Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

For example I want to install the latest version of "numpy". I type the following: "sudo apt-get install python-numpy". When I type this the first time it installs something and if I type this the second time it writes that I have already the latest version of numpy.

However, I see that my version of numpy is 1.1.1. and I know that it NOT the latest version. Why it happens and how this problem can be solved?

I can find the *tar.gz file with the latest version, I can extract files with the archive and than I need to rune one of the scripts which will be somewhere among the extracted files. But I do not like this way. It is too complicated. I do not know where I should put all these files, I do not know which dependencies I should install before I run the script for the installation of numpy, I do not know where numpy will be put after installation and so on.

Is there an easy way to get the latest version of numpy?

share|improve this question

migrated from stackoverflow.com Jan 13 '10 at 18:48

This question came from our site for professional and enthusiast programmers.

3  
should be on serverfault.com or superuser.com –  gingerbreadboy Jan 13 '10 at 18:43
    
As of 09-Jan-2013, here's a PPA with more recent numpy then LTS's sources. You generally find such things by searching for a package here, and then look at the bottom of the page: "Other versions of 'python-numpy' in untrusted archives.". –  Adobe Jan 9 '13 at 13:03
    
Finally there's an option of cloning the source, building rpm with bdist rpm (also don't forget to specify the correct fortran compiler), and tweak the package name in the setup.py) option to setup.py, then converting the rpm to deb with alien. Then installing the deb. This will get You latest numpy with You distribution understanding it correctly. –  Adobe Jan 9 '13 at 13:06
add comment

2 Answers 2

up vote 9 down vote accepted

Distributions (such as Ubuntu) do QA on their packages and so will not always have the latest version of everything available in their repositories as soon as they come out. The emphasis is on stability.

If you really need a more recent version for some reason, you can use Python's own package management stuff (setuptools, in case you need to install it), like so:

python easy_install numpy

This will install the latest version available on PyPI.

I would recommend removing the package installed by the distribution first, since they probably don't play well together in the event of conflicting paths.

share|improve this answer
2  
i.e. sudo apt-get purge python-numpy first –  Mark Mar 17 '10 at 5:04
add comment

As Ubuntu is released every 6 months, Cannonical includes the latest stable version of a package when they begin testing of the next release. For example, even today (Jan 2010), Ubuntu 8.04 LTS still runs Firefox 3 rather than Firefox 3.5.

There are some intrepid hackers who want Firefox 3.5 on ther Ubuntu 8.04 LTS, and these people are able to obtain and maintain mini-repositories on Launchpad. For example, here is a link explaining how to use a Launchpad repo to get Firefox 3.5.

As it turns out, numpy has a launchpad PPA repository, in fact it has three, one for Jaunty, one for Intrepid, and one for Hardy. But, these Launchpad PPA releases happen to be older versions (current for 9.10 is python-numpy 1:1.3.1-3, the PPA releases are 1.2.1 or 1.1.1). If you want the latest version of numpy available from a repository, you may need to seek fame and fortune as an intrepid hacker and put together a Launchpad PPA for the version you need.

I have been very thankful for the things people have put on launchpad, not being an intrepid hacker myself.

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.