On Tue, Feb 19, 2013 at 3:22 PM, Doug Blank <[email protected]> wrote:
> On Tuesday, February 19, 2013 12:45:50 PM UTC-5, Aaron Meurer wrote:
>
>> On Sun, Feb 17, 2013 at 8:57 PM, Doug Blank <[email protected]> wrote:
>> > Looks like a very active community here!
>> >
>> > For the time being, I just threw together a quick-n-dirty,
>> > not-very-general
>> > Numeric.py that just implements the few pieces I needed:
>> >
>> > http://svn.cs.brynmawr.edu/Calico/trunk/languages/Python/modules/ai/Numeric.py
>> >
>> > In my code, all of my uses are either one dimensional vectors
>> > (represented
>> > by a Python array in Numeric.array.array) or a two dimensional matrix
>> > (represented as a list of Python arrays in Numeric.array.array).
>> >
>> > A full pure-python Numeric.array replacement would be a lot of work, I
>> > suspect. But it would be useful for many projects to fall back to.
>> >
>> > On a different point completely, sympy looks like it would be an
>> > excellent
>> > candidate to include in our educational system,
>> > http://calicoproject.org/
>> > which includes IronPython, but that makes Python code callable from
>> > other
>> > languages (like Ruby, Scheme, Basic, Logo, etc).
>>
>> I don't know what the status of SymPy on IronPython is. The latest
>> information I know is
>> http://code.google.com/p/sympy/issues/detail?id=60, but the only way I
>> can test is with mono, and I suspect some of the issues are really
>> just mono bugs and not IronPython issues. If you can try running the
>> SymPy tests in IronPython and report back to that issue, that would
>> help. Hopefully it at least imports (that's more than can be said for
>> Jython).
>>
>
> SymPy is working fine on IronPython, at least under Mono, on Mac, Windows,
> and Linux. I have included SymPy into our educational IDE, and it appears to
> be working fine.
>
> I didn't run the tests yet, but everything I tried worked very well. I set
> up the print defaults to use unicode, and it looks nice in our GUI.
>
> I did have to make at least a couple of edits, and I had to get a couple of
> current standard python libraries. One edit:
That's awesome.
>
> In my configuration, sys.version_info[0] is equal to 3, and that caused some
> issues on start-up. I'm not sure if I caused the sys.version_info to be set
> that way (I have turned on all of the Python3 options that there are
> (with_statement, division, print_function, etc) even though it is not
> Python3 yet). I commented out the fail:
If the Python is supposed to be 2, then sys.version_info[0] should
give 2. You might look deeper into why this is happening. At any
rate, a better fix might be to get it to return 2. One way would be to
monkey-patch it. Something along the lines of
from collections import namedtuple
sys.version_info = namedtuple('version_info', 'major minor micro
releaselevel serial')(2, 7, 3, 'final', 0)
(I didn't completely test that, but it should be something similar).
Or better yet, figure out why it is wrong.
>
> else: # Python 3
> pass
> #if not HAS_2TO3:
> # Here we can also check for specific Python 3 versions, if needed
>
> And I made sure I had the latest signal.py from Python 2.7. There was one
> issue with the custom importer. I can give a full diff.
>
> I'll run the tests and report any issues. I'd be glad to help refine/test
> the start-up code to include IronPython.
Great. If you could post a full log of any test failures and doctest
failures to that issue, that would be great. Then we can work on
getting it to work. If it really works on mono, maybe I will be able
to test it myself (last time I checked it didn't, but it's been a
while).
Aaron Meurer
>
> -Doug
>
>
>>
>> Aaron Meurer
>>
>> >
>> > Thanks!
>> >
>> > -Doug
>> >
>> >
>> > On Saturday, February 16, 2013 2:55:04 PM UTC-5, Matthew wrote:
>> >>
>> >> You mean 'float64' and all that?
>> >>
>> >> I think that that's a completely separable decision from the container
>> >> class. I think the only decision that an NDArray class would have to
>> >> make
>> >> is whether or not to insist on type homogeneity. Other than that it
>> >> should
>> >> just hold things, grant convenient access, provide views, etc.... It
>> >> should
>> >> probably also implement something like tensor.dot. I think that the
>> >> actual
>> >> elementwise computation piece should be parametrized or rely on the
>> >> `__mul__` attribute of the elements.
>> >>
>> >> Perhaps I didn't understand your question.
>> >>
>> >>
>> >> On Sat, Feb 16, 2013 at 11:43 AM, Aaron Meurer <[email protected]>
>> >> wrote:
>> >>>
>> >>> Would we also emulate the dtype API?
>> >>>
>> >>> Aaron Meurer
>> >>>
>> >>> On Feb 16, 2013, at 12:38 PM, Stefan Krastanov
>> >>> <[email protected]> wrote:
>> >>>
>> >>> > I mostly agree that some kind of ndarray object would be nice. A
>> >>> > very
>> >>> > simple, non-symbolic (i.e. not subclass of Basic) one. It would be
>> >>> > even better if numpy agrees on some Abstract Base Class that we can
>> >>> > mimic.
>> >>> >
>> >>> > --
>> >>> > You received this message because you are subscribed to the Google
>> >>> > Groups "sympy" group.
>> >>> > To unsubscribe from this group and stop receiving emails from it,
>> >>> > send
>> >>> > an email to [email protected].
>> >>> > To post to this group, send email to [email protected].
>> >>>
>> >>> > Visit this group at http://groups.google.com/group/sympy?hl=en.
>> >>> > For more options, visit https://groups.google.com/groups/opt_out.
>> >>> >
>> >>> >
>> >>>
>> >>> --
>> >>> You received this message because you are subscribed to the Google
>> >>> Groups
>> >>> "sympy" group.
>> >>> To unsubscribe from this group and stop receiving emails from it, send
>> >>> an
>> >>> email to [email protected].
>> >>> To post to this group, send email to [email protected].
>> >>>
>> >>> Visit this group at http://groups.google.com/group/sympy?hl=en.
>> >>> For more options, visit https://groups.google.com/groups/opt_out.
>> >>>
>> >>>
>> >>
>> > --
>> > You received this message because you are subscribed to the Google
>> > Groups
>> > "sympy" group.
>> > To unsubscribe from this group and stop receiving emails from it, send
>> > an
>> > email to [email protected].
>> > To post to this group, send email to [email protected].
>> > Visit this group at http://groups.google.com/group/sympy?hl=en.
>> > For more options, visit https://groups.google.com/groups/opt_out.
>> >
>> >
>
> --
> You received this message because you are subscribed to the Google Groups
> "sympy" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> To post to this group, send email to [email protected].
> Visit this group at http://groups.google.com/group/sympy?hl=en.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
--
You received this message because you are subscribed to the Google Groups
"sympy" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/sympy?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
