Is there any place where one can get Windows binaries for NumPy, even
if it's not the latest release? SourceForge only has source and Linux
binary versions, and the LLNL site doesn't seem to exist any more.
Konrad Hinsen | E-Mail: hinsen(a)cnrs-orleans.fr
Centre de Biophysique Moleculaire (CNRS) | Tel.: +33-18.104.22.168.69
Rue Charles Sadron | Fax: +33-22.214.171.124.17
45071 Orleans Cedex 2 | Deutsch/Esperanto/English/
France | Nederlands/Francais
As I have received several encouraging emails of interested people,
I'm excited to announce the availability of an experimental Numerical
Python revision on the NumPy SourceForge site.
I've placed it under the directoryname numpy2 (checkout that directory
according to the CVS instructions).
It is NOT COMPLETED yet, so only people interested in a particular view of
a future arrayobject or who want to help fashion that view should check it
It builds on the current NumPy. Operations should not change on the
Python level. Important new operations will be added. C-API is changed
but support for backward-near-compatibility is desired and sought after.
Only an NDArray class and a Ufunc class along with standard operations on
them is set for inclusion in the new development. Other components will
be added modules.
This is a Python class with C-calls placed for speed of certain
operations. It is built around a data object which can be any Python
object exporting a buffer interface (including a buffer object) and a
dimensions/strides object which is also relies on the buffer interface.
A C-API gives C-users access to the pointers for this information for any
subclass of NDArray.
This is a Python class that encapsulates the N-D looping structure with
the broadcasting rules. It will allow wrapping of fast C functions as
well as Python functions. It basically consists of two attributes: a
select function (for type coercion and specification) and a compute
function that does the actual computation.
Comments and feedback (and especially code) are welcomed.
Is the fast umath module dead? If so lets pull the source code!
I ask since the fast_umath source is still there but apparently not
updated. The calls PyUFunc_FromFuncAndData have the wrong number of
Rob Managan <mailto://email@example.com>
LLNL ph: 925-423-0903
P.O. Box 808, L-095 FAX: 925-422-3389
Livermore, CA 94551-0808
Thomas Shepard writes:
>> Thought you may like to see this.
>> Do you know of any other users of the Gist module besides yourself?
> Everyone who uses Yorick uses the gist package within Yorick, and there
> are a lot of such people at LLNL and worldwide. Since some major new LLNL
> production codes will be python based, I predict significant interest
> in the Python Gist package.
I do wonder how many users of Yorick there are. I am sure I'm the
only one in the Berkeley astronomy department, for example.
I will take a look at the Gist source, and see if I can understand it
well enough to support it. I haven't used Gist all that much, but I
finally found a problem that I couldn't solve with the other graphics
packages I know, so I've started using it again.
> Regarding xfiles.llnl.gov, I am pretty sure that the server is now
> behind the new LLNL firewall and therefore invisible to the outside
> world. But it still exists. It would be a good idea to arrange for
> these packages to be available somewhere where outsiders can get them.
I found LLNLDistribution11 on the ftp-icf site, so I'm set for now. I
could look into sourceforge as a distribution site.
Then again, I'd like to graduate by December, so I'm likely to be busy
on my own projects, stamping down gamma-rays and unruly pair plasmas,
without much time for purely programming work. ;-)
Johann Hibschman johann(a)physics.berkeley.edu
>I read with interest your message about Gist. I've been using it with
>Python on an SGI and a Linux box. Do packages like Ezplot, etc, run on the
I think the answer to that is no, unless there is a version of Ezplot
I don't know about, which is quite possible. Gist provides all the
functionality available in Ezplot and, in my opinion, is easier to
use than Ezplot (although Ezplot is not difficult either).
There are publicly available Yorick versions that run on windows and
macs which use Gist with the corresponding graphics engines. I don't
know if anyone has made python modules using the windows or mac versions
of the Gist graphics engines. If not, I don't think it would be
very hard to do.
I wonder if the upcoming mac OS X will support Xlib. Does anybody know?
PS: Another great feature of Gist I forgot to mention:
It produces great-looking publication-quality output directly.
The choice of things like tick marks, line thicknesses, fonts,
etc. (which is quasi-infinitely customizable) that Gist makes
is very good.
In contrast, things that other packages I have used do:
Use the thinnest possible lines for everything, presumably because
they think someone is going to scale data off the plot using a ruler
and need to keep it precise. This results in thin lines that don't
Insist on writing unrequested text to the plots (think "core dump").
Write some text on top of other text.
Thomas D. Shepard
LLNL A Division
925 423 4018
>Thought you may like to see this.
>Do you know of any other users of the Gist module besides yourself?
Everyone who uses Yorick uses the gist package within Yorick, and there
are a lot of such people at LLNL and worldwide. Since some major new LLNL
production codes will be python based, I predict significant interest
in the Python Gist package.
Currently, I have heard that Marty Marinak uses the Python Gist package
within an LLNL python-based code. I haven't tried to do a survey, so there
may well be others. I intend to use it for my work, which would then
expose it to my co-workers and collaborators, many of whom already
use Yorick with its Gist interface.
Gist is a very efficient graphics package, due in part to being written
directly to Xlib rather than layered on top of some other interface, and
due also to the skill of Gist's author (Dave Munro). (It is also
very easy to write additional efficient graphics engines for Gist, as has
been done by Steve Langer for windows and macOS.)
Gist also has a very well designed interface (both "user friendly" and
efficient, an unusual combination) with features that were tailored
specifically to the needs of LLNL physicists.
In my opinion, it would be a tragedy if Gist were lost.
Regarding xfiles.llnl.gov, I am pretty sure that the server is now
behind the new LLNL firewall and therefore invisible to the outside
world. But it still exists. It would be a good idea to arrange for
these packages to be available somewhere where outsiders can get them.
>---------- Forwarded message ----------
>Date: Mon, 8 May 2000 15:33:19 -0700
>From: "L. Busby" <busby(a)icf.llnl.gov>
>Cc: motteler(a)icf.llnl.gov, numpy-discussion(a)lists.sourceforge.net
>Subject: Re: [Numpy-discussion] LLNL distribution?
>[ Johann Hibschman <johann(a)physics.berkeley.edu> asks ]
>>Is there still a publically available LLNL distribution of NumPy?
>>Or, more precisely, now that NumPy has been spun off as a separate
>>project, where should I look for updates to the Gist module? Is
>>it still available?
>>xfiles.llnl.gov doesn't seem to be there anymore...
>The older versions of the LLNL distribution, including the Gist module,
>continue to be available at ftp-icf.llnl.gov:/pub/python.
>There haven't been any updates to Gist in well over a year. We (Zane
>Motteler and I) have zero local customers for it, and a multitude of
>other programmatic responsibilities. We realize that it would be nice
>simply to dis-entangle Gist from the old distribution and make it
>available as a separate Python package, perhaps using Distutils, or
>not, or as an RPM, or as an SRPM, or as a Debian package. There is no
>time and no support for this activity. If you the reader would like to
>undertake the project, you have our blessing.
>Numpy-discussion mailing list
Thomas D. Shepard
LLNL A Division
925 423 4018
I am looking at suggestions. I have already looked at another NumPy doc that
looks like Python online docs.
What I really would like to do is have:
-A way of retrieving from a stable web page the TOC of documentation that I'd
then put in a custom gnome-tree-like widget and when the user clicks on one item
it'd open a link to online corresponding page.
-An entry widget that'd have a search item that look in index of docs (online)
and return all possible choices then when user chooses one show the
-A bookmarking facility so that users can go back quickly to places they have
-The choice to download the whole .tar.gz thing to browse off-line.
All of the above is easy to implement, and I have already started doing so
with GMatH (http://gmath.sourceforge.net), provided there is a known stable
place where to look for these docs.
I already have a place to look for, but I forgot the address.
All I need is some nifty .py thing that'd fetch the TOC, and index (from web
site) and search through that index. I am not very good with urllib or httpblib,
so any help is welcome.
> From: "Paul F. Dubois" <pauldubois(a)home.com>
> To: "Hassan Aurag" <aurag(a)CRM.UMontreal.CA>,
> Subject: RE: [Numpy-discussion] HTML/SGML docs
> Date: Tue, 9 May 2000 07:33:04 -0700
> MIME-Version: 1.0
> Content-Transfer-Encoding: 7bit
> X-Priority: 3 (Normal)
> X-MSMail-Priority: Normal
> Importance: Normal
> X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2615.200
> The document source is Framemaker, which can produce HTML and PDF.
> It would be best to have an collaborative framework such as the Zope one
> recently announced, if it works well; I haven't tried that yet.
> > -----Original Message-----
> > From: numpy-discussion-admin(a)lists.sourceforge.net
> > [mailto:firstname.lastname@example.org]On Behalf Of Hassan
> > Aurag
> > Sent: Monday, May 08, 2000 7:58 PM
> > To: numpy-discussion(a)lists.sourceforge.net
> > Subject: [Numpy-discussion] HTML/SGML docs
> > Hi,
> > I think I have asked this question a couple of thousand times, but are
> > there plans to have HTML/SGML docs, or in other words, can we expect
> > to have the source form of this beast.
> > I have reached the point where I know how to integrate a nice gtkhtml
> > (gnome html in fact) widget ad browse html docs (HURRAY!). Now I'd
> > like to us it for something. The nifty thing would be for me to have
> > some kind of docbook thinggy. Which means a huge collection of html
> > pages with automagic links to prev, next, up, home..... Then I'd show
> > the contents in one window using a tree-like thing and the rest in
> > another (a real html widget).
> > Thanks again
> > _______________________________________________
> > Numpy-discussion mailing list
> > Numpy-discussion(a)lists.sourceforge.net
> > http://lists.sourceforge.net/mailman/listinfo/numpy-discussion
------------- End Forwarded Message -------------