NumPy 1.0b4 now available

Tue Aug 29 21:00:03 EDT 2006

Robert Kern wrote:
> mensanator at aol.com wrote:
> > Travis E. Oliphant wrote:
> >> The 4th beta release of NumPy 1.0 has just been made available.
> >>
> >> NumPy 1.0 represents the culmination of over 18 months of work to unify
> >> the Numeric and Numarray array packages into a single best-of-breed
> >> array package for Python.
> >>
> >> NumPy supports all the features of Numeric and Numarray with a healthy
> >> dose of it's own improved features.
> >
> > So how come this support doesn't extend to making a numpy
> > version of NumTuT?
>
> Because NumTut has no redeeming value? Certainly none over this:
>
>    http://www.scipy.org/Numpy_Example_List

Ok, it's just that when I go to the download page, I see

Latest File Releases

The latest file releases

Package    Release  Date               Notes / Monitor  Downloads
Numarray   1.5.2    August 24, 2006          -          Download
Numeric    24.2     November 11, 2005        -          Download
NumPy      1.0b4    July 21, 2006            -          Download
tutorial   NumTut   August 14, 2001          -          Download

which certainly makes no indication that NumTut has no redeeming value.
Even the Notes simply tell you to look at the Numeric documentation.
I hardly think that's good enough. It should bear a large red label:

WARNING! Not compatible with NumPy (and has no redeeming value)

>
> >> It's time to start porting your applications to use NumPy as Numeric is
> >> no longer maintained and Numarray will only be maintained for a few more
> >> months.
> >
> > Which would be a good reason to convert NumTut.
>
> No. NumTut would still need to be *useful*.

Then why are you floggin' the bloody thing?

>
> >> Porting is not difficult especially using the compatibility layers
> >> numpy.oldnumeric and numpy.numarray and the alter_code1.py modules in
> >> those packages.
> >
> > Ah, I see. I'm supposed to convert it myself.
> >
> > So when I place my cursor over the link
>
> What link?

This one (from Numpy Home Page):

<br />
Numeric
users should find the transition very easy.  There is a <a

href="http://projects.scipy.org/scipy/numpy/browser/trunk/numpy/oldnumeric/alter_code1.py">module</a>
that can make most of the necessary changes to your Python code that
used Numeric to work with the new NumPy. <br />
<br />

>
> > and it identifies the
> > link as "alter_code1.py" meaning it's a Python source file,
> > naturally I download it only to discover that it's not a source
> > code file but an .html file masquerading as a .py file.
> >
> > So when I followow the link and see some stupid management
> > system, I'm completely bewildered. What the fuck am I supposed
> > to do with this? Oh, wait...I can download in other formats.
> >
> > Only the "Plain Text" format isn't plain text. It strips off the web
> > page stuff but doesn't translate the codes such as < rendering
> > the alleged plain text un-runnable. Ok, I should have downloaded
> > "Original Format". Would it be too much trouble to explain all that?
>
> Why are you trying to download the file from the Trac source browser? Get it
> from the real numpy source distribution.

Because the Home Page told me to go there? That usually means it
isn't included in the distribution. I certainly wouldn't have gone to
all that trouble if I knew I already had it. The wording on the Home
Page implied I didn't, so I didn't look.

The best Quality Control is performed by someone who's completely
clueless.

Remember, the programmer only *thinks* he knows how it works.

>
> > Oh, and when you run alter_code1.py, it does not, in fact, alter
> > the code. After "altering", the NumTut files continue to complain
> > that there is no module named "Numeric". Pardon me, but wasn't
> > alter_code1.py supposed to fix that? I guess not. Of course, you
> > can guess what happened next. Manually changing "Numeric" to
> > "numpy" fixes the import problem but it still won't run, something
> > about "types" not being defined.
>
> Yup, that's a bug. Apparently we're not detecting "from Numeric import *". The
> drop-in replacement for "Numeric" is not "numpy" but "numpy.oldnumeric".
>
> Making that replacement at least gets me to a RuntimeError in Tkinter. Which is
> exactly what I get with Numeric. Remember what I said about NumTut having no
> redeeming value?

Fine. I'll discard it. Lucky for me I have no actual Numeric code to
convert.

>
> >> The full C-API of Numeric is supported as is the C-API
> >> of Numarray.
> >>
> >> More information is available at  http://numpy.scipy.org
> >
> > Like the statement
> >
> > "There is a module called convertcode.py in NumPy that
> > can make the transition to NumPy easier (it will automatically
> > perform the search-and-replace style changes that need to
> > be made to python code that uses Numeric to make it work
> > with NumPy)."
> >
> > Which is a lie, there is no such module included in numpy.
> > Is alter_code1.py supposed to take its place? If so why hasn't
> > the Home Page been updated to reflect this?
>
> Are you volunteering to maintain the site?

Sure.

> I'd be happy to set you up with access.

And I'll be happy to change it. Say, remove the goofy links, direct
the users to look for the alter*.py programs in their distribution,
point out that NumTut has no redeeming value and cannot be converted
using alter*.py, so the user is on his own to make it work if he
really wants to try it? At this point, that's all I know about it.

>
> >> NumPy Developers
> >
> > And you have the GALL to CHARGE for the documentation!
>
> No, he's charging for a book. It's not the only source of documentation out
> there, you know.

Ok, I flew off the handle there. I realize you get what you pay for.
Not getting what you pay for irks me, however.

>
>    http://www.scipy.org/Documentation
>
> If you think the freely available material is lacking, please contribute.

I'll be happy to, if I can.

It was I, after all, who found the memory leak bug in gmpy that's now
fixed. Unfortunately, there's been no reply to my second bug find and
given the issues with the last update, I despair of ever seeing another
Windows version, meaning I'll probably be stuck on Python 2.4 forever.

>
> --
> Robert Kern
>
> "I have come to believe that the whole world is an enigma, a harmless enigma
>   that is made terrible by our own mad attempt to interpret it as though it had
>   an underlying truth."
>    -- Umberto Eco