-
-
Notifications
You must be signed in to change notification settings - Fork 1.2k
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
Python 3.6.8 - fix cross-compilation issue #3726
Conversation
(repost with correct ID) In #3673 comment #3673 (comment) we discussed how to check that the resulting cross-compiled binary works. I think we're missing proof that the ctypes and some cffi can load correctly as well. I am worried about the pathnames a bit, and I don't think repeating any errors from The key is whether the binary built on one arch runs correctly in the other, which you have shown to some extent, but may not highlight the nuance of the error previously noticed downstream of the last builds. On your x86_64-built python-3.6.8 for the armada370, please run the following, and please edit your top "is that OK?" to include this so that it's really obvious for the next iteration.
Specifically, I think the |
I edited my previous post. |
I think this is a successful sign though. Yet another check (platform.system) to try to detect issues:
Does the "machine:" value look correct on your armada370 ? My hidden agenda here is to be able to semi-automate later release validation. Looking for a value on your armada370 HW that says something different than the |
|
Looks good to me then; reduces the customization/patch load by 56 lines or so. Yves ? |
I'm working on another patch that also solves the problem of library and wheel names. |
@roleoroleo |
Unfortunately homeassistant does not work. |
Yeah I also noticed that. Is bcrypt really needed then? |
I created a new PR to solve the bcrypt problem. |
@roleoroleo /edit: /edit2: /edit: |
@@ -17,7 +17,8 @@ LICENSE = PSF | |||
|
|||
GNU_CONFIGURE = 1 | |||
ADDITIONAL_CFLAGS = -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -L $(STAGING_INSTALL_PREFIX)/lib -I $(STAGING_INSTALL_PREFIX)/include | |||
CONFIGURE_ARGS = --enable-shared --enable-ipv6 --without-ensurepip --enable-loadable-sqlite-extensions PYTHON_FOR_BUILD=$(HOSTPYTHON) PGEN_FOR_BUILD=$(HOSTPGEN) | |||
#CONFIGURE_ARGS = --enable-shared --enable-ipv6 --without-ensurepip --enable-loadable-sqlite-extensions PYTHON_FOR_BUILD=$(HOSTPYTHON) PGEN_FOR_BUILD=$(HOSTPGEN) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I would simply discard old line...
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sorry but I deleted all my forked repos and I don't known how to change the PR.
I am processing builds to confirm it works well... Then I would publish my own borgbackup (and maybe homeassistant) updates... |
Repository and branch no longer available. Superseded by #3731 |
Motivation: I'm trying to use Home Assistant anda the last version deprecates Python 3.5
Linked issues: #3673
I cross-compiled it on x86_64 using a debian 9.9 with docker.
And I was able to test it on armada370.
I realized that the wheel files and the libraries contained still have the wrong notation "linux_x86_64" even if they are arm binaries.
Looking at the previous python 3.5.6 package I noticed that the situation is similar.
This is the output of the sys import executed on my DS214se: