Windows, r1219: Completion in script buffer doesn't work

[Richard Stanton]

System: OS X 10.8.3, Emacs 24.3.1, python-mode 1219, running the Anaconda python distribution (though I used to use EPD, and had very similar results) and using ipython as my default python shell.

The problem: Completion hasn't worked for ages with python-mode. Here's an example Python script file, say complete.py:

a = [1,2,3]

If I press C-c C-c, an Ipython buffer opens quite normally. Now suppose I go back to the script buffer and type (in the next line)

a.

and then, with the cursor right after the dot, go to the menu and select "PyTools" -> "Completion" -> "Complete symbol". No completion buffer pops up. Instead

1) In the script buffer, the cursor moves forward 6 spaces, leaving 6 new blank spaces after the dot.

2) In the Ipython buffer, I see a new input line that looks like this:

In [7]: a.append;a.count;a.extend;a.index;a.insert;a.pop;a.remove;a.reverse;a.sort

3) In the minibuffer, I see the message "Completion function py-shell-complete uses a deprecated calling convention"

Am I missing something? I've been unable to get completion to work for months now, which is quite a significant problem.

Let me know if I can do anything to help track this down.

Best,

Richard Stanton

[Richard Stanton (a-stanton)]

Followup:

I just tried the same experiment using EPD instead of Anaconda on the same machine. It fails as well, but in a different way. On selecting "Complete Symbol" as above,

1) There's a *long* wait (about 10-15 seconds, on a fast machine with lots of RAM) with Emacs unresponsive to key presses during this period.

2) An *IPython Completions* buffer does eventually pop up, but tells me "There are no possible completions of what you have typed."

[Richard Stanton (a-stanton)]

I have found a workaround, which is not to use python-mode's completion at all! I just tried opening up the EIN package, then loaded complete.py, and used EIN to connect the buffer to an existing IPython noteboook. Attached (because it's so lovely!) is a screenshot of what happens now when I type the dot after the a. Not only do I get a pop-up window with all the possible completions, but I also get a description of what each of them means in a separate (yellow) pop-up window. It would be great if python-mode.el could offer the same functionality.

[Andreas Roehler (a-roehler)]

Am 04.04.2013 18:59, schrieb Richard Stanton:
> I have found a workaround, which is not to use python-mode's completion
> at all! I just tried opening up the EIN package, then loaded
> complete.py, and used EIN to connect the buffer to an existing IPython
> noteboook. Attached (because it's so lovely!) is a screenshot of what
> happens now when I type the dot after the a. Not only do I get a pop-up
> window with all the possible completions, but I also get a description
> of what each of them means in a separate (yellow) pop-up window. It
> would be great if python-mode.el could offer the same functionality.
>
> ** Attachment added: "Screen Shot 2013-04-04 at 9.52.05 AM.png"
> https://bugs.launchpad.net/python-mode/+bug/1164183/+attachment/3621208/+files/Screen%20Shot%202013-04-04%20at%209.52.05%20AM.png
>

Thanks, Richard, nice indeed.

So what remains to do from python-mode side? Maybe just mention that way?
IMO we must stick with the default tools, can't require notebook.

IPython OTOH should be an option, will see what to do,

Andreas

[Richard Stanton (a-stanton)]

I agree we shouldn't require external tools or notebooks. This is a useful
work-around, but it would be better if python-mode could provide the same
completion functionality without needing an external package. Maybe you
could just borrow the code from EIN for completion using Ipython.

On 4/4/13 10:21 AM, "Andreas Roehler" <email address hidden> wrote:

>Am 04.04.2013 18:59, schrieb Richard Stanton:
>> I have found a workaround, which is not to use python-mode's completion
>> at all! I just tried opening up the EIN package, then loaded
>> complete.py, and used EIN to connect the buffer to an existing IPython
>> noteboook. Attached (because it's so lovely!) is a screenshot of what
>> happens now when I type the dot after the a. Not only do I get a pop-up
>> window with all the possible completions, but I also get a description
>> of what each of them means in a separate (yellow) pop-up window. It
>> would be great if python-mode.el could offer the same functionality.
>>
>> ** Attachment added: "Screen Shot 2013-04-04 at 9.52.05 AM.png"
>>
>>https://bugs.launchpad.net/python-mode/+bug/1164183/+attachment/3621208/+
>>files/Screen%20Shot%202013-04-04%20at%209.52.05%20AM.png
>>
>
>Thanks, Richard, nice indeed.
>
>So what remains to do from python-mode side? Maybe just mention that way?
>IMO we must stick with the default tools, can't require notebook.
>
>IPython OTOH should be an option, will see what to do,
>
>Andreas
>
>--
>You received this bug notification because you are subscribed to the bug
>report.
>https://bugs.launchpad.net/bugs/1164183
>
>Title:
> r1219: Completion in script buffer doesn't work
>
>Status in An Emacs mode for editing Python code:
> New
>
>Bug description:
> System: OS X 10.8.3, Emacs 24.3.1, python-mode 1219, running the
> Anaconda python distribution (though I used to use EPD, and had very
> similar results) and using ipython as my default python shell.
>
> The problem: Completion hasn't worked for ages with python-mode.
> Here's an example Python script file, say complete.py:
>
> a = [1,2,3]
>
>
> If I press C-c C-c, an Ipython buffer opens quite normally. Now suppose
>I go back to the script buffer and type (in the next line)
>
> a.
>
> and then, with the cursor right after the dot, go to the menu and
> select "PyTools" -> "Completion" -> "Complete symbol". No completion
> buffer pops up. Instead
>
> 1) In the script buffer, the cursor moves forward 6 spaces, leaving 6
> new blank spaces after the dot.
>
> 2) In the Ipython buffer, I see a new input line that looks like this:
>
> In [7]:
>
>a.append;a.count;a.extend;a.index;a.insert;a.pop;a.remove;a.reverse;a.sort
>
> 3) In the minibuffer, I see the message "Completion function py-shell-
> complete uses a deprecated calling convention"
>
> Am I missing something? I've been unable to get completion to work for
> months now, which is quite a significant problem.
>
> Let me know if I can do anything to help track this down.
>
> Best,
>
> Richard Stanton
>
>To manage notifications about this bug go to:
>https://bugs.launchpad.net/python-mode/+bug/1164183/+subscriptions

[Andreas Roehler (a-roehler)]

Being shameless enough to do that, if suitable :)

Am 04.04.2013 19:41, schrieb Richard Stanton:
> I agree we shouldn't require external tools or notebooks. This is a useful
> work-around, but it would be better if python-mode could provide the same
> completion functionality without needing an external package. Maybe you
> could just borrow the code from EIN for completion using Ipython.
>
>
> On 4/4/13 10:21 AM, "Andreas Roehler" <email address hidden> wrote:
>
>> Am 04.04.2013 18:59, schrieb Richard Stanton:
>>> I have found a workaround, which is not to use python-mode's completion
>>> at all! I just tried opening up the EIN package, then loaded
>>> complete.py, and used EIN to connect the buffer to an existing IPython
>>> noteboook. Attached (because it's so lovely!) is a screenshot of what
>>> happens now when I type the dot after the a. Not only do I get a pop-up
>>> window with all the possible completions, but I also get a description
>>> of what each of them means in a separate (yellow) pop-up window. It
>>> would be great if python-mode.el could offer the same functionality.
>>>
>>> ** Attachment added: "Screen Shot 2013-04-04 at 9.52.05 AM.png"
>>>
>>> https://bugs.launchpad.net/python-mode/+bug/1164183/+attachment/3621208/+
>>> files/Screen%20Shot%202013-04-04%20at%209.52.05%20AM.png
>>>
>>
>> Thanks, Richard, nice indeed.
>>
>> So what remains to do from python-mode side? Maybe just mention that way?
>> IMO we must stick with the default tools, can't require notebook.
>>
>> IPython OTOH should be an option, will see what to do,
>>
>> Andreas
>>
>> --
>> You received this bug notification because you are subscribed to the bug
>> report.
>> https://bugs.launchpad.net/bugs/1164183
>>
>> Title:
>> r1219: Completion in script buffer doesn't work
>>
>> Status in An Emacs mode for editing Python code:
>> New
>>
>> Bug description:
>> System: OS X 10.8.3, Emacs 24.3.1, python-mode 1219, running the
>> Anaconda python distribution (though I used to use EPD, and had very
>> similar results) and using ipython as my default python shell.
>>
>> The problem: Completion hasn't worked for ages with python-mode.
>> Here's an example Python script file, say complete.py:
>>
>> a = [1,2,3]
>>
>>
>> If I press C-c C-c, an Ipython buffer opens quite normally. Now suppose
>> I go back to the script buffer and type (in the next line)
>>
>> a.
>>
>> and then, with the cursor right after the dot, go to the menu and
>> select "PyTools" -> "Completion" -> "Complete symbol". No completion
>> buffer pops up. Instead
>>
>> 1) In the script buffer, the cursor moves forward 6 spaces, leaving 6
>> new blank spaces after the dot.
>>
>> 2) In the Ipython buffer, I see a new input line that looks like this:
>>
>> In [7]:
>>
>> a.append;a.count;a.extend;a.index;a.insert;a.pop;a.remove;a.reverse;a.sort
>>
>> 3) In the minibuffer, I see the message "Completion function py-shell-
>> complete uses a deprecated calling convention"
>>
>> Am I missing something? I've been unable to get completion to work for
>> months now, which is quite a significant problem.
>>
>> Let me know if I can do anything to help track this down.
...

[Andreas Roehler (a-roehler)]

Hi Richard,

the Python code used to fetch completions doesn't work at windows for me.
No idea why not.
Will remove the assignment, so maybe someone else will see it and send a patch.

Cheers,
Andreas

[Andreas Roehler (a-roehler)]

Works for me with current trunk. Please re-open if not done.