Re: [Chicken-hackers] [PATCH] fix: eval.scm: use soname only with when USES_SONAME present
On Sat, Jan 25, 2014 at 10:36:43PM +0100, Peter Bex wrote: I believe this patch does basically the correct thing, so here's a signed-off copy. However, it breaks the Cygwin and MingW builds because they have their own Makefile fragments for generating chicken-defaults.h and set CUSTOM_CHICKEN_DEFAULTS to override the generation of that file from defaults.make. Here's another patch, updated for today's master. The small patch which adds C_INSTALL_POSTINSTALL_PROGRAM somewhere in the chicken-defaults.h rule (which my patch modified completely) caused my old patch to fail to apply. I hope I didn't break anything by my mucking about by hand! Cheers, Peter -- http://www.more-magic.net From 72ba40794f6a229a84a64404bff378402f602617 Mon Sep 17 00:00:00 2001 From: Kristian Lein-Mathisen kristianl...@gmail.com Date: Fri, 3 Jan 2014 15:16:51 +0100 Subject: [PATCH 1/3] fix: eval.scm: use soname only with when USES_SONAME present in Makefile Android, who'se software-version is 'linux, does not use sonames and crashes with (use unit) because the binary-version is incorrectly appended to the shared library filename. This patch explicitly propagates the Makefiles' USES_SONAME property through to eval for this to work properly. Signed-off-by: Peter Bex peter@xs4all.nl --- defaults.make |7 +++ eval.scm |4 ++-- 2 files changed, 9 insertions(+), 2 deletions(-) diff --git a/defaults.make b/defaults.make index f3c600b..a32dfbd 100644 --- a/defaults.make +++ b/defaults.make @@ -485,6 +485,13 @@ endif echo #ifndef C_BINARY_VERSION $@ echo # define C_BINARY_VERSION $(BINARYVERSION) $@ echo #endif $@ + echo #ifndef C_USES_SONAME $@ +ifdef USES_SONAME + echo # define C_USES_SONAME 1 $@ +else + echo # define C_USES_SONAME 0 $@ +endif + echo #endif $@ echo /* END OF FILE */ $@ endif diff --git a/eval.scm b/eval.scm index a835394..ed1b92d 100644 --- a/eval.scm +++ b/eval.scm @@ -57,6 +57,7 @@ (define-foreign-variable install-egg-home c-string C_INSTALL_EGG_HOME) (define-foreign-variable installation-home c-string C_INSTALL_SHARE_HOME) (define-foreign-variable binary-version int C_BINARY_VERSION) +(define-foreign-variable uses-soname? bool C_USES_SONAME) (define-foreign-variable install-lib-name c-string C_INSTALL_LIB_NAME) (define ##sys#core-library-modules @@ -1066,8 +1067,7 @@ (define dynamic-load-libraries (let ((ext -(if (and (memq (software-version) '(linux netbsd openbsd freebsd haiku hurd)) - (not (zero? binary-version))) ; allow configless build +(if uses-soname? (string-append ##sys#load-library-extension . -- 1.7.10.4 From 0890a33ae04d2ada3d4a81b8570e81c9c715fa9d Mon Sep 17 00:00:00 2001 From: Peter Bex peter@xs4all.nl Date: Fri, 24 Jan 2014 23:24:29 +0100 Subject: [PATCH 2/3] Fix SONAME handling on Windows This removes unnecessarily duplicated rules for chicken-defaults.h from Cygwin Makefile and fixes MingW makefile to set the new USES_SONAME definition. The only actual difference between Makefile.cygwin and defaults.make was that it set TARGET_RUN_LIB_HOME to TARGET_PREFIX/bin - now make this dependent on DLLSINPATH, which also controls where the library is copied during install. --- Makefile.cygwin | 138 --- Makefile.mingw | 12 +++-- csc.scm | 11 ++--- defaults.make |8 4 files changed, 20 insertions(+), 149 deletions(-) diff --git a/Makefile.cygwin b/Makefile.cygwin index 6989ab8..abe7fa0 100644 --- a/Makefile.cygwin +++ b/Makefile.cygwin @@ -80,7 +80,6 @@ APPLY_HACK_OBJECT = apply-hack.$(ARCH)$(O) # select default and internal settings -CUSTOM_CHICKEN_DEFAULTS=1 include $(SRCDIR)/defaults.make LIBCHICKEN_SO_LIBRARIES = $(LIBRARIES) @@ -127,141 +126,4 @@ ifdef HACKED_APPLY endif cat chicken-defaults.h $@ -chicken-defaults.h: - echo /* generated */ $@ - echo #define C_CHICKEN_PROGRAM \$(CHICKEN_PROGRAM)$(EXE)\ $@ - echo #ifndef C_INSTALL_CC $@ - echo # define C_INSTALL_CC \$(C_COMPILER)\ $@ - echo #endif $@ - echo #ifndef C_INSTALL_CXX $@ - echo # define C_INSTALL_CXX \$(CXX_COMPILER)\ $@ - echo #endif $@ - echo #ifndef C_INSTALL_POSTINSTALL_PROGRAM $@ - echo # define C_INSTALL_POSTINSTALL_PROGRAM \$(POSTINSTALL_PROGRAM)\ $@ - echo #endif $@ - echo #ifndef C_INSTALL_CFLAGS $@ - echo # define C_INSTALL_CFLAGS \$(C_COMPILER_OPTIONS) $(C_COMPILER_OPTIMIZATION_OPTIONS)\ $@ - echo #endif $@ - echo #ifndef C_INSTALL_LDFLAGS $@ - echo # define C_INSTALL_LDFLAGS \$(LINKER_OPTIONS) $(LINKER_OPTIMIZATION_OPTIONS)\ $@ - echo #endif $@ - echo #ifndef C_INSTALL_PREFIX $@ - echo # define C_INSTALL_PREFIX \$(PREFIX)\ $@ - echo #endif $@ - echo #ifndef C_INSTALL_SHARE_HOME $@ - echo # define C_INSTALL_SHARE_HOME
[Chicken-hackers] [PATCH] Remove list argument check from C_u_i_memq
This was erroneously added in 0a52536b7cb6b3d5a35ecc8f4c11131041ae873a From 32bc17d3c3a313a4f869079f3777ed7dd2b8b9f4 Mon Sep 17 00:00:00 2001 From: Moritz Heidkamp mor...@twoticketsplease.de Date: Wed, 5 Feb 2014 23:26:39 +0100 Subject: [PATCH] Remove list argument check from C_u_i_memq This was erroneously added in 0a52536b7cb6b3d5a35ecc8f4c11131041ae873a --- runtime.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/runtime.c b/runtime.c index 9575d80..f184d71 100644 --- a/runtime.c +++ b/runtime.c @@ -5600,9 +5600,6 @@ C_regparm C_word C_fcall C_u_i_memq(C_word x, C_word lst) else lst = C_u_i_cdr(lst); } - if(lst!=C_SCHEME_END_OF_LIST) -barf(C_BAD_ARGUMENT_TYPE_ERROR, memv, lst); - return C_SCHEME_FALSE; } -- 1.8.5.2 ___ Chicken-hackers mailing list Chicken-hackers@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-hackers
Re: [Chicken-hackers] [PATCH] Replace unsafe string functions with their safer counterparts
Hi, On Thu, 23 Jan 2014 12:28:01 +0100 Christian Kellermann ck...@pestilenz.org wrote: I propose the following patch. It has been tested on OpenBSD and Linux. Should work without troubles on all other OSes as well. Please give it a try. Sorry for taking so long to give it a try on Windows. Unfortunately, it seems to break things on that platform. The environment variable tests in posix-tests.scm caught the issue. Here's an example: #;1 (use posix) ; loading c:/local/chicken-master-mingw/lib/chicken/7/posix.import.so ; ... ; loading library posix ... #;2 (setenv FOO bar) #;3 (get-environment-variable FOO) x\x01=b\x03 Reverting 8efbbc50d050d28c34ea0c5a58d0de92ec9f918a makes tests pass again. Tested on cygwin, mingw (Windows XP 32bit). Best wishes. Mario -- http://parenteses.org/mario ___ Chicken-hackers mailing list Chicken-hackers@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-hackers
Re: [Chicken-hackers] [PATCH] fix: eval.scm: use soname only with when USES_SONAME present
On Wed, 5 Feb 2014 20:11:16 +0100 Peter Bex peter@xs4all.nl wrote: On Sat, Jan 25, 2014 at 10:36:43PM +0100, Peter Bex wrote: I believe this patch does basically the correct thing, so here's a signed-off copy. However, it breaks the Cygwin and MingW builds because they have their own Makefile fragments for generating chicken-defaults.h and set CUSTOM_CHICKEN_DEFAULTS to override the generation of that file from defaults.make. Here's another patch, updated for today's master. The small patch which adds C_INSTALL_POSTINSTALL_PROGRAM somewhere in the chicken-defaults.h rule (which my patch modified completely) caused my old patch to fail to apply. I hope I didn't break anything by my mucking about by hand! Thanks Peter and Kristian. I tested and applied your patches. They seem to work fine on Windows XP 32bit (mingw, mingw-msys and cygwin). I also tested them on Linux x86-64 (gcc and clang), just in case. Best wishes. Mario -- http://parenteses.org/mario ___ Chicken-hackers mailing list Chicken-hackers@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-hackers
Re: [Chicken-hackers] [PATCH] Replace unsafe string functions with their safer counterparts
On Wed, 05 Feb 2014 23:56:11 + Mario Domenech Goulart mario.goul...@gmail.com wrote: On Thu, 23 Jan 2014 12:28:01 +0100 Christian Kellermann ck...@pestilenz.org wrote: I propose the following patch. It has been tested on OpenBSD and Linux. Should work without troubles on all other OSes as well. Please give it a try. Sorry for taking so long to give it a try on Windows. Unfortunately, it seems to break things on that platform. The environment variable tests in posix-tests.scm caught the issue. Here's an example: #;1 (use posix) ; loading c:/local/chicken-master-mingw/lib/chicken/7/posix.import.so ; ... ; loading library posix ... #;2 (setenv FOO bar) #;3 (get-environment-variable FOO) x\x01=b\x03 Reverting 8efbbc50d050d28c34ea0c5a58d0de92ec9f918a makes tests pass again. Tested on cygwin, mingw (Windows XP 32bit). I forgot to mention that on cygwin I had to define HAVE_STRLCAT and HAVE_STRLCPY in Makefile.cygwin, otherwise I'd get errors like http://paste.call-cc.org/paste?id=87b50bb84351693193183e7c58f2293b6cd52148 After defining HAVE_STRLCAT and STRLCPY and rebuilding chicken, I get the environment variables issue. Best wishes. Mario -- http://parenteses.org/mario ___ Chicken-hackers mailing list Chicken-hackers@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-hackers
Re: [Chicken-hackers] [PATCH] Remove list argument check from C_u_i_memq
On Wed, Feb 05, 2014 at 11:28:15PM +0100, Moritz Heidkamp wrote: This was erroneously added in 0a52536b7cb6b3d5a35ecc8f4c11131041ae873a Thanks, I've pushed it. Feel free to push patches that are as trivial as this directly in the future. Cheers, Peter -- http://www.more-magic.net ___ Chicken-hackers mailing list Chicken-hackers@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-hackers
