[
https://issues.apache.org/jira/browse/ARROW-7422?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Abishek Muthian closed ARROW-7422.
----------------------------------
Fix Version/s: 0.15.1
Resolution: Not A Bug
[~apitrou], sorry for the trouble this might not be a bug and due to user error
regarding environmental variables when using sudo to install pyarrow after
building.
I'm not sure why unavailability env vars can cause compiler to detect incorrect
CPU capabilities. But, if I use sudo -E to export env vars from the build
session this error doesn't occur.
Although there are similar errors such as the one I've posted here on the
github, since there's not clear evidence of an error within the project I'll
close this as 'Not A Bug'.
I've written a
[gist|https://gist.github.com/heavyinfo/04e1326bb9bed9cecb19c2d603c8d521]
detailing all the setups needed to build and install arrow, pyarrow in ARM if
anyone needed.
> [Python] Improper CPU flags failing pyarrow install in ARM devices
> ------------------------------------------------------------------
>
> Key: ARROW-7422
> URL: https://issues.apache.org/jira/browse/ARROW-7422
> Project: Apache Arrow
> Issue Type: Bug
> Components: Python
> Affects Versions: 0.15.1
> Environment: ARMv8 Jetson Nano with CUDA capable GPU
> Ubuntu 18.04.1
> gcc version 7.4.0
> cmake version 3.10.2
> Reporter: Abishek Muthian
> Priority: Major
> Fix For: 0.15.1
>
>
> Improper CPU flags failing build in-spite of routines explicitly checking
> whether compiler supports those flags. Since pyarrow package in the pip has
> configuration errors, building from the source is the only option but the
> installation fails due to several issues, this being one of them.
> Had to manually remove SSE4, AltiVec related flags from the
> *DefineOptions.cmake* & *SetupCxxFlags.cmake* to avoid this error in-spite of
> routines to check whether compiler can handle this flags.
> There are other errors after doing this(they don't seem to be related to CPU
> flags, so I have opened separate issue).
> Log is in the external issue URL.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
