matplotlib-users

[Matplotlib-users] Has nxutils been removed from mpl?

[Jorge S. <jor...@ya...>]

Hi,
I've had an import error with some of my code, specifically with nxutils (I need
it for using points_inside_poly). 

In [1]: import matplotlib.nxutils as nx
---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
/home/jscandal/<ipython-input-1-c56b9729b42e> in <module>()
----> 1 import matplotlib.nxutils as nx

ImportError: No module named nxutils

I built mpl from trunk and tried also version 1.1 from my distro, but neither
seems to have nxutils. I still find the examples in the docs using it, though,
so I am confused.
Has nxutils been replaced or removed? 

Thanks

Jorge

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Benjamin R. <ben...@ou...>]

On Wed, Mar 7, 2012 at 10:46 AM, Jorge Scandaliaris
<jor...@ya...>wrote:

> Hi,
> I've had an import error with some of my code, specifically with nxutils
> (I need
> it for using points_inside_poly).
>
> In [1]: import matplotlib.nxutils as nx
> ---------------------------------------------------------------------------
> ImportError                               Traceback (most recent call last)
> /home/jscandal/<ipython-input-1-c56b9729b42e> in <module>()
> ----> 1 import matplotlib.nxutils as nx
>
> ImportError: No module named nxutils
>
> I built mpl from trunk and tried also version 1.1 from my distro, but
> neither
> seems to have nxutils. I still find the examples in the docs using it,
> though,
> so I am confused.
> Has nxutils been replaced or removed?
>
> Thanks
>
> Jorge
>
>
Last I heard, it was supposed to be deprecated in master (although it was
accidentially removed at one point).  Indeed, it appears to be missing from
the current master.  The deprecation process must be followed! Personally,
I think such removal is short-sighted (nxutil's functions are much faster
than their python counter-parts).

However, it was still in v1.1.x.  I suspect that when you tested v1.1.x,
you were really running the source build.  Verify which version you are
testing with:

import matplotlib
print matplotlib.__version__
print matplotlib.__file__

Ben Root

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Jorge S. <jor...@ya...>]

Benjamin Root <ben.root@...> writes:

> 
< snip >
>
> import matplotlib
> print matplotlib.__version__
> print matplotlib.__file__
>
> Ben Root
> 
> 

You're right, 1.1 still has nxutils. What are the python counterparts you
mention, are they in mpl?

Thanks,

Jorge

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Benjamin R. <ben...@ou...>]

On Wed, Mar 7, 2012 at 11:31 AM, Jorge Scandaliaris
<jor...@ya...>wrote:

> Benjamin Root <ben.root@...> writes:
>
> >
> < snip >
> >
> > import matplotlib
> > print matplotlib.__version__
> > print matplotlib.__file__
> >
> > Ben Root
> >
> >
>
> You're right, 1.1 still has nxutils. What are the python counterparts you
> mention, are they in mpl?
>
> Thanks,
>
> Jorge
>
>
Essentially, you make a Path object using the vertices, and then use its
contains_point() method.

Ben Root

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Michael D. <md...@st...>]

On 03/07/2012 11:59 AM, Benjamin Root wrote:
>
>
> On Wed, Mar 7, 2012 at 10:46 AM, Jorge Scandaliaris 
> <jor...@ya... <mailto:jor...@ya...>> wrote:
>
>     Hi,
>     I've had an import error with some of my code, specifically with
>     nxutils (I need
>     it for using points_inside_poly).
>
>     In [1]: import matplotlib.nxutils as nx
>     ---------------------------------------------------------------------------
>     ImportError                               Traceback (most recent
>     call last)
>     /home/jscandal/<ipython-input-1-c56b9729b42e> in <module>()
>     ----> 1 import matplotlib.nxutils as nx
>
>     ImportError: No module named nxutils
>
>     I built mpl from trunk and tried also version 1.1 from my distro,
>     but neither
>     seems to have nxutils. I still find the examples in the docs using
>     it, though,
>     so I am confused.
>     Has nxutils been replaced or removed?
>
>     Thanks
>
>     Jorge
>
>
> Last I heard, it was supposed to be deprecated in master (although it 
> was accidentially removed at one point).  Indeed, it appears to be 
> missing from the current master.  The deprecation process must be 
> followed! Personally, I think such removal is short-sighted (nxutil's 
> functions are much faster than their python counter-parts).
I agree that the deprecation process should have been followed better.  
However, I'm not sure what you mean by them being faster than their 
Pyhton counterparts.  Both functions in nxutils are replaced by 
functions in _path.cpp, which are also written in C++ but are more 
complete and don't have broken corner cases.  They may be slightly 
slower  because they have a few more checks and handle Bezier curves, 
but they are not significantly slower.

Mike

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[John H. <jd...@gm...>]

On Wed, Mar 7, 2012 at 12:31 PM, Michael Droettboom <md...@st...> wrote:

> I agree that the deprecation process should have been followed better.
> However, I'm not sure what you mean by them being faster than their Pyhton
> counterparts.  Both functions in nxutils are replaced by functions in
> _path.cpp, which are also written in C++ but are more complete and don't
> have broken corner cases.  They may be slightly slower  because they have a
> few more checks and handle Bezier curves, but they are not significantly
> slower.
>


What about providing a python module nxutils in master (1.2) that has the
same signature as the old nxutils extension code, calls the path
functionality, and raises a deprecation warning with suggested a suggested
code migration?  And then remove it for 1.3

JDH

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Michael D. <md...@st...>]

On 03/07/2012 01:39 PM, John Hunter wrote:
>
>
> On Wed, Mar 7, 2012 at 12:31 PM, Michael Droettboom <md...@st... 
> <mailto:md...@st...>> wrote:
>
>     I agree that the deprecation process should have been followed
>     better.  However, I'm not sure what you mean by them being faster
>     than their Pyhton counterparts.  Both functions in nxutils are
>     replaced by functions in _path.cpp, which are also written in C++
>     but are more complete and don't have broken corner cases.  They
>     may be slightly slower  because they have a few more checks and
>     handle Bezier curves, but they are not significantly slower.
>
>
>
> What about providing a python module nxutils in master (1.2) that has 
> the same signature as the old nxutils extension code, calls the path 
> functionality, and raises a deprecation warning with suggested a 
> suggested code migration?  And then remove it for 1.3
>
Yeah.  I think that's a good way out of this.  I'll file a bug and 
self-assign.

Mike

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Benjamin R. <ben...@ou...>]

On Wed, Mar 7, 2012 at 12:44 PM, Michael Droettboom <md...@st...> wrote:

>  On 03/07/2012 01:39 PM, John Hunter wrote:
>
>
>
> On Wed, Mar 7, 2012 at 12:31 PM, Michael Droettboom <md...@st...>wrote:
>
>>  I agree that the deprecation process should have been followed better.
>> However, I'm not sure what you mean by them being faster than their Pyhton
>> counterparts.  Both functions in nxutils are replaced by functions in
>> _path.cpp, which are also written in C++ but are more complete and don't
>> have broken corner cases.  They may be slightly slower  because they have a
>> few more checks and handle Bezier curves, but they are not significantly
>> slower.
>>
>
>
>  What about providing a python module nxutils in master (1.2) that has
> the same signature as the old nxutils extension code, calls the path
> functionality, and raises a deprecation warning with suggested a suggested
> code migration?  And then remove it for 1.3
>
>  Yeah.  I think that's a good way out of this.  I'll file a bug and
> self-assign.
>
> Mike
>
>
Thanks for that.  As for timing, I have no hard numbers.  In fact, after
some checking, I just realized that my app that has been using nxutils
still even though I am on master!  Maybe my slowdown was a coincidence with
other changes I have made?

Everyone should remember to clear out nxutils.so because the build/clean
process won't necessarially do it unless you blow away your install dir.

Ben Root

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Jorge S. <jor...@ya...>]

Benjamin Root <ben.root@...> writes:

<snip>
> 
> Essentially, you make a Path object using the vertices, and then use its 
> contains_point() method.
> Ben Root
> 

OK, but given that contains_point works with a *single* point at a time, I have
to call it for all my points which is a bit more cumbersome, or am I missing
something?

Jorge

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Dharhas P. <Dha...@tw...>]

+ one on this issue. One of the big advantages of the nxutils points in poly is that you could pass it a large numpy array of points and get back a mask. We found this to be significantly faster than using looping through the single point in poly algorithms from packages like shapely. Echoing Jorge's question how would we do this using contains_point(). 


- dharhas 


>>> Jorge Scandaliaris <jor...@ya...> 3/8/2012 3:33 AM >>>
Benjamin Root <ben.root@...> writes:

<snip>
>
> Essentially, you make a Path object using the vertices, and then use its
> contains_point() method.
> Ben Root
>

OK, but given that contains_point works with a *single* point at a time, I have
to call it for all my points which is a bit more cumbersome, or am I missing
something?

Jorge




------------------------------------------------------------------------------
Virtualization & Cloud Management Using Capacity Planning
Cloud computing makes use of virtualization - but cloud computing
also focuses on allowing computing to be delivered as a service.
http://www.accelacomm.com/jaw/sfnl/114/51521223/
_______________________________________________
Matplotlib-users mailing list
Mat...@li...
https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[John H. <jd...@gm...>]

On Thu, Mar 8, 2012 at 10:21 AM, Dharhas Pothina <
Dha...@tw...> wrote:

>
>  + one on this issue. One of the big advantages of the nxutils points in
> poly is that you could pass it a large numpy array of points and get back a
> mask. We found this to be significantly faster than using looping through
> the single point in poly algorithms from packages like shapely. Echoing
> Jorge's question how would we do this using contains_point().
>

One could modify src/_path.cpp to expose a "points_in_path" method so the
looping would be done in C++.  The underlying work is being done
in point_in_path_impl and exposed to python as  point_in_path.  It would
not be too much work to expose a points_in_path method, but we would have
to think about what arguments to take and to return.  _path.cpp does not
currently depend on numpy, and doing things via python list would be slower
than what we currently have in nxutils which is array aware.

JDH

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Benjamin R. <ben...@ou...>]

On Thu, Mar 8, 2012 at 10:21 AM, Dharhas Pothina <
Dha...@tw...> wrote:

>
>  + one on this issue. One of the big advantages of the nxutils points in
> poly is that you could pass it a large numpy array of points and get back a
> mask. We found this to be significantly faster than using looping through
> the single point in poly algorithms from packages like shapely. Echoing
> Jorge's question how would we do this using contains_point().
>
>
>  - dharhas
>
>
>
> >>> Jorge Scandaliaris <jor...@ya...> 3/8/2012 3:33 AM >>>
>
> Benjamin Root <ben.root@...> writes:
>
> <snip>
> >
> > Essentially, you make a Path object using the vertices, and then use its
> > contains_point() method.
> > Ben Root
> >
>
> OK, but given that contains_point works with a *single* point at a time, I
> have
> to call it for all my points which is a bit more cumbersome, or am I
> missing
> something?
>
> Jorge
>
>
>
+1 as well.  I just took another look at the Path object and I see no such
function.  The lack of this function is a problem for me as well in my
existing apps.  In order to deprecate nxutils, this functionality needs to
be added to Path.  Otherwise, nxutils *must* be reinstated before the next
release.

Ben Root

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[John H. <jd...@gm...>]

On Thu, Mar 8, 2012 at 10:32 AM, Benjamin Root <ben...@ou...> wrote:

>
> +1 as well.  I just took another look at the Path object and I see no such
> function.  The lack of this function is a problem for me as well in my
> existing apps.  In order to deprecate nxutils, this functionality needs to
> be added to Path.  Otherwise, nxutils *must* be reinstated before the next
> release.
>
>
Michael has already agreed to make a nxutils compatibility layer that would
have the same interface as the old nxutils.  So we are talking about
performance, not core functionality.

We should remember that Michael did the lion's share of the work on
 porting mpl to python 3 (
https://github.com/matplotlib/matplotlib/pull/565/commits).  He elected not
to port all of the C++ if he could replace some of the functionality with
the core.  So those who rely on bare metal speed the you are getting in
nxutils should step up to either :

1) help with the port of nxutils to python 3

2) help with exposing methods in _path.cpp that are almost as fast or faster

3) live with slower speeds in the compatibility layer he has agreed to write

4) ask (nicely) for someone to help you

I prefer option 2 because this is fairly easy and avoids code redundancy.
 It would take just a few lines of extra code to do this with the python
sequence protocol as inputs and python lists as return values.  It would
take a bit more to support numpy arrays as input and output, and we should
get input from Michael about the desirability of making _path.cpp depend on
numpy.  I don't see the harm, but I'd like to verify.

In my opinion, a slower implementation in a nxutils.py compatibility module
is not a release stopper, even if it is undesirable.

JDH

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Ryan M. <rm...@gm...>]

On Thu, Mar 8, 2012 at 10:47 AM, John Hunter <jd...@gm...> wrote:
> I prefer option 2 because this is fairly easy and avoids code redundancy.
>  It would take just a few lines of extra code to do this with the python
> sequence protocol as inputs and python lists as return values.  It would
> take a bit more to support numpy arrays as input and output, and we should
> get input from Michael about the desirability of making _path.cpp depend on
> numpy.  I don't see the harm, but I'd like to verify.
>
> In my opinion, a slower implementation in a nxutils.py compatibility module
> is not a release stopper, even if it is undesirable.

 +1
Python 3 support is more important and, quite frankly,
overdue--especially given how long it's been sitting in the tree.

Ryan

-- 
Ryan May
Graduate Research Assistant
School of Meteorology
University of Oklahoma

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Benjamin R. <ben...@ou...>]

On Thu, Mar 8, 2012 at 10:47 AM, John Hunter <jd...@gm...> wrote:

>
>
> On Thu, Mar 8, 2012 at 10:32 AM, Benjamin Root <ben...@ou...> wrote:
>
>>
>> +1 as well.  I just took another look at the Path object and I see no
>> such function.  The lack of this function is a problem for me as well in my
>> existing apps.  In order to deprecate nxutils, this functionality needs to
>> be added to Path.  Otherwise, nxutils *must* be reinstated before the next
>> release.
>>
>>
> Michael has already agreed to make a nxutils compatibility layer that
> would have the same interface as the old nxutils.  So we are talking about
> performance, not core functionality.
>
> We should remember that Michael did the lion's share of the work on
>  porting mpl to python 3 (
> https://github.com/matplotlib/matplotlib/pull/565/commits).  He elected
> not to port all of the C++ if he could replace some of the functionality
> with the core.  So those who rely on bare metal speed the you are getting
> in nxutils should step up to either :
>
> 1) help with the port of nxutils to python 3
>
> 2) help with exposing methods in _path.cpp that are almost as fast or
> faster
>
> 3) live with slower speeds in the compatibility layer he has agreed to
> write
>
> 4) ask (nicely) for someone to help you
>
> I prefer option 2 because this is fairly easy and avoids code redundancy.
>  It would take just a few lines of extra code to do this with the python
> sequence protocol as inputs and python lists as return values.  It would
> take a bit more to support numpy arrays as input and output, and we should
> get input from Michael about the desirability of making _path.cpp depend on
> numpy.  I don't see the harm, but I'd like to verify.
>
> In my opinion, a slower implementation in a
> nxutils.py compatibility module is not a release stopper, even if it is
> undesirable.
>
> JDH
>

Don't get me wrong.  If help is needed, I can certainly provide it since it
is my itch.  I am just a little exasperated with how this issue has been
handled up to now.  The shim is a very good idea and it should have been
done back when the py3k merge happened.  I didn't press the issue then
because I was traveling and didn't have time to examine the issue closely,
and having _nxutils.so still in my build hide the problem from me (my own
fault).

As for shim implementation, I would be willing to accept a slightly slower
function now (with the promise of improvements later), but if the
implementation is too much slower, then effort will need to be made to get
it up to acceptable levels.  I know of several users **cough**my future
employer**cough** who uses functionality such as this, and they would not
be happy if their products are dragged down by such a bottleneck.

Probably about time I dug more into CXX wrapped stuff...

Ben Root

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Michael D. <md...@st...>]

_path.cpp already has a  number of methods that use Numpy arrays (John 
is mistaken that it doesn't depend on Numpy), so adding another is no 
problem.

It was my understanding that nxutils was an internal usage module and 
not part of the public API, and therefore could be removed as long as 
all of the internal references were updated.  It doesn't have a '_' 
prefix, so of course it being private wasn't explicit, but there are 
number of things in matplotlib that fall into that category.  
Apparently, I was mistaken and there are a lot of users using it from 
outside.  It shouldn't be a problem to build a compatibility shim -- I'd 
much rather do that than have multiple functions -- and adding a 
points_in_poly to _path.cpp should be pretty straightforward.

Mike

On 03/08/2012 12:16 PM, Benjamin Root wrote:
>
>
> On Thu, Mar 8, 2012 at 10:47 AM, John Hunter <jd...@gm... 
> <mailto:jd...@gm...>> wrote:
>
>
>
>     On Thu, Mar 8, 2012 at 10:32 AM, Benjamin Root <ben...@ou...
>     <mailto:ben...@ou...>> wrote:
>
>
>         +1 as well.  I just took another look at the Path object and I
>         see no such function.  The lack of this function is a problem
>         for me as well in my existing apps.  In order to deprecate
>         nxutils, this functionality needs to be added to Path. 
>         Otherwise, nxutils *must* be reinstated before the next release.
>
>
>     Michael has already agreed to make a nxutils compatibility layer
>     that would have the same interface as the old nxutils.  So we are
>     talking about performance, not core functionality.
>
>     We should remember that Michael did the lion's share of the work
>     on  porting mpl to python 3
>     (https://github.com/matplotlib/matplotlib/pull/565/commits).  He
>     elected not to port all of the C++ if he could replace some of the
>     functionality with the core.  So those who rely on bare metal
>     speed the you are getting in nxutils should step up to either :
>
>     1) help with the port of nxutils to python 3
>
>     2) help with exposing methods in _path.cpp that are almost as fast
>     or faster
>
>     3) live with slower speeds in the compatibility layer he has
>     agreed to write
>
>     4) ask (nicely) for someone to help you
>
>     I prefer option 2 because this is fairly easy and avoids code
>     redundancy.  It would take just a few lines of extra code to do
>     this with the python sequence protocol as inputs and python lists
>     as return values.  It would take a bit more to support numpy
>     arrays as input and output, and we should get input from Michael
>     about the desirability of making _path.cpp depend on numpy.  I
>     don't see the harm, but I'd like to verify.
>
>     In my opinion, a slower implementation in a
>     nxutils.py compatibility module is not a release stopper, even if
>     it is undesirable.
>
>     JDH
>
>
> Don't get me wrong.  If help is needed, I can certainly provide it 
> since it is my itch.  I am just a little exasperated with how this 
> issue has been handled up to now.  The shim is a very good idea and it 
> should have been done back when the py3k merge happened.  I didn't 
> press the issue then because I was traveling and didn't have time to 
> examine the issue closely, and having _nxutils.so still in my build 
> hide the problem from me (my own fault).
>
> As for shim implementation, I would be willing to accept a slightly 
> slower function now (with the promise of improvements later), but if 
> the implementation is too much slower, then effort will need to be 
> made to get it up to acceptable levels.  I know of several users 
> **cough**my future employer**cough** who uses functionality such as 
> this, and they would not be happy if their products are dragged down 
> by such a bottleneck.
>
> Probably about time I dug more into CXX wrapped stuff...
>
> Ben Root
>
>
> ------------------------------------------------------------------------------
> Virtualization&  Cloud Management Using Capacity Planning
> Cloud computing makes use of virtualization - but cloud computing
> also focuses on allowing computing to be delivered as a service.
> http://www.accelacomm.com/jaw/sfnl/114/51521223/
>
>
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Michael D. <md...@st...>]

On 03/08/2012 12:35 PM, Michael Droettboom wrote:
> It shouldn't be a problem to build a compatibility shim -- I'd much 
> rather do that than have multiple functions

Oops -- hit send too soon.  I meant to say "I'd much rather do that than 
have multiple functions that do virtually the same thing"

Mike

>
> Mike
>
> On 03/08/2012 12:16 PM, Benjamin Root wrote:
>>
>>
>> On Thu, Mar 8, 2012 at 10:47 AM, John Hunter <jd...@gm... 
>> <mailto:jd...@gm...>> wrote:
>>
>>
>>
>>     On Thu, Mar 8, 2012 at 10:32 AM, Benjamin Root <ben...@ou...
>>     <mailto:ben...@ou...>> wrote:
>>
>>
>>         +1 as well.  I just took another look at the Path object and
>>         I see no such function.  The lack of this function is a
>>         problem for me as well in my existing apps.  In order to
>>         deprecate nxutils, this functionality needs to be added to
>>         Path.  Otherwise, nxutils *must* be reinstated before the
>>         next release.
>>
>>
>>     Michael has already agreed to make a nxutils compatibility layer
>>     that would have the same interface as the old nxutils.  So we are
>>     talking about performance, not core functionality.
>>
>>     We should remember that Michael did the lion's share of the work
>>     on  porting mpl to python 3
>>     (https://github.com/matplotlib/matplotlib/pull/565/commits).  He
>>     elected not to port all of the C++ if he could replace some of
>>     the functionality with the core.  So those who rely on bare metal
>>     speed the you are getting in nxutils should step up to either :
>>
>>     1) help with the port of nxutils to python 3
>>
>>     2) help with exposing methods in _path.cpp that are almost as
>>     fast or faster
>>
>>     3) live with slower speeds in the compatibility layer he has
>>     agreed to write
>>
>>     4) ask (nicely) for someone to help you
>>
>>     I prefer option 2 because this is fairly easy and avoids code
>>     redundancy.  It would take just a few lines of extra code to do
>>     this with the python sequence protocol as inputs and python lists
>>     as return values.  It would take a bit more to support numpy
>>     arrays as input and output, and we should get input from Michael
>>     about the desirability of making _path.cpp depend on numpy.  I
>>     don't see the harm, but I'd like to verify.
>>
>>     In my opinion, a slower implementation in a
>>     nxutils.py compatibility module is not a release stopper, even if
>>     it is undesirable.
>>
>>     JDH
>>
>>
>> Don't get me wrong.  If help is needed, I can certainly provide it 
>> since it is my itch.  I am just a little exasperated with how this 
>> issue has been handled up to now.  The shim is a very good idea and 
>> it should have been done back when the py3k merge happened.  I didn't 
>> press the issue then because I was traveling and didn't have time to 
>> examine the issue closely, and having _nxutils.so still in my build 
>> hide the problem from me (my own fault).
>>
>> As for shim implementation, I would be willing to accept a slightly 
>> slower function now (with the promise of improvements later), but if 
>> the implementation is too much slower, then effort will need to be 
>> made to get it up to acceptable levels.  I know of several users 
>> **cough**my future employer**cough** who uses functionality such as 
>> this, and they would not be happy if their products are dragged down 
>> by such a bottleneck.
>>
>> Probably about time I dug more into CXX wrapped stuff...
>>
>> Ben Root
>>
>>
>> ------------------------------------------------------------------------------
>> Virtualization&  Cloud Management Using Capacity Planning
>> Cloud computing makes use of virtualization - but cloud computing
>> also focuses on allowing computing to be delivered as a service.
>> http://www.accelacomm.com/jaw/sfnl/114/51521223/
>>
>>
>> _______________________________________________
>> Matplotlib-users mailing list
>> Mat...@li...
>> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>
>
> ------------------------------------------------------------------------------
> Virtualization&  Cloud Management Using Capacity Planning
> Cloud computing makes use of virtualization - but cloud computing
> also focuses on allowing computing to be delivered as a service.
> http://www.accelacomm.com/jaw/sfnl/114/51521223/
>
>
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Eric F. <ef...@ha...>]

On 03/08/2012 07:16 AM, Benjamin Root wrote:
>
>
> On Thu, Mar 8, 2012 at 10:47 AM, John Hunter <jd...@gm...
> <mailto:jd...@gm...>> wrote:
>
>
>
>     On Thu, Mar 8, 2012 at 10:32 AM, Benjamin Root <ben...@ou...
>     <mailto:ben...@ou...>> wrote:
>
>
>         +1 as well.  I just took another look at the Path object and I
>         see no such function.  The lack of this function is a problem
>         for me as well in my existing apps.  In order to deprecate
>         nxutils, this functionality needs to be added to Path.
>         Otherwise, nxutils *must* be reinstated before the next release.
>
>
>     Michael has already agreed to make a nxutils compatibility layer
>     that would have the same interface as the old nxutils.  So we are
>     talking about performance, not core functionality.
>
>     We should remember that Michael did the lion's share of the work on
>       porting mpl to python 3
>     (https://github.com/matplotlib/matplotlib/pull/565/commits).  He
>     elected not to port all of the C++ if he could replace some of the
>     functionality with the core.  So those who rely on bare metal speed
>     the you are getting in nxutils should step up to either :
>
>     1) help with the port of nxutils to python 3
>
>     2) help with exposing methods in _path.cpp that are almost as fast
>     or faster
>
>     3) live with slower speeds in the compatibility layer he has agreed
>     to write
>
>     4) ask (nicely) for someone to help you
>
>     I prefer option 2 because this is fairly easy and avoids code
>     redundancy.  It would take just a few lines of extra code to do this
>     with the python sequence protocol as inputs and python lists as
>     return values.  It would take a bit more to support numpy arrays as
>     input and output, and we should get input from Michael about the
>     desirability of making _path.cpp depend on numpy.  I don't see the
>     harm, but I'd like to verify.
>
>     In my opinion, a slower implementation in a
>     nxutils.py compatibility module is not a release stopper, even if it
>     is undesirable.
>
>     JDH
>
>
> Don't get me wrong.  If help is needed, I can certainly provide it since
> it is my itch.  I am just a little exasperated with how this issue has
> been handled up to now.  The shim is a very good idea and it should have
> been done back when the py3k merge happened.  I didn't press the issue
> then because I was traveling and didn't have time to examine the issue
> closely, and having _nxutils.so still in my build hide the problem from
> me (my own fault).
>
> As for shim implementation, I would be willing to accept a slightly
> slower function now (with the promise of improvements later), but if the
> implementation is too much slower, then effort will need to be made to
> get it up to acceptable levels.  I know of several users **cough**my
> future employer**cough** who uses functionality such as this, and they
> would not be happy if their products are dragged down by such a bottleneck.

For my own purposes I decided to simply copy the pnpoly core and write a 
cython wrapper, since nxutils did not seem to have a secure future in 
mpl.  If anyone is interested in this approach, see the files _pnpoly.c
and utility.pyx (last function therein) in the num/src subdirectory of 
my "pycurrents" repo: 
http://currents.soest.hawaii.edu/hgstage/pycurrents/shortlog/tip.  The 
wrapper is brand-new and very lightly tested; I suspect it could be 
improved considerably.  (Also, at present it has only the points_in_poly 
function.)

>
> Probably about time I dug more into CXX wrapped stuff...

mpl strategy question: stick with CXX and hand-wrapping exclusively, or 
start using cython?

Eric

>
> Ben Root
>
>
>
> ------------------------------------------------------------------------------
> Virtualization&  Cloud Management Using Capacity Planning
> Cloud computing makes use of virtualization - but cloud computing
> also focuses on allowing computing to be delivered as a service.
> http://www.accelacomm.com/jaw/sfnl/114/51521223/
>
>
>
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Benjamin R. <ben...@ou...>]

On Thu, Mar 8, 2012 at 11:16 AM, Benjamin Root <ben...@ou...> wrote:

>
>
> On Thu, Mar 8, 2012 at 10:47 AM, John Hunter <jd...@gm...> wrote:
>
>>
>>
>> On Thu, Mar 8, 2012 at 10:32 AM, Benjamin Root <ben...@ou...> wrote:
>>
>>>
>>> +1 as well.  I just took another look at the Path object and I see no
>>> such function.  The lack of this function is a problem for me as well in my
>>> existing apps.  In order to deprecate nxutils, this functionality needs to
>>> be added to Path.  Otherwise, nxutils *must* be reinstated before the next
>>> release.
>>>
>>>
>> Michael has already agreed to make a nxutils compatibility layer that
>> would have the same interface as the old nxutils.  So we are talking about
>> performance, not core functionality.
>>
>> We should remember that Michael did the lion's share of the work on
>>  porting mpl to python 3 (
>> https://github.com/matplotlib/matplotlib/pull/565/commits).  He elected
>> not to port all of the C++ if he could replace some of the functionality
>> with the core.  So those who rely on bare metal speed the you are getting
>> in nxutils should step up to either :
>>
>> 1) help with the port of nxutils to python 3
>>
>> 2) help with exposing methods in _path.cpp that are almost as fast or
>> faster
>>
>> 3) live with slower speeds in the compatibility layer he has agreed to
>> write
>>
>> 4) ask (nicely) for someone to help you
>>
>> I prefer option 2 because this is fairly easy and avoids code redundancy.
>>  It would take just a few lines of extra code to do this with the python
>> sequence protocol as inputs and python lists as return values.  It would
>> take a bit more to support numpy arrays as input and output, and we should
>> get input from Michael about the desirability of making _path.cpp depend on
>> numpy.  I don't see the harm, but I'd like to verify.
>>
>> In my opinion, a slower implementation in a
>> nxutils.py compatibility module is not a release stopper, even if it is
>> undesirable.
>>
>> JDH
>>
>
> Don't get me wrong.  If help is needed, I can certainly provide it since
> it is my itch.  I am just a little exasperated with how this issue has been
> handled up to now.  The shim is a very good idea and it should have been
> done back when the py3k merge happened.  I didn't press the issue then
> because I was traveling and didn't have time to examine the issue closely,
> and having _nxutils.so still in my build hide the problem from me (my own
> fault).
>
> As for shim implementation, I would be willing to accept a slightly slower
> function now (with the promise of improvements later), but if the
> implementation is too much slower, then effort will need to be made to get
> it up to acceptable levels.  I know of several users **cough**my future
> employer**cough** who uses functionality such as this, and they would not
> be happy if their products are dragged down by such a bottleneck.
>
> Probably about time I dug more into CXX wrapped stuff...
>
> Ben Root
>


Looking over the code, it looks like we could generalize
point_in_path_impl() into points_in_path_impl().  The current code iterates
through the path vertices to test a single point.  Putting this action
inside a loop (for each point that we want to test) would mean that this
iterator has to be processed each time, which I suspect would kill
performance when the number of vertices is far greater than the number of
test points.

Tinkering....

Ben Root

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Michael D. <md...@st...>]

There is a proposed solution to all of this here:

https://github.com/matplotlib/matplotlib/pull/746

Please test -- I don't have any nxutils-using code myself, and 
matplotlib itself has none.  We should probably convert some of the 
nxutils code in the wild into some unit tests.

Mike

On 03/08/2012 12:37 PM, Benjamin Root wrote:
>
>
> On Thu, Mar 8, 2012 at 11:16 AM, Benjamin Root <ben...@ou... 
> <mailto:ben...@ou...>> wrote:
>
>
>
>     On Thu, Mar 8, 2012 at 10:47 AM, John Hunter <jd...@gm...
>     <mailto:jd...@gm...>> wrote:
>
>
>
>         On Thu, Mar 8, 2012 at 10:32 AM, Benjamin Root
>         <ben...@ou... <mailto:ben...@ou...>> wrote:
>
>
>             +1 as well.  I just took another look at the Path object
>             and I see no such function.  The lack of this function is
>             a problem for me as well in my existing apps.  In order to
>             deprecate nxutils, this functionality needs to be added to
>             Path.  Otherwise, nxutils *must* be reinstated before the
>             next release.
>
>
>         Michael has already agreed to make a nxutils compatibility
>         layer that would have the same interface as the old nxutils.
>          So we are talking about performance, not core functionality.
>
>         We should remember that Michael did the lion's share of the
>         work on  porting mpl to python 3
>         (https://github.com/matplotlib/matplotlib/pull/565/commits).
>          He elected not to port all of the C++ if he could replace
>         some of the functionality with the core.  So those who rely on
>         bare metal speed the you are getting in nxutils should step up
>         to either :
>
>         1) help with the port of nxutils to python 3
>
>         2) help with exposing methods in _path.cpp that are almost as
>         fast or faster
>
>         3) live with slower speeds in the compatibility layer he has
>         agreed to write
>
>         4) ask (nicely) for someone to help you
>
>         I prefer option 2 because this is fairly easy and avoids code
>         redundancy.  It would take just a few lines of extra code to
>         do this with the python sequence protocol as inputs and python
>         lists as return values.  It would take a bit more to support
>         numpy arrays as input and output, and we should get input from
>         Michael about the desirability of making _path.cpp depend on
>         numpy.  I don't see the harm, but I'd like to verify.
>
>         In my opinion, a slower implementation in a
>         nxutils.py compatibility module is not a release stopper, even
>         if it is undesirable.
>
>         JDH
>
>
>     Don't get me wrong.  If help is needed, I can certainly provide it
>     since it is my itch.  I am just a little exasperated with how this
>     issue has been handled up to now.  The shim is a very good idea
>     and it should have been done back when the py3k merge happened.  I
>     didn't press the issue then because I was traveling and didn't
>     have time to examine the issue closely, and having _nxutils.so
>     still in my build hide the problem from me (my own fault).
>
>     As for shim implementation, I would be willing to accept a
>     slightly slower function now (with the promise of improvements
>     later), but if the implementation is too much slower, then effort
>     will need to be made to get it up to acceptable levels.  I know of
>     several users **cough**my future employer**cough** who uses
>     functionality such as this, and they would not be happy if their
>     products are dragged down by such a bottleneck.
>
>     Probably about time I dug more into CXX wrapped stuff...
>
>     Ben Root
>
>
>
> Looking over the code, it looks like we could generalize 
> point_in_path_impl() into points_in_path_impl().  The current code 
> iterates through the path vertices to test a single point.  Putting 
> this action inside a loop (for each point that we want to test) would 
> mean that this iterator has to be processed each time, which I suspect 
> would kill performance when the number of vertices is far greater than 
> the number of test points.
>
> Tinkering....
>
> Ben Root
>
>
>
> ------------------------------------------------------------------------------
> Virtualization&  Cloud Management Using Capacity Planning
> Cloud computing makes use of virtualization - but cloud computing
> also focuses on allowing computing to be delivered as a service.
> http://www.accelacomm.com/jaw/sfnl/114/51521223/
>
>
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Re: [Matplotlib-users] Has nxutils been removed from mpl?

[Dharhas P. <Dha...@tw...>]

As someone who has never contributed to matplotlib before, are there any instructions on how to contribute to writing tests. We have some data and scripts we could probably convert to tests for the nxutils points in poly functions. Should I just do i pull the branch nxutilsbackwards branch and make a pull request after adding tests on github?  

- dharhas

>>> Michael Droettboom <md...@st...> 3/8/2012 5:51 PM >>>
There is a proposed solution to all of this here:

https://github.com/matplotlib/matplotlib/pull/746

Please test -- I don't have any nxutils-using code myself, and matplotlib itself has none.  We should probably convert some of the nxutils code in the wild into some unit tests.

Mike

On 03/08/2012 12:37 PM, Benjamin Root wrote: 





On Thu, Mar 8, 2012 at 11:16 AM, Benjamin Root <ben...@ou...> wrote:






On Thu, Mar 8, 2012 at 10:47 AM, John Hunter <jd...@gm...> wrote:






On Thu, Mar 8, 2012 at 10:32 AM, Benjamin Root <ben...@ou...> wrote:




+1 as well.  I just took another look at the Path object and I see no such function.  The lack of this function is a problem for me as well in my existing apps.  In order to deprecate nxutils, this functionality needs to be added to Path.  Otherwise, nxutils *must* be reinstated before the next release.

 



Michael has already agreed to make a nxutils compatibility layer that would have the same interface as the old nxutils.  So we are talking about performance, not core functionality. 



We should remember that Michael did the lion's share of the work on  porting mpl to python 3 (https://github.com/matplotlib/matplotlib/pull/565/commits).  He elected not to port all of the C++ if he could replace some of the functionality with the core.  So those who rely on bare metal speed the you are getting in nxutils should step up to either : 



1) help with the port of nxutils to python 3 



2) help with exposing methods in _path.cpp that are almost as fast or faster 



3) live with slower speeds in the compatibility layer he has agreed to write 



4) ask (nicely) for someone to help you 



I prefer option 2 because this is fairly easy and avoids code redundancy.  It would take just a few lines of extra code to do this with the python sequence protocol as inputs and python lists as return values.  It would take a bit more to support numpy arrays as input and output, and we should get input from Michael about the desirability of making _path.cpp depend on numpy.  I don't see the harm, but I'd like to verify. 



In my opinion, a slower implementation in a nxutils.py compatibility module is not a release stopper, even if it is undesirable.   



JDH 
 


Don't get me wrong.  If help is needed, I can certainly provide it since it is my itch.  I am just a little exasperated with how this issue has been handled up to now.  The shim is a very good idea and it should have been done back when the py3k merge happened.  I didn't press the issue then because I was traveling and didn't have time to examine the issue closely, and having _nxutils.so still in my build hide the problem from me (my own fault).

As for shim implementation, I would be willing to accept a slightly slower function now (with the promise of improvements later), but if the implementation is too much slower, then effort will need to be made to get it up to acceptable levels.  I know of several users **cough**my future employer**cough** who uses functionality such as this, and they would not be happy if their products are dragged down by such a bottleneck.

Probably about time I dug more into CXX wrapped stuff...

Ben Root
 



Looking over the code, it looks like we could generalize point_in_path_impl() into points_in_path_impl().  The current code iterates through the path vertices to test a single point.  Putting this action inside a loop (for each point that we want to test) would mean that this iterator has to be processed each time, which I suspect would kill performance when the number of vertices is far greater than the number of test points.

Tinkering....

Ben Root




------------------------------------------------------------------------------    
Virtualization & Cloud Management Using Capacity Planning  
Cloud computing makes use of virtualization - but cloud computing 
also focuses on allowing computing to be delivered as a service.
http://www.accelacomm.com/jaw/sfnl/114/51521223/



_______________________________________________
Matplotlib-users mailing list
Mat...@li...
https://lists.sourceforge.net/lists/listinfo/matplotlib-users