[Qgis-developer] Re: plug-in requirements

[Barry Rowlingson]

volkan kepoglu wrote:

> It would be nice to have a
> standard method for
>> plugins to test and report requirements before they install themselves
> properly. Checking for > things like rpy and specific R packages would be
> one of those things.

Here's some more thinking:

plugins should define a "requirements" function in their __init__.py 
file which will look something like this:

def requirements():
   reqs = {
     'qgis': ['>=0.9.0','<0.11.0'] ,
     'qgis-plugins': [],
     'python': '>=2.3.0',
     'python-packages': ['rpy'],
     'rpy-r': '>=2.4.0',
     'r-packages': ['DCluster'],
     'other': otherRequirements,
     }
   return reqs

  - so basically it returns a dict of name:value pairs which the 
requirements testing system can get. Note that the system can display 
the requirements as text, and also test the requirements. Obviously the 
system needs a way of testing each dependency type.

  The 'other' type is special, and its value is a function in the plugin 
that is called with one argument, 'test' and returns a list of two 
items. The first item is the test description, and the second item is 
the test status. Test status is a list of two items, the first is True 
or False for the test and the second is a note string. If the function 
is called with 'test=False', the status should be (None,None) and the 
test isn't performed.

  For example, here's an otherRequirements function that tests if the 
os.name parameter is 'nt' or 'posix':

  def otherRequirements(test=False):
   status=(None,None)
   if test:
     import os
     status=(False,'OS not valid')
     if os.name=="nt" or os.name=="posix":
       status=(True,'OS is valid')
   return ('OS is nt or posix',status)

  I'm not sure that returning a dict() from requirements is correct, 
since it might be neessary to impose an ordering on the testing (no 
point testing for R packages if rpy isn't installed, for example).

  Any other ideas?

Barry