bottleneck on Mac OSX 10.8

84 views
Skip to first unread message

Keith Goodman

unread,
Aug 27, 2012, 4:00:36 PM8/27/12
to bottl...@googlegroups.com
Has anyone gotten bottleneck to work on OSX 10.8?

Here's a report that says it doesn't work:

https://github.com/kwgoodman/bottleneck/issues/49

dbv

unread,
May 23, 2013, 4:04:37 PM5/23/13
to bottl...@googlegroups.com
Hi!  Just tried to install Bottleneck 0.6 on OSX 10.8.3 and obtained a similar set of errors - in fact, 684 warning messages!

I am attempting to install Bottleneck 0.6 for 32-bit Python 2.7.5 under OSX.  As 32-bit binaries are required, I guess this means downloading the Bottleneck source files and compiling with Cython as 32-bit?  Thx.

Keith Goodman

unread,
May 23, 2013, 4:25:57 PM5/23/13
to bottl...@googlegroups.com
On Thu, May 23, 2013 at 1:04 PM, dbv <dinesh...@hotmail.com> wrote:
Hi!  Just tried to install Bottleneck 0.6 on OSX 10.8.3 and obtained a similar set of errors - in fact, 684 warning messages!

I am attempting to install Bottleneck 0.6 for 32-bit Python 2.7.5 under OSX.  As 32-bit binaries are required, I guess this means downloading the Bottleneck source files and compiling with Cython as 32-bit?  Thx.

So downloading from github and then running

$ make all

should build it locally (in same directory as source).

You'll need cython. 32-bit will be automatically detected.

dbv

unread,
May 24, 2013, 10:26:46 AM5/24/13
to bottl...@googlegroups.com
Built successfully on OSX 10.8.3.  

It did take over 30 mins to build though (running on a pretty fast mac mini).

Btw, does Cython automatically detect if 32-bit or 64-bit compilation is needed?


Keith Goodman

unread,
May 24, 2013, 12:24:24 PM5/24/13
to bottl...@googlegroups.com
On Fri, May 24, 2013 at 7:26 AM, dbv <dinesh...@hotmail.com> wrote:

Built successfully on OSX 10.8.3.

Yay!
 

It did take over 30 mins to build though (running on a pretty fast mac mini).

Ugh.
 

Btw, does Cython automatically detect if 32-bit or 64-bit compilation is needed?

Bottleneck detects it runtime through the use of np.int_. On my 64-bit system:

In [1]: np.int_?
Type:       type
Base Class: <type 'type'>
String Form:<type 'numpy.int64'>
Namespace:  Interactive
File:       /usr/local/lib/python2.7/dist-packages/numpy/__init__.py
Docstring:  64-bit integer. Character code 'l'. Python int compatible.

dbv

unread,
May 28, 2013, 9:13:29 AM5/28/13
to bottl...@googlegroups.com
Btw, a "pip install bottleneck" on a 32-bit ubuntu 12.04 took a couple of minutes to install.  Hth.
Reply all
Reply to author
Forward
0 new messages