It'd be nice if 1.1.6.pre1 in particular got some testing love by people running ruby-head.

On Friday, 24 August 2012 02:28:01 UTC+10, Jon wrote:

:)

[jon@archee ffi-git]$ git log -1 --oneline
20c243a Add back ffiStatus lvar to rbffi_MethodHandle_Init()

[jon@archee ffi-git]$ gem i pkg/ffi-1.2.0.devmaster.gem --user-install
Building native extensions.  This could take a while...
Successfully installed ffi-1.2.0.devmaster
1 gem installed


> eek => MethodHandle.c:340:5: error: 'ffiStatus' undeclared (first use in this function)
>
>
> [jon@archee ~]$ gem i ffi --pre --user-install
> Fetching: ffi-1.2.0.dev3.gem (100%)
> Building native extensions.  This could take a while...
> ERROR:  Error installing ffi:
>         ERROR: Failed to build gem native extension.
>
>         /usr/local/bin/ruby extconf.rb
> checking for ffi_call() in -lffi... yes
> checking for ffi_prep_closure()... yes
> checking for ffi_raw_call()... yes
> checking for ffi_prep_raw_closure()... yes
> checking for rb_thread_blocking_region()... yes
> checking for ruby_native_thread_p()... yes
> checking for rb_thread_call_with_gvl()... yes
> checking for ffi_prep_cif_var()... yes
> creating extconf.h
> creating Makefile
>
> make
> compiling Pointer.c
> compiling Platform.c
> compiling Types.c
> compiling MappedType.c
> compiling Buffer.c
> compiling Call.c
> Call.c: In function 'do_blocking_call':
> Call.c:281:5: warning: 'rb_thread_blocking_region' is deprecated (declared at /usr/local/include/ruby-2.0.0/ruby/intern.h:823) [-Wdeprecated-declarations]
> compiling LongDouble.c
> compiling ClosurePool.c
> compiling StructByReference.c
> compiling ArrayType.c
> compiling Variadic.c
> compiling MemoryPointer.c
> compiling Struct.c
> compiling AbstractMemory.c
> compiling MethodHandle.c
> MethodHandle.c: In function 'rbffi_MethodHandle_Init':
> MethodHandle.c:340:5: error: 'ffiStatus' undeclared (first use in this function)
> MethodHandle.c:340:5: note: each undeclared identifier is reported only once for each function it appears in
> make: *** [MethodHandle.o] Error 1
>
>
>
> > Can you try 1.2.0.dev3 ?  It should eliminate the warnflags MRI puts in
> > there.
> >
> > On Friday, 10 August 2012 06:16:10 UTC+10, Jon wrote:
> > >
> > > With ruby trunk the last few ffi gem releases have failed to build on my
> > > Arch 3.4.7 32bit dev system with gcc 4.7.1 (20120721).
> > >
> > > Given the details and workaround info below, the root cause appears to be
> > > recent `configure.in` ansi compliance mods:
> > >
> > >   https://github.com/ruby/ruby/blob/trunk/configure.in#L619-635
> > >
> > > Looks like others are running into as well
> > >
> > >   https://github.com/flori/json/issues/142
> > >
> > > While I've only briefly investigated, I don't have time to fully look into
> > > it. But I don't want to forget about it.
> > >
> > >
> > >
> > > *** GEM UPDATE FAILURE ***
> > >
> > > [jon@archee ~]$ gem update ffi --user-install
> > > Updating installed gems
> > > Updating ffi
> > > Fetching: ffi-1.1.5.gem (100%)
> > > Building native extensions.  This could take a while...
> > > ERROR:  Error installing ffi:
> > >         ERROR: Failed to build gem native extension.
> > >
> > >         /usr/local/bin/ruby extconf.rb
> > > checking for ffi_call() in -lffi... yes
> > > checking for ffi_prep_closure()... yes
> > > checking for ffi_raw_call()... yes
> > > checking for ffi_prep_raw_closure()... yes
> > > checking for rb_thread_blocking_region()... yes
> > > checking for ruby_native_thread_p()... yes
> > > checking for rb_thread_call_with_gvl()... yes
> > > creating extconf.h
> > > creating Makefile
> > >
> > > make
> > > compiling Pointer.c
> > > In file included from Pointer.c:34:0:
> > > rbffi_endian.h:46:4: error: #error "Cannot determine the endian-ness of
> > > this platform"
> > > Pointer.c: In function 'ptr_order':
> > > Pointer.c:339:59: error: 'BYTE_ORDER' undeclared (first use in this
> > > function)
> > > Pointer.c:339:59: note: each undeclared identifier is reported only once
> > > for each function it appears in
> > > Pointer.c:339:72: error: 'BIG_ENDIAN' undeclared (first use in this
> > > function)
> > > Pointer.c:351:25: error: 'LITTLE_ENDIAN' undeclared (first use in this
> > > function)
> > > Pointer.c:368:1: warning: control reaches end of non-void function
> > > [-Wreturn-type]
> > > make: *** [Pointer.o] Error 1
> > >
> > >
> > > Gem files will remain installed in
> > > /home/jon/.gem/ruby/2.0.0/gems/ffi-1.1.5 for inspection.
> > > Results logged to
> > > /home/jon/.gem/ruby/2.0.0/gems/ffi-1.1.5/ext/ffi_c/gem_make.out
> > > Nothing to update
> > >
> > >
> > >
> > > *** WORKAROUND ***
> > >
> > > gem update ffi --user-install -- --with-cflags='--std=gnu99'
> > >
> > >
> > >
> > > *** SYSTEM INFO ***
> > >
> > > # rbconfig.rb contents:
> > >   CONFIG["warnflags"] = "-Wall -Wextra -Wno-unused-parameter
> > > -Wno-parentheses -Wno-long-long -Wno-missing-field-initializers
> > > -Wunused-variable -Werror=pointer-arith -Werror=write-strings
> > > -Werror=declaration-after-statement -Werror=implicit-function-declaration
> > > -ansi -std=iso9899:199409"
> > >   CONFIG["WERRORFLAG"] = "-Werror"
> > >
> > >
> > > [jon@archee ~]$ gem env
> > > RubyGems Environment:
> > >   - RUBYGEMS VERSION: 1.8.24
> > >   - RUBY VERSION: 2.0.0 (2012-08-10 patchlevel -1) [i686-linux]
> > >   - INSTALLATION DIRECTORY: /usr/local/lib/ruby/gems/2.0.0
> > >   - RUBY EXECUTABLE: /usr/local/bin/ruby
> > >   - EXECUTABLE DIRECTORY: /usr/local/bin
> > >   - RUBYGEMS PLATFORMS:
> > >     - ruby
> > >     - x86-linux
> > >   - GEM PATHS:
> > >      - /usr/local/lib/ruby/gems/2.0.0
> > >      - /home/jon/.gem/ruby/2.0.0
> > >   - GEM CONFIGURATION:
> > >      - :update_sources => true
> > >      - :verbose => true
> > >      - :benchmark => false
> > >      - :backtrace => false
> > >      - :bulk_threshold => 1000
> > >      - :sources => ["http://rubygems.org"]
> > >      - "gem" => "--no-ri --no-rdoc"
> > >   - REMOTE SOURCES:
> > >      - http://rubygems.org
> > >
> > >
> > > Jon
> > >
> > > ---
> > > Fail fast. Fail often. Fail publicly. Learn. Adapt. Repeat.
> > > http://thecodeshop.github.com | http://jonforums.github.com/
> > > twitter: @jonforums
> > >
>
>
> Jon
>
> ---
> Fail fast. Fail often. Fail publicly. Learn. Adapt. Repeat.
> http://thecodeshop.github.com | http://jonforums.github.com/
> twitter: @jonforums


Jon

---
Fail fast. Fail often. Fail publicly. Learn. Adapt. Repeat.
http://thecodeshop.github.com | http://jonforums.github.com/
twitter: @jonforums