cffi-devel

cffi-devel@common-lisp.net

1113 discussions

[cffi-devel] New patches: 6-Sep-2006
by loliveira＠common-lisp.net 07 Sep '06

07 Sep '06

Wed Sep 6 07:40:27 EDT 2006 Luis Oliveira <loliveira(a)common-lisp.net> * Use the RTEST package nickname instead of RT because of Lispworks. M ./tests/bindings.lisp -1 +1 Wed Sep 6 06:13:36 EDT 2006 Luis Oliveira <loliveira(a)common-lisp.net> * Backwards compatibility for cffi-clisp's foreign-funcall M ./src/cffi-clisp.lisp -1 +9 An updated tarball of CFFI's source can be downloaded here: http://common-lisp.net/project/cffi/tarballs/cffi-060906.tar.gz Darcsweb URL: http://common-lisp.net/cgi-bin/darcsweb/darcsweb.cgi?r=cffi;a=summary

1 0

[cffi-devel] New patches: 5-Sep-2006
by loliveira＠common-lisp.net 06 Sep '06

06 Sep '06

Tue Sep 5 06:34:18 EDT 2006 Luis Oliveira <loliveira(a)common-lisp.net> * Support for recent CVS versions of CLISP - As of 2006-09-03, CLISP's FFI::FOREIGN-LIBRARY-FUNCTION takes an additional argument. Accomodated this change. Patch courtesy of Kevin Rosenberg. M ./src/cffi-clisp.lisp -1 +1 Tue Sep 5 06:11:50 EDT 2006 Luis Oliveira <loliveira(a)common-lisp.net> * [u]llong alignment issues on darwin/ppc - CLISP: special case for :[unsigned-]long-long on darwin/ppc. - SBCL: missing special case for :unsigned-long-long. - New regression test: STRUCT.ALIGNMENT.8. M ./src/cffi-clisp.lisp -2 +3 M ./src/cffi-sbcl.lisp -2 +3 M ./tests/libtest.c +24 M ./tests/struct.lisp +29 An updated tarball of CFFI's source can be downloaded here: http://common-lisp.net/project/cffi/tarballs/cffi-060905.tar.gz Darcsweb URL: http://common-lisp.net/cgi-bin/darcsweb/darcsweb.cgi?r=cffi;a=summary

1 0

[cffi-devel] ecl patch for callback support
by Brian Spilsbury 05 Sep '06

05 Sep '06

Hello, a cursory glance at the archives didn't show a patch for ecl callbacks, so perhaps the one included below will be useful. This seems to make cl-cairo happier, anyhow. :) Regards, Brian. #### --- cffi-ecl.lisp 2006-04-12 05:14:18.000000000 +0900 +++ cffi-ecl-fix.lisp 2006-09-05 20:09:17.000000000 +0900 @@ -228,6 +228,8 @@ ;;;# Callbacks +(defvar *callbacks* (make-hash-table)) + ;;; Create a package to contain the symbols for callback functions. ;;; We want to redefine callbacks with the same symbol so the internal ;;; data structures are reused. @@ -257,7 +259,7 @@ (gethash name *callbacks*) (unless winp (error "Undefined callback: ~S" name)) - (ffi:callback name))) + (ffi:callback symbol))) ;;;# Foreign Globals

2 1

[cffi-devel] New patches: 31-Aug-2006
by loliveira＠common-lisp.net 01 Sep '06

01 Sep '06

Thu Aug 31 11:31:30 EDT 2006 Luis Oliveira <loliveira(a)common-lisp.net> * New TODO items about LOAD-FOREIGN-LIBRARY. M ./TODO +3 Thu Aug 31 11:30:39 EDT 2006 Luis Oliveira <loliveira(a)common-lisp.net> * New macro: INCF-POINTER (with documentation) M ./doc/cffi-manual.texinfo +52 M ./src/package.lisp +1 M ./src/types.lisp -1 +3 M ./src/utils.lisp -13 +13 M ./tests/libtest.c -1 +1 M ./tests/memory.lisp +11 Thu Aug 31 11:28:37 EDT 2006 Luis Oliveira <loliveira(a)common-lisp.net> * Misc manual cleanups like s/=>/@result{}/, etc... M ./doc/cffi-manual.texinfo -44 +50 M ./doc/cffi-sys-spec.texinfo -5 +3 Thu Aug 31 11:24:23 EDT 2006 Luis Oliveira <loliveira(a)common-lisp.net> * Update cffi-uffi-compat to reflect recent UFFI changes. - Better handling of shared library extensions on Windows. Patch courtesy of Kevin Rosenberg. M ./uffi-compat/uffi-compat.lisp -4 +4 An updated tarball of CFFI's source can be downloaded here: http://common-lisp.net/project/cffi/tarballs/cffi-060831.tar.gz Darcsweb URL: http://common-lisp.net/cgi-bin/darcsweb/darcsweb.cgi?r=cffi;a=summary

1 0

[cffi-devel] result reporting for random numeric tests in tests/random-tester.lisp
by Stephen Compall 31 Aug '06

31 Aug '06

I'm almost finished with a random test generator in tests/random-tester.lisp. However, I'd like to know how you would like results to be reported by the main driver function, `run-random-numeric-tests'. Shall it print RT test names and expect you to analyze the temporary C/Lisp files? Shall it print type signatures as well? Shall it try to find what types the failed tests have in common? I'd like to provide all convenient information without overengineering any further than I already may have. There is also the possibility that I have misinterpreted how "random" should be parameterized with regard to tests. Everything I have so far: http://scompall.nocandysw.com/cffi/randtest-0.darcs.patch is a darcs patch http://scompall.nocandysw.com/cffi/randtest-0.diff is `darcs whatsnew' output on the file log: A random numeric test generator (except outermost driver code) * Everything on pages "Temporary file generation" and "Test driver and result handling", and most of "Random test selection" is new. * Add a "stream" argument to C-FUNCTION and C-CALLBACK. * `output-c-code' is broken, and I don't know whether it's useful, so snap it in half more permanently. -- Stephen Compall http://scompall.nocandysw.com/blog

2 2

[cffi-devel] callbacks calling translate-to-foreign
by Jan Rychter 14 Aug '06

14 Aug '06

I have a problem defining callbacks that use user-defined types. After quite a bit of debugging I traced it down to this: (defcallback %callback-resize :void ((ee ecore-evas)) ...) ecore-evas is defined using defctype, I also have translate-from-foreign and translate-to-foreign defined for it. Now, everything works fine for normal function calls, types get translated back and forth, but when the callback gets called, CFFI calls translate-to-foreign on ee instead of the seemingly correct translate-from-foreign. This obviously fails, as it's trying to translate-to-foreign a foreign pointer. I can't understand the defcallback macrology, so I can't trace this any further -- am I doing something wrong, or is there a bug in CFFI? --J.

1 1

[cffi-devel] testing equality on system objects
by Sean Champ 14 Aug '06

14 Aug '06

Hello, I was migrating some code from one of my own codebases into an archive for the Tioga project, when I noticed the following function. It's used in facilitating a function called EQLP, in comparison onto ALIEN objects. I would like to submit this function, now, in case it would be of use in CFFI; if it would be, I could be glad to remove the original form of it from my own code. This is an equality-predicate function, operating in comparison onto system objects, viz ALIEN-VALUE objects. #+(or CMU SBCL) (shadowing-import #+CMU '(ext:fixnump ext:bignump) #+SBCL '(sb-int:fixnump sb-int:bignump)) #+CMU (use-package '#:SB-ALIEN) #+SBCL (use-package '#:SB-ALIEN) ;; (hopefully that will cover all of the symbol-availability.) ;; (I have excerpted this from another package.) (defun system-objects-eqlp (obj1 obj2 external-type) (declare (type (or alien-value float integer) obj1 obj2) (type alien-type external-type)) (typecase external-type ;; handle, first, those types for which OBJ1 and OBJ2 both must ;; have regular CL representations (alien-integer-type ;; NB: ALIEN-ENUM-TYPE is handled here (when (and (integerp obj1) (integerp obj2)) (cond ((and (fixnump obj1) (fixnump obj2)) (eql (the fixnum obj1) (the fixnum obj2))) ((and (bignump obj1) (bignump obj2)) (eql (the bignum obj1) (the bignum obj2)))))) (alien-float-type (when (and (floatp obj1) (floatp obj2) (eql (the float obj1) (the float obj2))) (values t))) (t (let ((o1-foreign-type (alien-value-type obj1)) (o2-foreign-type (alien-value-type obj2))) (when (and (alien-type-= o1-foreign-type external-type) (alien-type-= o2-foreign-type external-type)) (typecase external-type (alien-fun-type ;; OBJ1 and OBJ2 are already determined as having ;; ALIEN-VALUE-TYPEs that are ALIEN-TYPE-= . ;; ;; Presumably, this is the only check remaining, then (sap= (alien-sap obj1) (alien-sap obj2))) (alien-record-type (dolist (type-field (alien-record-type-fields external-type) (values t)) ;; referenced on #'SB-ALIEN:SLOT (let ((field-offset (alien-record-field-offset type-field)) (field-type (alien-record-field-type type-field))) (unless (system-objects-eqlp ;; referenced on DEREF and DEREF-GUTS (extract-alien-value (alien-value-sap obj1) field-offset field-type) (extract-alien-value (alien-value-sap obj2) field-offset field-type) field-type) (return (values nil)))))) (alien-pointer-type (let ((dest-type (alien-pointer-type-to external-type))) ;; NB: This is made with the assumption that the offset 0 will be ;; applicable in all times in which this method will be called ;; -- namely, for all 'alien pointers' not representing 'alien ;; arrays' ;; (cond (dest-type (system-objects-eqlp ;; referenced on #'DEREF & #'DEREF-GUTS ;; ;; Why this is enough, in lieu of DEREF : ;; 1) the EXTERNAL-TYPE is certainly an ALIEN-POINTER-TYPE ;; 2) there are no INDICES that must be provided, as if for ;; an equivalent DEREF call ;; (extract-alien-value (alien-value-sap obj1) 0 dest-type) (extract-alien-value (alien-value-sap obj2) 0 dest-type) dest-type)) ;; Conslusion: OBJ1 and OBJ2 are both 'null pointers' (t t)))) (alien-array-type (let* ((dimensions (copy-list (alien-array-type-dimensions external-type))) (etype (alien-array-type-element-type external-type)) (dim-ptr (1- (length dimensions)))) (loop (when (zerop dim-ptr) (return t)) (let ((obj1-v (apply #'deref obj1 dimensions)) (obj2-v (apply #'deref obj2 dimensions))) (unless (or (and (null obj1-v) (null obj2-v)) (system-objects-eqlp obj1-v obj2-v etype)) (return nil))) (when (zerop (nth dim-ptr dimensions)) (decf dim-ptr)) (decf (nth dim-ptr dimensions))))) (t ;; NB: this would be a mater of a program deficiency, not of any ;; innate quality of the underlying system (error "System is not prepared to determine equivalence ~ for objects of type ~S" external-type)) )))))) If there would be criticism about how that function operates, I could be glad to know, so as to resolve it. Looking at it, now, I consider that the EXTERNAL-TYPE argument was required for kluding onto the situation of an ALIEN numeric type -- alien type representation, onto a native value. In no other case does it appear to be used. I am not sure, immediateley, of how system-objects-eqlp would need to be revised, if to be applicable in lisp hosts being neither SBCL nor CMUCL. Now that I have become aware of CFFI, I had thought that that function might bear mentioning. I consider that CFFI would be an appropriate location for that item of code, there. As it was, I had implemented that function, to have it take care of what I had wanted an EQLP function to do, on system objects. Considering that EQLP implements a method using system-objects-eqlp, , I might also request that the following would be added, with it, if that item would be added to the CFFI sources: #+TAL (defmethod tal:eqlp ((obj1 alien-value) (obj2 alien-value)) (system-objects-eqlp obj1 obj2 (alien-value-type obj1))) For what it's worth, EQLP uses an extension onto the MOP methods protocol, so as to try and save some cycles when called on objects not of the same type; that behavior might resemble something after some compiler optimizations in CMUCL and SBCL; it would be operable in a cross-MOP-enabled-platform manner. For what it's worth, I intend to publish the source for EQLP and the documentation for it, as part to a 'tal-base' system in the the Tioga Auxiliary Library (TAL). TAL will be provided in a source-tree to the Tioga project http://www.common-lisp.net/project/tioga/ On release of it, notification should be made by way of the Tioga-announce mailing list http://common-lisp.net/mailman/listinfo/tioga-announce I mention the above, upon my wanting to explain what the phantom EQLP function is. I consider that the behaviors of EQLP may be better explained of the documentation for it, however. (The source-code and the documentation for it are scheduled for release, but they are not yet available, at this hour. ) In the top of it, EQLP is a particular equality-predicate function. If that system-objects-eqlp function would appear to be applicable to the CFFI project, I would be glad if that it could be useful. I would remove it, then, removing it from the TAL codebase from which it was drawn. I consider that it would be more suitably located, if somewhere onto CFFI, more than if it was into a component to TAL. -- Sean Some code I had used as to double-check if systems-objects-eqlp was working on stat objects. The following was defined onto SBCL. (defun lstat-without-clos-intermediary (designator) ;; patched onto sb-posix:lstat (sb-posix::with-alien-stat ext-stat-obj () (let ((r (sb-alien::alien-funcall (sb-alien::extern-alien "lstat" (function sb-alien::int sb-alien::c-string (* sb-posix::alien-stat))) (namestring designator) ext-stat-obj))) (if (minusp r) (simple-err 'sb-posix:syscall-error "lstat on file ~S" designator) (values ext-stat-obj))))) (defparameter *stat-1* (lstat-without-clos-intermediary "/etc/passwd")) (defparameter *stat-2* (lstat-without-clos-intermediary "/etc/passwd")) (system-objects-eqlp *stat-1* *stat-2* (alien-value-type *stat-1*))

1 0

[cffi-devel] small doc patch
by Jack Unrue 04 Aug '06

04 Aug '06

The attached patch is a small doc fix for the manual in which I made the name of the first parameter for free-translated-object consistent with how free-converted-object and translate-to-foreign are documented. -- Jack Unrue

1 0

[cffi-devel] Test FOREIGN-SYMBOL-POINTER.1 failed
by Yaroslav Kavenchuk 01 Aug '06

01 Aug '06

cffi-060626.tar.gz SBCL 0.9.15.8 mingw, win2k Test FOREIGN-SYMBOL-POINTER.1 failed Form: (COMPARE-AGAINST-ABS (FOREIGN-SYMBOL-POINTER "abs")) Expected value: T Actual value: NIL. Thanks! -- WBR, Yaroslav Kavenchuk.

1 0

[cffi-devel] Enums, structs, and field sizes
by Dan Knapp 15 Jul '06

15 Jul '06

I just uncovered this issue with cffi-net. This is more of a design question than a bug; I know at least one way that I could work around it, but it's distressingly inelegant. To get the cffi-net low-level primitives up and running, we feed cffi-grovel the names of some preprocessor constants which it turns into an enum called address-family, with values such as :af-inet -> 2. This enum is used as a parameter to a bunch of different function calls, and it's also a member of the struct sockaddr-in. We tell the groveller that the "family" slot should be of the type address-family, so instead of figuring out the size of it and putting in :uint8, :uint16, or whatever, it declares the slot to be of type address-family. The problem is that the slot of course has a correct size. In this case it happens to be :uint8. In the absence of other guidance, cffi assumes that an enum type should be a regular integer, which seems to wind up as :sint32. So when it comes time to access the slot, too much memory is used and it overwrites adjacent slots. Since cffi does provide the option to give a base type for the enum, we could specify that type. If we do that by hand, though, it defeats the purpose of having cffi-grovel; it would introduce a portability issue and would have to be manually maintained for new architectures with surprising new slot sizes. We could make cffi-grovel smart enough to notice that the enum is used for a slot, and take some appropriate action. It could set the base type of the enum, but this would cause trouble with the functions that use it as a parameter type and expect it to be int-sized. It could create a new enum, address-family-uint8, that duplicates everything in address-family but is a different size. This has the benefit of working, so if we can't come up with anything better it's probably what we'll do. We also can't reasonably hook in any special-case code for when this field is accessed, because cffi only dispatches on type when it's deciding how to access a field, and if we create a new type we're back to one of the above scenarios. It occured to me that the reason this is a problem at all is that enums are considered to be types. Perhaps they really need to be something outside the type system, that can be overlayed on a type as needed? Anyway, since there's no particular hurry, we thought we'd post to the list and ask for suggestions. It would be comforting to find a clean way to do this. -- Dan Knapp

1 0

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

cffi-devel