Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.python > #71708 > unrolled thread

using a new computer and bringing needed libraries to it

Back to article view | Back to comp.lang.python

Contents

  using a new computer and bringing needed libraries to it CM <cmpython@gmail.com> - 2014-05-17 16:53 -0700
    Re: using a new computer and bringing needed libraries to it Ned Batchelder <ned@nedbatchelder.com> - 2014-05-17 20:17 -0400
      Re: using a new computer and bringing needed libraries to it Rustom Mody <rustompmody@gmail.com> - 2014-05-17 20:29 -0700
        Re: using a new computer and bringing needed libraries to it Ben Finney <ben@benfinney.id.au> - 2014-05-18 17:39 +1000
    Re: using a new computer and bringing needed libraries to it Terry Reedy <tjreedy@udel.edu> - 2014-05-17 20:29 -0400
    Re: using a new computer and bringing needed libraries to it Chris Angelico <rosuav@gmail.com> - 2014-05-18 10:30 +1000
    Pip requirements: Machine-readable configuration versus human-audience documentation (was: using a new computer and bringing needed libraries to it) Ben Finney <ben@benfinney.id.au> - 2014-05-18 12:01 +1000

#71708 — using a new computer and bringing needed libraries to it

If I want to switch my work from one computer to a new one, and I have lots of various libraries installed on the original computer, what's the best way to switch that all to the new computer?  I'm hoping there is some simple way like just copying the Python/Lib/site-packages folder, but I'm also guessing this isn't sufficient. I was hoping I wouldn't have to just one-by-one install all of those libraries again on the newer computer.

I probably want to develop on BOTH these computers for the time being, too.  One is at home and one is at a "remote site"/secret lair.  And then I'll be doing it again when I buy a newer computer at some point.

Thanks.

[toc] | [next] | [standalone]

#71711

On 5/17/14 7:53 PM, CM wrote:
> If I want to switch my work from one computer to a new one, and I have lots of various libraries installed on the original computer, what's the best way to switch that all to the new computer?  I'm hoping there is some simple way like just copying the Python/Lib/site-packages folder, but I'm also guessing this isn't sufficient. I was hoping I wouldn't have to just one-by-one install all of those libraries again on the newer computer.
>
> I probably want to develop on BOTH these computers for the time being, too.  One is at home and one is at a "remote site"/secret lair.  And then I'll be doing it again when I buy a newer computer at some point.
>
> Thanks.
>

Make a list of the packages you need.  Put it in a file called 
requirements.txt.  Then install them with:

     $ pip install -r requirements.txt

Keep that file up-to-date as you add new requirements.

-- 
Ned Batchelder, http://nedbatchelder.com

[toc] | [prev] | [next] | [standalone]

#71719

On Sunday, May 18, 2014 5:47:05 AM UTC+5:30, Ned Batchelder wrote:
> On 5/17/14 7:53 PM, CM wrote:
> 

> > If I want to switch my work from one computer to a new one, and I
> > have lots of various libraries installed on the original computer,
> > what's the best way to switch that all to the new computer?  I'm
> > hoping there is some simple way like just copying the
> > Python/Lib/site-packages folder, but I'm also guessing this isn't
> > sufficient. I was hoping I wouldn't have to just one-by-one
> > install all of those libraries again on the newer computer.

> > I probably want to develop on BOTH these computers for the time
> > being, too.  One is at home and one is at a "remote site"/secret
> > lair.  And then I'll be doing it again when I buy a newer computer
> > at some point.

> 
> Make a list of the packages you need.  Put it in a file called 
> requirements.txt.  Then install them with:
> 
> 
>      $ pip install -r requirements.txt
> 
> 
> Keep that file up-to-date as you add new requirements.


What about things installed at a lower level than pip, eg apt-get?

[toc] | [prev] | [next] | [standalone]

#71728

Rustom Mody <rustompmody@gmail.com> writes:

> On Sunday, May 18, 2014 5:47:05 AM UTC+5:30, Ned Batchelder wrote:
> > Make a list of the [Python-specific] packages you need. Put it in a
> > file called requirements.txt. […]
>
> What about things installed at a lower level than pip, eg apt-get?

That's an important issue. Requirements of a project, such as packages
that need to be installed from the operating system (e.g. “you need
Python 3.2 or later for this project”), are ideal for documenting in
plain human-targeted text in a document called “requirements.txt”.

Which is why I advocate using a *different* filename, more explicit
about its special purpose (e.g. ‘pip_requirements’), for the
Pip-specific (and thereby Python-specific) machine-readable
configuration file.

-- 
 \       “But Marge, what if we chose the wrong religion? Each week we |
  `\          just make God madder and madder.” —Homer, _The Simpsons_ |
_o__)                                                                  |
Ben Finney

[toc] | [prev] | [next] | [standalone]

#71714

On 5/17/2014 7:53 PM, CM wrote:
> If I want to switch my work from one computer to a new one, and I
> have lots of various libraries installed on the original computer,
> what's the best way to switch that all to the new computer?  I'm
> hoping there is some simple way like just copying the
> Python/Lib/site-packages folder, but I'm also guessing this isn't
> sufficient.

Have your tried it? Since Python only cares about the contents of 
site-packages, copying should be fine, at least as far as python is 
concerned. I have copied pythonx.y/Lib/site-packages to 
pythonx.(y+1)/Lib/site-packages more than once.

In each site-packages, I also have python.pth containing, in my case, 
"F:/Python". packages and modules in F:/Python are imported the same as 
if they were in each site-packages. This avoids copying and lets me try 
the same file on multiple versions.

Copying does not copy registry entries or anything outside of 
site-packages. I do not know whether pip, for instance, does either.

-- 
Terry Jan Reedy

[toc] | [prev] | [next] | [standalone]

#71715

On Sun, May 18, 2014 at 10:17 AM, Ned Batchelder <ned@nedbatchelder.com> wrote:
> Make a list of the packages you need.  Put it in a file called
> requirements.txt.  Then install them with:
>
>     $ pip install -r requirements.txt
>
> Keep that file up-to-date as you add new requirements.

+1. And the "keep up-to-date" bit can be done very well with source
control; that way, you don't need to wonder whether you added one over
here or deleted one over there - the commit history will tell you.

ChrisA

[toc] | [prev] | [next] | [standalone]

#71716 — Pip requirements: Machine-readable configuration versus human-audience documentation (was: using a new computer and bringing needed libraries to it)

Ned Batchelder <ned@nedbatchelder.com> writes:

> Make a list of the packages you need.  Put it in a file called
> requirements.txt.  Then install them with:
>
>     $ pip install -r requirements.txt
>
> Keep that file up-to-date as you add new requirements.

Since these requirements are specifically for Python, more specifically
for Pip, and even more specifically are supposed to be in a
machine-readable foramt and not just an arbitrary free-form text
document, can we recommend instead some more specific filename?

‘requirements.txt’ is already used in many projects to document *for a
human reader* the project-wide requirements, not jsut for Python, and we
should not arrogate a general name like that to a specific tool like
Pip.

I'd recommend (and have already begun to use) the name
‘pip_requirements’ or the like. I know that there is heaps of
Pip-specific documentation out there already recommending the more
general name, but I'd like that to change.

-- 
 \             “I believe our future depends powerfully on how well we |
  `\     understand this cosmos, in which we float like a mote of dust |
_o__)                 in the morning sky.” —Carl Sagan, _Cosmos_, 1980 |
Ben Finney

[toc] | [prev] | [standalone]

Back to top | Article view | comp.lang.python

csiph-web