Adding -m option to %run, similar to -m for python interpreter. #819

jstenar · 2011-09-22T17:42:12Z

I have long been missing an option to %run that is similar to -m for the regular python interpreter.

The -m option searches for a module on the python path and launches that module as a script.

This pull request is a first shot at this implementation. can you see anything that is missing?

I added some helper functions at top-level for locating modules, should they be somewhere else?

/Jörgen

takluyver · 2011-09-29T20:12:25Z

IPython/core/magic.py

                                          mode='list',list_all=1)
+        if opts.has_key("m"):
+            modulename = opts.get("m")[0]
+            arg_lst = [find_mod(modulename)]


This is currently going to replace all the arguments with the path to the module. Does Python itself allow you to do this sort of thing:

python -m somepkg.somemodule several args here

@jstenar, one more thing for when you get a chance: let's use if "m" in opts instead of has_key, which is the more modern way to spell it out. We may have lingering uses of has_key but we shouldn't be using that for new code.

And further, since you know that opts has m, then you can use modulename = opts['m'][0]. You only need to use get in dict access when you don't know for sure the key is available.

Should I remove old has_key uses in magic_run as well?

That's probably worth doing while you're working on it.

I just pushed some changes that address these issues. I also did some pep-8 cleaning of magic_run.

takluyver · 2011-09-29T20:16:02Z

Another quick question - if the user for some reason does %run -m (with no module name) or %run -m misspelt.modname, how gracefully will this fail?

jstenar · 2011-09-29T23:11:08Z

Thomas skrev 2011-09-29 22:16:

Another quick question - if the user for some reason does %run -m (with no module name) or %run -m misspelt.modname, how gracefully will this fail?

Thanks for your comments, you raise several good points I need to
adress. I guess that is why these reviews are a good idea :)

Unfortunatley I will not be able to look into this until late next week.

/Jörgen

takluyver · 2011-09-29T23:36:11Z

That's fine, there's no rush. I was just checking through the list of pull requests. Hope you're well!

jstenar · 2011-10-08T16:07:00Z

Running %run -m without argument will fail with an error during parse_options:

In [1]: %run -m
UsageError: option -m requires argument ( allowed: "nidtN:b:pD:l:rs:T:em:" )

fperez · 2011-10-08T20:39:58Z

@jstenar, that error mode seems OK, as it gives the user enough information on what the issue is, so I'm OK with that. Did you see a problem with it?

On the other hand, due to other recent merges, this PR doesn't currently merge, so you'll need to give it a rebase and do a push --force. Let us know if you need a hand with any of that...

Added helper functions at top-level, should they be somewhere else?

jstenar · 2011-10-09T12:43:13Z

I think I did the rebase correctly (my first one).

While I'm mucking about in magic_run I could take a stab at issue #710 as well or should I do that in a separate branch?

takluyver · 2011-10-09T16:18:14Z

I think let's do that in a separate branch - magic_run is some of the more thorny code, so it's worth keeping a clear eye on what's going on with it.

jstenar · 2011-10-09T18:03:59Z

Thomas skrev 2011-10-09 18:18:

I think let's do that in a separate branch - magic_run is some of the more thorny code, so it's worth keeping a clear eye on what's going on with it.

ok

fperez · 2011-10-09T19:36:03Z

IPython/core/magic.py

+def find_module(name, path=None):
+    """imp.find_module variant that only return path of module.
+    Return None if module is missing or does not have .py or .pyw extension
+    """


Let's try to follow the dev guidelines for docstrings; here it is for reference:

http://ipython.org/ipython-doc/dev/development/doc_guide.html#docstring-format

This ensures that we'll have consistency throughout, and that auto-generated API docs have the right signatures.

We just use the numpy standard, so it's nothing particularly complicated or special-case.

fperez · 2011-10-09T20:00:12Z

@jstenar, thanks for the updated work! I think we're almost there. But I see there are three new functions in the magic module now, so in addition to my in-line feedback, a couple of more general comments:

they are fairly generic utility functions related to module handling, so they should probably go elsewhere. I'd suggest a new module in IPython.utils, named something like modpaths.py. You can find a template with the necessary boilerplate at /docs/source/development/template.py.
Now that you have a nice new module with just three functions in it, the next thing is to ensure that you also add in utils/tests a file named test_modpaths.py that tests each one of these in isolation.

This way, the code will be easier to reuse elsewhere and will be validated. We want to move away from putting more and more tools into massive files like magic.py, that is already very unweildy.

So with a bit more iteration, we're close to having this solid for merge. Thanks for being patient, but that's how we ensure we build a more robust project moving forward!

jstenar · 2011-10-18T21:02:42Z

Fernando Perez skrev 2011-10-09 22:00:

@jstenar, thanks for the updated work! I think we're almost there. But I see there are three new functions in the magic module now, so in addition to my in-line feedback, a couple of more general comments:

they are fairly generic utility functions related to module handling, so they should probably go elsewhere. I'd suggest a new module in IPython.utils, named something like modpaths.py. You can find a template with the necessary boilerplate at /docs/source/development/template.py.

Now that you have a nice new module with just three functions in it, the next thing is to ensure that you also add in utils/tests a file named test_modpaths.py that tests each one of these in isolation.

This way, the code will be easier to reuse elsewhere and will be validated. We want to move away from putting more and more tools into massive files like magic.py, that is already very unweildy.

So with a bit more iteration, we're close to having this solid for merge. Thanks for being patient, but that's how we ensure we build a more robust project moving forward!

I have pushed changes related to these comments

fperez · 2011-10-18T23:30:17Z

@jstenar: yes, I'd seen that, and the PR was merged 4 days ago :) Good work!

jstenar · 2011-10-19T17:03:25Z

Fernando Perez skrev 2011-10-19 01:30:

@jstenar: yes, I'd seen that, and the PR was merged 4 days ago :) Good work!

cool, guess I missed that:)

The new -m option searches for a module on the python path and launches that module as a script. This is similar to Python's -m command-line flag in behavior, but while staying inside IPython. Conflicts: IPython/core/magic.py Closes ipythongh-819.

takluyver reviewed Sep 29, 2011
View reviewed changes

Jörgen Stenarson added 3 commits October 9, 2011 14:35

Adding -m option to %run, similar to -m for python interpreter.

55eff15

Added helper functions at top-level, should they be somewhere else?

Improve error handling for %run -m

42da76c

Replace has_key with in in magic_run and some pep-8 fixes.

269e9c9

fperez reviewed Oct 9, 2011
View reviewed changes

Jörgen Stenarson added 2 commits October 9, 2011 22:10

Fixing import statements and improving docstrings.

4811dd7

Moving helper functions to utils.module_paths, adding tests.

ea3a0fc

fperez closed this in a4cfe5c Oct 13, 2011

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Adding -m option to %run, similar to -m for python interpreter. #819

Adding -m option to %run, similar to -m for python interpreter. #819

jstenar commented Sep 22, 2011

takluyver Sep 29, 2011

fperez Oct 8, 2011

fperez Oct 8, 2011

jstenar Oct 8, 2011

takluyver Oct 8, 2011

jstenar Oct 8, 2011

takluyver commented Sep 29, 2011

jstenar commented Sep 29, 2011

takluyver commented Sep 29, 2011

jstenar commented Oct 8, 2011

fperez commented Oct 8, 2011

jstenar commented Oct 9, 2011

takluyver commented Oct 9, 2011

jstenar commented Oct 9, 2011

fperez Oct 9, 2011

fperez commented Oct 9, 2011

jstenar commented Oct 18, 2011

fperez commented Oct 18, 2011

jstenar commented Oct 19, 2011

Adding -m option to %run, similar to -m for python interpreter. #819

Adding -m option to %run, similar to -m for python interpreter. #819

Conversation

jstenar commented Sep 22, 2011

takluyver Sep 29, 2011

Choose a reason for hiding this comment

fperez Oct 8, 2011

Choose a reason for hiding this comment

fperez Oct 8, 2011

Choose a reason for hiding this comment

jstenar Oct 8, 2011

Choose a reason for hiding this comment

takluyver Oct 8, 2011

Choose a reason for hiding this comment

jstenar Oct 8, 2011

Choose a reason for hiding this comment

takluyver commented Sep 29, 2011

jstenar commented Sep 29, 2011

takluyver commented Sep 29, 2011

jstenar commented Oct 8, 2011

fperez commented Oct 8, 2011

jstenar commented Oct 9, 2011

takluyver commented Oct 9, 2011

jstenar commented Oct 9, 2011

fperez Oct 9, 2011

Choose a reason for hiding this comment

fperez commented Oct 9, 2011

jstenar commented Oct 18, 2011

fperez commented Oct 18, 2011

jstenar commented Oct 19, 2011