Le 30 déc. 2012 à 18:55, Steven Silvester <[email protected]> a
écrit :


Pierre,

Do you mean performance in terms of speed or capability?


I meant in terms of speed, because - from my perspective - the
introspection features powered by "rope" are the things I hate the most in
Spyder: it sometimes slows me down when I work on a slow machine, and they
are undoubtedly far slower than the Pydev equivalent features.

In terms of speed, it appears to be at best as fast, but generally much
slower than rope.  In terms of capability, the recursiveness of Jedi is
pretty impressive.  From the example on the site,

import time


class Foo(object):

global time

asdf = time

 def asdfy():

return Foo

 xorz = getattr(asdfy()(), 'asdf')

xorz.strftime()


I get the proper docstring when using Jedi, but not with rope. Perhaps the
easiest short-term solution is to use rope as is, and fall back on Jedi
when no matches are found?


Why not?
I see that you have open an issue on the GoogleCode project website. Things
will have to be done with the necessary precautions (library detection) as
for "rope".

Cheers
-Pierre

On Sunday, December 30, 2012 10:52:44 AM UTC-6, Pierre Raybaut wrote:
>
> Hi,
>
> Thanks a lot for this interesting patch.
> I'm especially interested in it as I was about to test the Jedi library to
> see what it has to offer. Simply by reading the feature list on the Jedi
> project website, I don't see a big difference with Rope features (am I
> wrong?) but I was hoping that it would have better performance. So what's
> about the performance gain from Jedi to Rope?
>
> Cheers,
> -Pierre
>
> Le 29 déc. 2012 à 21:45, Steven Silvester <[email protected]<javascript:>>
> a écrit :
>
> I took a stab at replacing rope with the jedi library, because it looks to
> be more promising based on the feature list at
> https://github.com/davidhalter/jedi.
> The replacement is fully functional with support for completions, call
> tips, docstrings and goto definition.
> However, it lacks two things found in the rope version: fast initial
> loading of large libraries like numpy and PyQt, and docstring support for
> functions with dynamic docstrings like functools.partial.
> Fixing these would take some similar patching as was done to the rope
> library, perhaps pushing back improvements to the jedi library.
>
> What do you guys think, is this worth pursuing further?
>
> I have attached the diff for the files affected:
> spyderlib/widgets/sourcecode/codeeditor.py
> spyderlib/widgets/editor.py
> spyderlib/plugins/inspector.py
>
>
> Regards,
> Steve Silvester
>
>  --
> You received this message because you are subscribed to the Google Groups
> "spyder" group.
> To view this discussion on the web visit
> https://groups.google.com/d/msg/spyderlib/-/V5TQxcxhG4AJ.
> To post to this group, send email to [email protected]<javascript:>
> .
> To unsubscribe from this group, send email to
> [email protected] <javascript:>.
> For more options, visit this group at
> http://groups.google.com/group/spyderlib?hl=en.
>
> <jedi.patch>
>
>  --
You received this message because you are subscribed to the Google Groups
"spyder" group.
To view this discussion on the web visit
https://groups.google.com/d/msg/spyderlib/-/HXlGLZ-2tEEJ.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/spyderlib?hl=en.

-- 
You received this message because you are subscribed to the Google Groups 
"spyder" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/spyderlib?hl=en.

Reply via email to