variadic argument c is missing:

(LET ((#:__FD919 FD))
  (LET ((#:__REQUEST920 +USBDEVFS-CONTROL+))
    (CFFI-SYS:%FOREIGN-FUNCALL "ioctl"
                               (:INT #:__FD919 :UNSIGNED-LONG #:__REQUEST920
                                (AUTOWRAP:PTR C) :INT)
                               :CONVENTION :CDECL)))

Here are the software versions, I'm using:

This is SBCL 1.3.5.24-fdc84dc
Linux localhost 4.5.4-1-ARCH #1 SMP PREEMPT Wed May 11 22:21:28 CEST 2016 x86_64
 GNU/Linux
quicklisp/dists/quicklisp/software/cl-autowrap-20160421-git/

Obscurity of a successful library wrapping

What do you get, when CL-AUTOWRAP succeeds?

Currently it is a Lisp image with the C library loaded and the
corresponding Lisp artifacts defined.

The problem with this is that this is rather obscure.

It would be nicer to have some text generated, as this will facilitate:

visual enumerability
jump-to-source

Latest commit breaks ZMQ4L

I'm going to take a look into this myself, but I just wanted to share: The latest commit (with hash cd30484) breaks ZMQ4L with the following error:

[package zmq4]..> Error: :UNSIGNED-INT can't be destructured against the lambda list (AUTOWRAP::ANON-RECORD-TYPE AUTOWRAP::ANON-PARAMS COMMON-LISP:&REST AUTOWRAP::ANON-FIELD-LIST), because it does not contain at least 2 elements.
> While executing: CCL::PREPARE-TO-DESTRUCTURE, in process listener(1).

I'm going to take a look into fixing this, but you as the creator & maintainer of the package might know exactly what is broken by looking at it!

Unable to exclude a symbol from an included source.

My project is autowrapped-ode and my use case is as follows:

ODE comes in double or single float precision flavours
ODE defines its base floating point type as dReal which is float or double depending on the above
ODE defines all of its other types in terms of dReal
I can detect the precision using this code
The above means I can define dReal correctly using this code

However c-include then redefines dReal and I can see a way to exclude it. The reason is that #'read-parse-forms has the following condition.

(and (or (included-p name exclude-definitions)
         (included-p location exclude-sources))
     (not (or (included-p name include-definitions)
              (included-p location include-sources))))

Which means that if I included the source I cant exclude a symbol.

One possibility is we could allow :include-sources to be used like this:

:include-sources ("odeinit.h"
                  ("common.h" :except "dReal"))

If this is something you would like but do not have time to implement I think I can add this. Please let me know.

Thanks again for this incredible project

:exclude-sources does not really seem to work sometimes

While attempting to bind clutter, like this:

(autowrap:c-include "clutter-raw.h"
                      :exclude-arch ("i686-pc-linux-gnu"
                                     "i686-pc-win32"
                                     "x86_64-pc-win64"
                                     "i686-apple-darwin9"
                                     "x86_64-apple-darwin9")
                      :exclude-sources ("/usr/include/clutter-1.0/clutter/clutter-keysyms.h")
                      :sysincludes '("/usr/include/clutter-1.0/clutter"
                                     "/usr/include/clutter-1.0"
                                     "/usr/include/cogl"
                                     "/usr/include/pango-1.0"
                                     "/usr/include/atk-1.0"
                                     "/usr/include/glib-2.0"
                                     "/usr/include/json-glib-1.0"
                                     "/usr/lib64/glib-2.0/include"
                                     "/usr/include/cairo"
                                     "/usr/include/freetype2"
                                     "/usr/include/libpng15"
                                     "/usr/include/pixman-1"))

..where clutter-raw.h is:

#include <clutter/clutter.h>
#include <cogl/cogl.h>
#include <GL/gl.h>

..the :exclude-sources directive does not seem to have effect,
as the misconstrued constant definitions still seem to seep through
(and deliver the damage.. but it's another issue..)

`autowrap:enum-key` not working

It looks like autowrap:enum-key doesn't work as documented.

(autowrap:enum-key 'log-level 1)

Note: skipping map value 1 of enum LOG-LEVEL to value due to
undefined foreign type: LOG-LEVEL
   [Condition of type AUTOWRAP:UNDEFINED-FOREIGN-TYPE-CONTEXTUALISED]

Looking at the code, enum-key calls require-type which calls
find-type which consults *foreign-types*, a hash whose keys are lists.
Even stranger, the entries for enums look like (:enum (enum-name)).

CL-AUTOWRAP vs. Clutter: some breakage

Good day!

I'm trying to make cl-autowrap generate bindings for libclutter & family.

I've created a cl-autowrap.lisp file, as per the CL-AUTOWRAP documentation,
as can be seen in the following repository:

https://github.com/deepfire/cffi-clutter/tree/cl-autowrap

..and then I do:

sbcl --eval '(require :cl-autowrap)' --eval '(load (compile-file "cl-autowrap.lisp"))'

The binding generation fails in many ways, as is reflected in the attached file.

Reporting of skipped records

During compilation cl-autowrap reports skipped records:

; compiling (AUTOWRAP:C-INCLUDE (QUOTE #) ...); Note: skipping record
                                              ; :struct #:ANON-TYPE-4
                                              ; field :__POS of type
                                              ; __OFF64_T due to undefined
                                              ; foreign type: __OFF64_T
; Total of 0 compile-time skipped definitions
; Total of 1 compile-time missing entities:
;   __OFF64_T

It would be helpful for debugging if this note would include the name of the
file that introduced that record (esp. useful for anon-types).

Symbol conflicts with CFFI

The following is what is required to use both CFFI and AUTOWRAP packages:

  (:shadowing-import-from :autowrap
                          #:foreign-type-size #:foreign-pointer
                          #:defcallback #:callback
                          #:define-foreign-type)

Autowrap of nfs/termbox fails to include structs

While autowrapping termbox (https://github.com/nsf/termbox), I can't get any structs to be wrapped, resulting in a compile-time error and the following note from SBCL:

Note: skipping determine the basic type of foreign type TB-EVENT due to
undefined foreign type: TB-EVENT

All that is in my primary Lisp file is the following:

(asdf:load-system :cl-plus-c)
(use-package :plus-c)
(cffi-sys:%load-foreign-library
 :libtest #p"/home/username/termbox-directory/usr/local/lib/libtermbox.so")
(autowrap:c-include "/home/username/termbox-directory/usr/local/include/termbox.h")
(plus-c:c-let ((event tb-event :free t))
       (print event))

I'm probably doing something wrong, but for the life of me I can't determine what.

Argument X is not a REAL: NIL

Hi.
I'm trying to wrap libdrm. Here is where autowrap stucks:

Argument X is not a REAL: NIL
   [Condition of type SIMPLE-TYPE-ERROR]

Restarts:
 0: [RETRY] Retry SLIME REPL evaluation request.
 1: [*ABORT] Return to SLIME's top level.
 2: [ABORT] abort thread (#<THREAD "new-repl-thread" RUNNING {10078B2363}>)

Backtrace:
  0: (SB-KERNEL:TWO-ARG-> NIL 0)
      Locals:
        SB-DEBUG::ARG-0 = NIL
        SB-DEBUG::ARG-1 = 0
  1: (SB-VM::GENERIC->)
      [No Locals]
  2: (AUTOWRAP:DEFINE-FOREIGN-ENUM #:ANON-TYPE-1279 NIL ((:ADD-COMMAND . 0) (:RM-COMMAND . 1) (:INST-HANDLER . 2) (:UNINST-HANDLER . 3)))
      Locals:
        SB-DEBUG::ARG-0 = #:ANON-TYPE-1279
        SB-DEBUG::ARG-1 = NIL
        SB-DEBUG::ARG-2 = ((:ADD-COMMAND . 0) (:RM-COMMAND . 1) (:INST-HANDLER . 2) (:UNINST-HANDLER . 3))
  3: (AUTOWRAP::%ENSURE-TYPE (AUTOWRAP::ENUM (NIL :ID 23) (:ADD-COMMAND . 0) (:RM-COMMAND . 1) (:INST-HANDLER . 2) (:UNINST-HANDLER . 3)) "record ~(~S~) ~S field ~S of type ~S" (:STRUCT DRM-CONTROL :FUNC (..
      Locals:
        SB-DEBUG::ARG-0 = (AUTOWRAP::ENUM (NIL :ID 23) (:ADD-COMMAND . 0) (:RM-COMMAND . 1) (:INST-HANDLER . 2) (:UNINST-HANDLER . 3))
        SB-DEBUG::ARG-1 = "record ~(~S~) ~S field ~S of type ~S"
        SB-DEBUG::ARG-2 = (:STRUCT DRM-CONTROL :FUNC (AUTOWRAP::ENUM (NIL :ID 23) (:ADD-COMMAND . 0) (:RM-COMMAND . 1) (:INST-HANDLER . 2) (:UNINST-HANDLER . 3)))
  4: (AUTOWRAP::PARSE-RECORD-FIELDS :STRUCT DRM-CONTROL ((:FUNC (AUTOWRAP::ENUM # # # # #) :BIT-SIZE 32 :BIT-OFFSET 0 ...) (:IRQ :INT :BIT-SIZE 32 :BIT-OFFSET 32 ...)))
      Locals:
        SB-DEBUG::ARG-0 = :STRUCT
        SB-DEBUG::ARG-1 = DRM-CONTROL
        SB-DEBUG::ARG-2 = ((:FUNC (AUTOWRAP::ENUM (NIL :ID 23) (:ADD-COMMAND . 0) (:RM-COMMAND . 1) (:INST-HANDLER . 2) (:UNINST-HANDLER . 3)) :BIT-SIZE 32 :BIT-OFFSET 0 ...) (:IRQ :INT :BIT-SIZE 32 :BIT-OFFSET 32 ...))
  5: ((LAMBDA NIL :IN AUTOWRAP:DEFINE-FOREIGN-RECORD))
  6: (AUTOWRAP::CALL-WITH-WRAP-ATTEMPT DRM-CONTROL #<CLOSURE (LAMBDA NIL :IN AUTOWRAP:DEFINE-FOREIGN-RECORD) {10081B23CB}> "record (~A (~S))" (:STRUCT DRM-CONTROL))

Here is a piece of code which causes this condition:

struct drm_control {
    enum {
        DRM_ADD_COMMAND,
        DRM_RM_COMMAND,
        DRM_INST_HANDLER,
        DRM_UNINST_HANDLER
    } func;
    int irq;
};

And respective JSON code, generated by autowrap:

{ "tag": "struct",
  "ns": 0,
  "name": "drm_control",
  "id": 0,
  "location": "/usr/include/libdrm/drm.h:164:8",
  "bit-size": 64,
  "bit-alignment": 32,
  "fields": [{ "tag": "field",
           "name": "func",
           "bit-offset": 0,
           "bit-size": 32,
           "bit-alignment": 32,
           "type": { "tag": "enum",
                     "ns": 1936876800,
             "name": "",
             "id": 23,
             "location": "/usr/include/libdrm/drm.h:165:2",
             "fields": [{ "tag": "field",
                      "name": "DRM_ADD_COMMAND",
                      "value": 0 },
                    { "tag": "field",
                      "name": "DRM_RM_COMMAND",
                      "value": 1 },
                    { "tag": "field",
                      "name": "DRM_INST_HANDLER",
                      "value": 2 },
                    { "tag": "field",
                      "name": "DRM_UNINST_HANDLER",
                      "value": 3 }] } },
             { "tag": "field",
           "name": "irq",
           "bit-offset": 32,
           "bit-size": 32,
           "bit-alignment": 32,
           "type": { "tag": ":int",
                     "bit-size": 32,
             "bit-alignment": 32 } }] },

{ "tag": "enum",
  "ns": 24,
  "name": "",
  "id": 23,
  "location": "/usr/include/libdrm/drm.h:165:2",
  "fields": [{ "tag": "field",
           "name": "DRM_ADD_COMMAND",
           "value": 0 },
         { "tag": "field",
           "name": "DRM_RM_COMMAND",
           "value": 1 },
         { "tag": "field",
           "name": "DRM_INST_HANDLER",
           "value": 2 },
         { "tag": "field",
           "name": "DRM_UNINST_HANDLER",
           "value": 3 }] },

:ANON-TYPE-42 is not the name of a class

Ok. After many blind alley and discovering the proper clang incantations and loading the library I finally got the test (LWM)

Here is the transcript.

CL-USER 5 > (in-package "CL-AUTOWRAP-TESTS")
#<The CL-AUTOWRAP-TESTS package, 24/64 internal, 38/64 external>

CL-AUTOWRAP-TESTS 6 > (cffi:use-foreign-library cl-autowrap-tests-lib)
#<CFFI:FOREIGN-LIBRARY CL-AUTOWRAP-TESTS-LIB "libcl-autowrap-test.dylib">

CL-AUTOWRAP-TESTS 7 > (get-foo)
#<FOO-T {#X00150F40}>

CL-AUTOWRAP-TESTS 8 > (free-foo *)

CL-AUTOWRAP-TESTS 9 > (let ((foo (get-foo)))
  (setf (foo-t.a foo) 5)             ;; foo.a = 5;
  (setf (foo-t.x[].b0 foo 0) #b10)   ;; foo.x[0].b0 = 2;
  (print (foo-t.x[].s.x foo 1))      ;; anonymous struct
  (foo-t.x[].s foo 0)                ;; => child wrapper
  (foo-t.x[].s& foo 0)               ;; &(foo.x[0].s) => pointer
  (free-foo foo))

0 
Error: #:ANON-TYPE-2247244 is not the name of a class
  1 (continue) Try finding the class #:ANON-TYPE-2247244 again
  2 (abort) Return to level 0.
  3 Return to top loop level 0.

Type :b for backtrace or :c <option number> to proceed.
Type :bug-form "<subject>" for a bug report template or :? for other options.

CL-AUTOWRAP-TESTS 11 : 1 > :b
Call to ERROR
Call to CERROR
Call to FIND-CLASS
Call to CLOS::MAKE-INSTANCE-SYMBOL
Interpreted call to FOO-T.X[].S
Call to LET
Call to EVAL
Call to CAPI::CAPI-TOP-LEVEL-FUNCTION
Call to CAPI::INTERACTIVE-PANE-TOP-LOOP
Call to MP::PROCESS-SG-FUNCTION

Looks like uninterned symbol are not quite palatable as class names in LWM.

Cheers
MA

Wrapping gdk-pixbuf..

Good day Ryan!

I'm trying to wrap more stuff, this time GDK-Pixbuf + Clutter.
The bigger problem, is that without certain ignores, GDK-Pixbuf does not
get wrapped at all - none of its symbols appear in the spec files.
The smaller problem is that some glib types do not get wrapped no matter what ignores,
for instance the g* types -- gboolean, guint* etc.

The overall problem is that I feel ill-equipped for deduction of what goes
wrong, and why c2ffi chooses to omit something from its spec files..

Here is what I have (alternatively, check out [email protected]:deepfire/cffi-clutter.git):

clutter-raw.h

#include <glib.h>
#include <gdk-pixbuf/gdk-pixbuf.h>
#include <clutter/clutter.h>
#include <cogl/cogl.h>
#include <GL/gl.h>

clutter-raw.lisp

(cl:eval-when (:compile-toplevel :load-toplevel :execute)
  (cffi-sys:%load-foreign-library
   :libclutter "/usr/lib64/libgdk_pixbuf-2.0.so")
  (cffi-sys:%load-foreign-library
   :libclutter "/usr/lib64/libglib-2.0.so")
  (cffi-sys:%load-foreign-library
   :libclutter "/usr/lib64/libclutter-1.0.so.0"))

(autowrap:c-include "clutter-raw.h"
                    :exclude-arch ("i686-pc-linux-gnu"
                                   "i686-pc-win32"
                                   "x86_64-pc-win64"
                                   "i686-apple-darwin9"
                                   "x86_64-apple-darwin9")
                    :exclude-sources ("/usr/local/lib/clang/3.3/include/(?!stddef.h).*"
                                      ;; this (your) ignore weirdly enables GDK-Pixbuf wrapping
                                      ;; -- take it away, and none of its functions will appear in spec files.
                                      "/usr/include/(?!stdint.h|bits/types.h|sys/types.h|glib-2.0|clutter-1.0|cogl).*"
                                      "/usr/include/glib-2.0/glib/deprecated")
                    :sysincludes '("/usr/include/"
                                   "/usr/include/glib-2.0"
                                   "/usr/lib64/glib-2.0/include"
                                   "/usr/include/cairo"
                                   "/usr/include/cogl"
                                   "/usr/include/clutter-1.0"
                                   "/usr/include/pango-1.0"
                                   "/usr/include/atk-1.0"
                                   "/usr/include/gdk-pixbuf-2.0"
                                   "/usr/include/json-glib-1.0")
                    :exclude-constants (".*")
                    :constant-accessor const)

Nicer string literal passing needed

Whenever I have to pass a string literal to a function, I have to do things like this:

(with-foreign-strings ((text "Press <space> to reset the image position."))
  (clutter-text-new-with-text (make-pointer 0) text))

It'd be nicer if the call would detect them strings, and produce the W-F-S automatically..

Parameters named t do not work

Consider /tmp/foo.h containing:

int x(void *t);

CL-USER> (autowrap:c-include "/tmp/foo.h")
; in: AUTOWRAP:C-INCLUDE "/tmp/foo.h"
;     (SB-INT:NAMED-DS-BIND (:MACRO X . DEFMACRO)
;         (T)
;         (CDR #:EXPR)
;       (BLOCK X
;         (LET ((#:!FUN1850 #))
;           (WITH-SLOTS (#)
;               #:!FUN1850
;             (AUTOWRAP::FOREIGN-TO-FFI # '# # #:!FIELDS1851 #)))))
; --> SB-INT:BINDING*
; ==>
;   (LET* ((#:G0 (SB-C::CHECK-DS-LIST (CDR #:EXPR) 1 1 '(# T))) (T (POP #:G0)))
;     (BLOCK X
;       (LET ((#:!FUN1850 #))
;         (WITH-SLOTS (#)
;             #:!FUN1850
;           (AUTOWRAP::FOREIGN-TO-FFI # '# # #:!FIELDS1851 #)))))
;
; caught ERROR:
;   T names a defined constant, and cannot be used as a local variable.

;     (SB-INT:NAMED-LAMBDA (MACRO-FUNCTION X)
;         (#:EXPR #:ENV)
;       (DECLARE (SB-C::LAMBDA-LIST (T)))
;       (DECLARE (IGNORE #:ENV))
;       (SB-INT:NAMED-DS-BIND (:MACRO X . DEFMACRO)
;           (T)
;           (CDR #:EXPR)
;         (BLOCK X
;           (LET (#)
;             (WITH-SLOTS #
;                 #:!FUN1850
;               #)))))
;
; caught STYLE-WARNING:
;   The variable #:EXPR is defined but never used.
;
; compilation unit finished
;   caught 1 ERROR condition
;   caught 1 STYLE-WARNING condition
; Total of 0 compile-time skipped definitions
; Total of 0 compile-time missing entities
; Total of 0 load-time skipped definitions
; Total of 0 load-time missing entities

Error when using c function that return type bool

C function prototype

bool wlc_init(const struct wlc_interface *interface, int argc, char *argv[]);

My lisp code

(defun init (i argc argv)
  (wlc-ffi.functions:wlc-init i argc argv))

The error

; caught ERROR:
;   during macroexpansion of
;   (CFFI-SYS:%FOREIGN-FUNCALL "wlc_init" (:POINTER WLC-FFI::INTERFACE :INT ...)
;                              ...).
;   Use *BREAK-ON-SIGNALS* to intercept.
;   
;    :-BOOL fell through ECASE expression.
;    Wanted one of (:CHAR :UNSIGNED-CHAR :SHORT :UNSIGNED-SHORT :INT :UNSIGNED-INT
;                   :LONG :UNSIGNED-LONG :LONG-LONG :UNSIGNED-LONG-LONG :FLOAT
;                   :DOUBLE :POINTER :VOID).

Not sure if I'm doing something just wrong here but I have no idea how to add the type _Bool into cffi.
Here's the c-include if it matters, I also tried with with including "stdbool.h" but that had no effect on the error.

(autowrap:c-include "wlc/wlc.h"
 :accessor-package :wlc-ffi.accessors
 :function-package :wlc-ffi.functions
 :spec-path "/ss/prog/lisp/aw-wlc/"
 :exclude-sources ("/usr/local/lib/clang/([^/]*)/include/(?!stddef.h)"
           "/usr/include/")
 :include-sources ("stdint.h"
           "sys/types.h"
           "/usr/local/include/"))

Errors while autowrapping Chipmunk2D

I'm trying to autowrap Chipmunk2D (https://github.com/slembcke/Chipmunk2D). I'm trying to build it on OS X.

I had an error where <float.h> couldn't be found. I searched the path where the Chipmunk2D XCode project finds float.h (/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/lib/clang/6.1.0/include/) and added it to :sysincludes.

After solving that there are still many errors, but they all repeat. Basically, I have two types of errors:

CL-USER> (ql:quickload :cl-chipmunk)
To load "cl-chipmunk":
  Load 1 ASDF system:
    cl-chipmunk
; Loading "cl-chipmunk"
In file included from /var/folders/dv/8hnn7n357pq1g40tqsd9pk940000gn/T/TEMP-N8LQ13WQ:1:
In file included from /Users/simon/projects/cl-chipmunk/src/spec/chipmunk.h:1:
In file included from /usr/local/include/chipmunk/chipmunk.h:53:
In file included from /usr/local/include/chipmunk/chipmunk_types.h:45:
In file included from /System/Library/Frameworks/ApplicationServices.framework/Headers/ApplicationServices.h:23:
In file included from /System/Library/Frameworks/CoreServices.framework/Headers/CoreServices.h:19:
In file included from /System/Library/Frameworks/CoreFoundation.framework/Headers/CoreFoundation.h:55:
In file included from /System/Library/Frameworks/CoreFoundation.framework/Headers/CFPropertyList.h:13:
In file included from /System/Library/Frameworks/CoreFoundation.framework/Headers/CFStream.h:15:
In file included from /usr/include/dispatch/dispatch.h:51:
/usr/include/dispatch/object.h:143:15: error: blocks support disabled - compile with -fblocks or pick a deployment target that supports them
typedef void (^dispatch_block_t)(void);
              ^
In file included from /var/folders/dv/8hnn7n357pq1g40tqsd9pk940000gn/T/TEMP-N8LQ13WQ:1:
In file included from /Users/simon/projects/cl-chipmunk/src/spec/chipmunk.h:1:
In file included from /usr/local/include/chipmunk/chipmunk.h:53:
In file included from /usr/local/include/chipmunk/chipmunk_types.h:45:
In file included from /System/Library/Frameworks/ApplicationServices.framework/Headers/ApplicationServices.h:23:
In file included from /System/Library/Frameworks/CoreServices.framework/Headers/CoreServices.h:23:
In file included from /System/Library/Frameworks/CoreServices.framework/Frameworks/AE.framework/Headers/AE.h:20:
In file included from /System/Library/Frameworks/CoreServices.framework/Frameworks/CarbonCore.framework/Headers/CarbonCore.h:127:
In file included from /System/Library/Frameworks/CoreServices.framework/Frameworks/CarbonCore.framework/Headers/DriverServices.h:26:
In file included from /System/Library/Frameworks/CoreServices.framework/Frameworks/CarbonCore.framework/Headers/MachineExceptions.h:22:
In file included from /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/lib/clang/6.1.0/include/emmintrin.h:31:
In file included from /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/lib/clang/6.1.0/include/xmmintrin.h:31:
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/lib/clang/6.1.0/include/mmintrin.h:46:12: error: invalid conversion between vector type '__m64' (vector of 1 'long long' value) and integer type 'int' of different size
    return (__m64)__builtin_ia32_vec_init_v2si(__i, 0);
           ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The invalid conversion error occurs a lot for different types, but they all seem to come from emmintrin.h, xmmintrin.h and mmintrin.h. Same for the blocks support disabled error.

I would like to follow the instructions of the compiler for the blocks error, but I don't know how to pass those compiler options to cl-autowrap.

I would really appreciate it if you could give me some guidance ;)

Thank you!

c-ref to array slot returns first element, not wrapped array

Run the following in order from the repl:

(autowrap:define-foreign-record 'example :struct 128 8
                                '((:a (:array :int 4) :bit-size 128
                                   :bit-offset 0 :bit-alignment 8)))

(autowrap:define-foreign-alias 'example '(:struct (example)))

(autowrap:define-wrapper (:struct (example)))

(defvar *test* (autowrap:alloc 'example))

(plus-c:c-ref *test* example :a)

The result from c-ref is 0 which may well be the first element of the array, but
the expected result is a wrapped array.

c-ref passes build-ref the first field which is :a

:a is a record-field but is not a frf-bitfield so it then calls

(build-ref (car rest) (foreign-type field)
           (autowrap::make-field-ref field current-ref) (cdr rest))

but as (car rest) is nil it calls 'build-ref (null foreign-array)' which doesn't
wrap the result.

Is this expected behavior?

CCL: Importing AUTOWRAP:FREE to #<Package "AUTOWRAP.MINIMAL"> would conflict with inherited symbol FREE

On quicklisp 2013-10-03 cl-autowrap has a regression on CCL:

Importing AUTOWRAP:FREE to #<Package "AUTOWRAP.MINIMAL"> would conflict with inherited symbol FREE

See also cl-test-grid page for cl-autowrap:
http://common-lisp.net/project/cl-test-grid/library/cl-autowrap.html

rpav / cl-autowrap Goto Github PK

cl-autowrap's People

Contributors

Stargazers

Watchers

Forkers

cl-autowrap's Issues

Recommend Projects

Recommend Topics

Recommend Org