arrow-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Pearu Peterson <>
Subject Re: Cant build py-arrow with CUDA
Date Wed, 13 Feb 2019 23:49:27 GMT
Hi Andy,

`make format` is just a convenience step when making development changes to
the arrow C++ files that fixes any source formatting issues that I might
have introduced when modifying the source files. You don't have to run it.

I cannot explain why `make format` is failing for you without seeing the
failure messages.

Note that the instructions were created just for myself, I just thought
that these might be useful for answering your questions.

Best regards,

On Thu, Feb 14, 2019 at 12:56 AM Andrew Palumbo <> wrote:

> Pearu,
> I just tried building py-arrow with cuda, using the cheat-sheet that you
> sent,  Thanks very much, all modules seem to be building, but I think there
> is a step that is missing,  I had a question that I was hoping you could
> answer; After setting up the environment, configuring and building, there
> is a comment in the `make format` line:
> make format # after changing cpp/ files
> could you explain what this means ( "After changing cpp/ files") I seem to be failing
> I can post a gist as wes mentioned if need be, but I believe that it has something to
 do with this comment.
> Thanks very much, again for the instructions.
> --andy
> ------------------------------
> *From:* Pearu Peterson <>
> *Sent:* Tuesday, February 12, 2019 5:24 PM
> *To:*
> *Subject:* Re: Cant build py-arrow with CUDA
> Hi,
> Here's mine cheat-sheet for building pyarrow with CUDA support (using
> conda dependencies):
> HTH,
> Pearu
> On Wed, Feb 13, 2019 at 12:11 AM Andrew Palumbo <>
> wrote:
> Hello,
> I've been trying to build py-arrow with CUDA support, unsuccessfully for
> some time.
> I've tried the directions here:
> with -DARROW_CUDA=ON in thr cmake build, with no luck.  I'm running
> Python3.6 on linux.
> I've gone through the cpp dir and the python dir and configured cmake with
> ccmake; No luck.  I have built py-arrow several times without CUDA support,
> but am very interested in the numba integration.
> Is there anything Obvious that I am missing?
> Thanks,
> Andy

View raw message