I have a quick question that is related to engineering support (I was reading the document you posted on the wiki about sd24, and I see the roadmap is planning to address those issues for SAGE 7.0.. ouch!): I don't know exactly how this is achieved, but I can see that
sage: type(1) <type 'sage.rings.integer.Integer'> this should come from the fact that the preparser replaces usual numbers with sage numerical types. Nonetheless, still we have: sage: type([1,2,3]) <type 'list'> This means that list are not replaced by anything internal, even if they are collector of python objects which are still sage numbers. I wonder which is the policy of SAGE towards arrays and matrices. I think there is a good base in numpy and scipy, so I wonder if it would make sense to let the preparser transform any list of numbers into a numpy array. That would greatly improve user ergonomy in case of raw numbers manipulation. That could be a not-default option, that could be activated using a function like: def numpy_mode(str): if str == 'on': from numpy import * "turn any list into a numpy array" elif str == 'off': "reverse behaviour" [I'm just talking on top of my head] Another slightly related issue is: could we make plot(numpy.array([1,2,3,4])) behave like list_plot(numpy_array([1,2,3,4]))? The way it is now, it just doesn't work... The reason I'm doing this is that I assume a big income of users when the windows port is ready (soon) and maybe there are a lot of non- mathematicians among them. Could it be that it make sense to create a wiki page with all the desiderata about SAGE/engineering? Maybe that may help in the long run, like the one I created for the physical units package! I can see that other scientists are in this community, I'm wondering if they may help, being even more experienced in contributing to SAGE. Cheers Maurizio On Jul 14, 1:57 am, William Stein <wst...@gmail.com> wrote: > On Wed, Jul 14, 2010 at 1:42 AM, Maurizio <maurizio.gran...@gmail.com> wrote: > > I just spend a couple of words about IDEs. I've personally spent a > > decent amount of time on Spyder and Eric, and my impressions are: > > - Eric is very well suited for general software development, it is not > > completely polished, and it lacks (at least explicitly, I didn't get > > those) useful features for scientific computing (which I'll later > > mention) > > - Spyder is theoretically exactly what I was looking for: it is a > > pythonized version of the Matlab GUI, which I felt very comfortable to > > use; nonetheless, the problems are there: > > 1) I find it very slow (even if I may have problems with matplotblib, > > my workstation is quad-core and generally fast), certainly slower than > > Eric (I think both are written in Python) > > 2) window management is awful: if you undock an internal subwindow, > > you are forced to not move it again within the area of the main Spyder > > window, otherwise it immediately redocks it > > 3) integrated plot management looks pretty, but (it's entirely not > > Spyder's fault) matplotlib just outputs pictures, so plot navigation > > is still orders of magnitude less evolved than Matlab's (I know it may > > sound silly, but is that so difficult to do something better??) > > on the pros side I count: > > 1) both internal and external console: the former one is useful to do > > experiments within the script you are editing, the latter is better to > > have a clean environment > > 2) variables management and browsing > > 3) enhanced editing (code completion, syntax highlighting, classes > > identification and browsing) > > Very interesting. > > 1. How does the speed of the Sage notebook running locally on your > computer compare to Spyder locally on your computer? > > 2. Are the plotting issues you mention the result of Spyder embedding > static png images (like the sage notebook does) or something more > subtle. The sage notebook might switch to HTML5 canvas rendering > soon.... I say might, because after having tried it a bunch, I'm > seriously concerned that HTML5 canvas matplotlib is slow -- > surprisingly, maybe much slower than using png's and image maps, which > we should have at least enabled long ago. > > 3. I have talked with people about making a Matlab-clone-ish version > of the Sage notebook. This would be web-based, but instead of feeling > Mathematica-like, it would feel much more Matlab-like. Thoughts? > > > > > I don't know what about outside Europe, but I find so strange that > > SAGE is unknown in scientific community, I find it very useful (from > > an engineering point of view), and I personally think that may be a > > perfect solution to be introduced inside universities at first (thanks > > to the wonderful internet-based notebook system). > > I was also very surprised. But it is simply a fact I observed. Well, it > wasn't > so much that Sage is *unknown* -- many people knew about it. What I > noticed at Euroscipy is that very few of the people there used Sage. > Not a single speaker > (except me) said they used Sage, and there were nearly about speakers > (including > lightning talks). > > > The problem I see now regarding scientific computing, is the not so > > seamless integration of numpy-scipy: do you think SAGE may improve > > numpy arrays management with cleaner syntax than regular python? I > > know you are usually against introducing syntax that is unacceptable > > in standard python, but I think that allowing users to avoid writing > > "np.array()" to do any kind of vector manipulation would be highly > > appreciated! > > Yes, this is definitely a Sage goal. I talked a lot about this goal > with Dag last weekend (he's one of the lead Cython developers). > > > I strongly support SAGE for science!! :) > > Thanks. > > > By the way (not completely off-topic) a colleague of mine is having > > some troubles in working with scipy.optimize within SAGE, but I have > > no details right now... I should better check! > > Yep, report it. > > > > > > > > > My 2 cents > > > Thanks > > > Maurizio > > > On 11 Lug, 20:41, Ondrej Certik <ond...@certik.cz> wrote: > >> On Sun, Jul 11, 2010 at 3:20 AM, William Stein <wst...@gmail.com> wrote: > >> > Hi, > > >> > 1. IDE's > >> > There are a number of IDEs that can be used for Python development: > > >> > * Spyder (free, cross platform) --http://code.google.com/p/spyderlib/ > >> > * Eric (free, cross platform) --http://eric-ide.python-projects.org/ > >> > * PyDev + Eclipse or Aptana (free, cross platform) --http://pydev.org/ > >> > * Wing IDE (non free, but has a 30-day trial) > >> > --http://www.wingware.com/ > >> > * XCode (free, closed, OS X only) > > >> > I'm at EuroScipy and many of the scientists and engineers giving talks > >> > mention some of these IDE's (especially Spyder). It would be of > >> > interest to make a page athttp://wiki.sagemath.orgabouteach of the > >> > above IDE's in the context of Sage. Which can be used with Sage? > >> > How? Do they work on anything but Linux, etc. Any volunteers? This > >> > could be a good student project (so possibly some funding for > >> > something at UW). > > >> > 2. Sage at EuroScipy: > > >> > Another thing -- though most talks mention Cython, not one single talk > >> > given about actual engineers/scientists doing work even mentioned Sage > >> > -- and there were over 30 talks. Perhaps there is no penetration at > >> > all of Sage into scientific computing, at least in Europe. Perhaps > >> > this will change in the next few years, given that NSF looks highly > >> > likely to fund this NSF granthttp://wstein.org/grants/compmath09/ > > >> > Sage was only mentioned in the first keynote by Langtangen, in which > >> > he explained that installing Python for his students is very hard. > >> > His personal solution -- force the students to install Ubuntu, either > >> > natively or in a Virtual Machine. Full stop. > >> >http://picasaweb.google.com/wstein/20100710EuroscipyDay1#549240022431... > >> > He made some (funny) jokes about being a dictator. > > >> > I personally disagree with his suggested "solution". Maple, Matlab, > >> > Mathematica do better, and so can we. > > >> Yeah, definitely. I am now working at the Lawrence Livermore National > >> Lab during the summer and I don't have a root access to my computer, > >> and it is not running Ubuntu. So his solution would be a complete > >> failure for me. > > >> I am running our latest git femhub:http://femhub.org/andthat > >> creates me a nice environment, and I use "femhub --shell", which is > >> like "sage -sh", except that the prompt looks better: > > >> FEMhub: ond...@raven:~/repos/hermes1d(master)$ > > >> Here are the packages that are in femhub: > > >>http://femhub.org/codes.php > > >> At least for me, it's now doing exactly what I need. > > >> Another problem is with gui ---- I couldn't get any working for > >> matplotlib. So I would like to get the html5 canvas working for > >> matplotlib. > > >> Also I would like to have some easy way to create guis, it should run > >> in the browser. Using extjs:http://www.sencha.com/products/js/, but > >> I'd like to somehow write it in Python, so that I don't have to mess > >> up with javascript. > > >> Ondrej > > > -- > > To post to this group, send an email to sage-devel@googlegroups.com > > To unsubscribe from this group, send an email to > > sage-devel+unsubscr...@googlegroups.com > > For more options, visit this group > > athttp://groups.google.com/group/sage-devel > > URL:http://www.sagemath.org > > -- > William Stein > Professor of Mathematics > University of Washingtonhttp://wstein.org -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org