4

So the title kind of says it all. We are working on a project that is starting to get a very large number of dependencies. So far we were using setuptools but more and more of these dependencies are either not easy installable (eg. wxPython) or are causing some problems on some machines with easy_install (eg. latest matplotlib or numpy).

So we are looking at another solution where we would create our own python distribution that would hold all our dependencies to make using our application as easy as possible to the users. (Something like Enthought or Python(x,y) but on a much lower scale). Now my question would be: how difficult would be to do somehthing like this, and if you would have any suggestion as to where to start/ some documentation on this domain/ some tools that might held with this.

Regards, Bogdan

Andrey Agibalov
  • 7,624
  • 8
  • 66
  • 111
Bogdan
  • 8,017
  • 6
  • 48
  • 64

2 Answers2

1

conda constructor will allow you to create Anaconda like installer based on YAML config - python with modules in one self extracting archive - outcome packet no need compilation, no root access. Just run, and update PATH.

knightdave
  • 11
  • 1
0

cx_Freeze will allow you to have not only a separate python distribution, but a stripped-down one

Mihai Stan
  • 1,052
  • 6
  • 7
  • From their web page I see distributions for windows and CentOS. Any way to make it work on Mac ? – Bogdan Sep 01 '11 at 09:46
  • py2app - was answer in the link Arafagion posted (http://stackoverflow.com/questions/106725/how-to-bundle-a-python-application-including-dependencies-for-windows) – Mihai Stan Sep 01 '11 at 11:18
  • Yea I already gave py2app about two tries but couldn't get it working properly. – Bogdan Sep 01 '11 at 15:33