Move towards mingw-w64 (GCC 4.6.3)

[Luis Lavena]

Hello,

I would like to move default DevKit used when checking out
RubyInstaller repository to mingw-w64 GCC 4.6.3 (identified as
mingw64-32-4.6.3)

The builds are provided by Ruben Van Boxem (rubemvb) and are
considered the upcoming packing for the mingw-w64 builds.

There is a heated conversation about GCC threading model (win32, posix
and possible a 3rd one). At this time posix threading forces any
compilation with C++ depend on pthread-compatible library winpthreads
instead of using native threading of Windows.

For the time being and to avoid these dependencies I'm recommending
win32 threading builds (which is the one currently defined)

There is also the goal of move to mingw-w64 entirely for both 32 and
64bits, which this default will help toward that goal.

If nobody objects I'm going to change the default during this weekend.

Regards,
--
Luis Lavena
AREA 17
-
Perfection in design is achieved not when there is nothing more to add,
but rather when there is nothing more to take away.
Antoine de Saint-Exupéry

[Jon]

On Saturday, May 19, 2012 4:30:57 PM UTC-4, Luis Lavena wrote:

Hello,

I would like to move default DevKit used when checking out
RubyInstaller repository to mingw-w64 GCC 4.6.3 (identified as
mingw64-32-4.6.3)

The builds are provided by Ruben Van Boxem (rubemvb) and are
considered the upcoming packing for the mingw-w64 builds.

There is a heated conversation about GCC threading model (win32, posix
and possible a 3rd one). At this time posix threading forces any
compilation with C++ depend on pthread-compatible library winpthreads
instead of using native threading of Windows.

For the time being and to avoid these dependencies I'm recommending
win32 threading builds (which is the one currently defined)

There is also the goal of move to mingw-w64 entirely for both 32 and
64bits, which this default will help toward that goal.

If nobody objects I'm going to change the default during this weekend.

 

Sadly, with mingw64-32-4.6.3 at ruby_1_9_3@35717 (mini-SHA: 5d43f34), I get 5 errors/fails when testing with:

 

  make test-all TESTS='openssl fiddle psych zlib io json mkmf pathname stringio erb fileutils ruby/test_io.rb ruby/test_io_m17n.rb ruby/test_file.rb'

 

No time to investigate due to other priorities.

 

Jon

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

[Luis Lavena]

On Sun, May 20, 2012 at 2:27 PM, Jon <jon.f...@gmail.com> wrote:
>
> Sadly, with mingw64-32-4.6.3 at ruby_1_9_3@35717 (mini-SHA: 5d43f34), I get
> 5 errors/fails when testing with:
>
>   make test-all TESTS='openssl fiddle psych zlib io json mkmf pathname
> stringio erb fileutils ruby/test_io.rb ruby/test_io_m17n.rb
> ruby/test_file.rb'
>
> No time to investigate due to other priorities.
>

I will investigate. At least I have now a dedicated i5 worker machine
to perform compilations, which is making things more easy as I can
fire parallel testing and not burn my lap :-D

Keep you posted.

[Luis Lavena]

On Sun, May 20, 2012 at 2:27 PM, Jon <jon.f...@gmail.com> wrote:
>

> Sadly, with mingw64-32-4.6.3 at ruby_1_9_3@35717 (mini-SHA: 5d43f34), I get
> 5 errors/fails when testing with:
>
>   make test-all TESTS='openssl fiddle psych zlib io json mkmf pathname
> stringio erb fileutils ruby/test_io.rb ruby/test_io_m17n.rb
> ruby/test_file.rb'
>

I can reproduce at r35732:

https://gist.github.com/2759782

Configure options where:
--enable-shared --disable-install-doc debugflags=-g
CPPFLAGS='-DFD_SETSIZE=32767'

Ruby was build with Knapsack binaries.

Of course the first "failure" is caused by OpenSSL configuration file
and looking into absolute path when build with knap-build.
(OT: If anyone knows how to make OpenSSL configure use a relative path
(so we can avoid the hardcoding) will be much appreciated.)

I didn't obtain the other two failures when used tdm-32-4.5.2, which
might be indicator of the problem you're mentioning.

I guess will have to hold my horses on mingw-w64... for now :-)

[Jon]

On Sunday, May 20, 2012 6:53:55 PM UTC-4, Luis Lavena wrote:

On Sun, May 20, 2012 at 2:27 PM, Jon <jon.f...@gmail.com> wrote:
>
> Sadly, with mingw64-32-4.6.3 at ruby_1_9_3@35717 (mini-SHA: 5d43f34), I get
> 5 errors/fails when testing with:
>
>   make test-all TESTS='openssl fiddle psych zlib io json mkmf pathname
> stringio erb fileutils ruby/test_io.rb ruby/test_io_m17n.rb
> ruby/test_file.rb'
>

I can reproduce at r35732:

https://gist.github.com/2759782

Configure options where:
--enable-shared --disable-install-doc debugflags=-g
CPPFLAGS='-DFD_SETSIZE=32767'

Ruby was build with Knapsack binaries.

Of course the first "failure" is caused by OpenSSL configuration file
and looking into absolute path when build with knap-build.
(OT: If anyone knows how to make OpenSSL configure use a relative path
(so we can avoid the hardcoding) will be much appreciated.)

I didn't obtain the other two failures when used tdm-32-4.5.2, which
might be indicator of the problem you're mentioning.

I guess will have to hold my horses on mingw-w64... for now :-)

 

I got the same two `ruby\test_file.rb` fails and a few more, but I don't recall OpenSSL fails.

 

I'll build again tomorrow and post a gist.

[Jon]

the build fails in all their glory

  https://gist.github.com/2763401

 

[Luis Lavena]

On Mon, May 21, 2012 at 2:27 PM, Jon <jon.f...@gmail.com> wrote:
>
>
> the build fails in all their glory
>
>   https://gist.github.com/2763401
>

I was testing against trunk, will do another test against ruby_1_9_3
later today, keep you posted.