New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Windows: bundled ffi_c.so is generating invalid memory access #182
Comments
So, should I yank the mingw binary builds? If it isn't working on 1.9.x, then having broken binaries is worse than having people build from source, correct? |
Correct, please yank them and I'll investigate further. Sorry for top posting. Sent from mobile.
|
Ok, the mingw32 build is yanked. |
@wmeissner would you mind share with me your current mingw installation version ( |
|
Hello @wmeissner, I'll be emailing you offline about this, I'm not able to replicate corrupt binaries so I think something borked in your setup. |
Any progress about this issue? Since one of my gems (libusb) depends on ffi, it would be great, if a precompiled ffi.gem would be available again. Equally to @luislavena, I'm also unable to reproduce the segv, when using a cross compiled binary gem. My compilation environment is:
|
Anyone who is still interested, please test the latest win32 build (https://rubygems.org/gems/ffi/versions/1.0.12.rc2-x86-mingw32). That is built using the toolchain recommended by the rake-compiler docs, so hopefully works. |
Hello @wmeissner I think there is a problem with the gem you packaged:
It is missing 1.9 binaries, which seems to because the compilation only targeted Ruby 1.8.x Did you set |
Forgot to mention, but I did email you @wmeissner offline and never got an answer from you 😢 |
Ok, I installed 1.9.3-p125, and pushed a new gem. 1.0.12.rc3 should have both 1.8 and 1.9 .so in it. |
@wmeissner seems to have worked:
|
Thanks @luislavena! If you have time, can you also confirm that everything builds fine using the devkit? |
@wmeissner not so great: |
@wmeissner sorry, was sick the past couple of days (and still I am). Will take a look over the weekend. |
No problem, @luislavena, you've already gone above and beyond! I think I fixed the devkit build problem - it also affected linux/macos with no libffi installed, and was due to the libffi-3.0.10 import ... it seems to do strange things when built in its own source tree, so that part of the build is always built in libffi-$(arch), and it seems to fix the problem. |
@wmeissner seems what you pushed for
|
I tested ffi-1.1.0.rc3 with libusb-0.1.3 using devkit and precompiled binary gems and the Ruby versions:
and all combinations worked as expected. Thanks! |
Thanks @larskanis and @luislavena for testing this out. |
Hello @wmeissner , this issue is obviously solved, since the ffi-1.1.0-x86-mingw32.gem is working on ruby-installer 1.8.7 and 1.9.3. |
@larskanis thanks for confirming this. Closing this out! |
The text was updated successfully, but these errors were encountered: