🌐 Mailman 3 April 2003 - Python-list - python.org

Jython, getting instance name from the interpreter
by bob_tommy＠hotmail.com May 4, 2003

May 4, 2003

Does anyone know to get the instance names from the interpreter? I want the getMyName method to return "instance" in this case, or a list of names if many... Thanks for any help!! Tommy public class InterPreter() { private static PythonInterpreter interp = new PythonInterpreter(); private static String myscript = "instance = MyClass()"; public InterPreter() {interp.exec(myscript);} public static String getInstanceName(Myclass mycl) { //hmm, how to do this? } } public class MyClass() { public String getMyName() { return InterPreter.getInstanceName(this); }

2 1

__setitem__ puzzle
by anton＠vredegoor.doge.nl May 3, 2003

May 3, 2003

I have subclassed the builtin list type to accept tuples for indexing. I am just experimenting with the code to see what I can do with it, so asking for what the code will be used for in order to provide a better way to do it will be answered with 'ni' :-) There's a problem with __setitem__ however. What's the proper way to write this class? Anton class tlist(list): def __getitem__(self,x): y = self[:] for z in x: y = y[z] return y def __setitem__(self,x,val): y = self[:] for z in x[:-1]: y = y[z] y[x[-1]] = val def test(): L = tlist(['aa',['bb',['cc','dd'],'ee'],'ff',['gg']]) indices = [(0,), (1,), (1, 0), (1, 1), (1, 1, 0), (1, 1, 1), (1, 2), (2,), (3,), (3, 0)] for x in indices: print L[x] print L[(1,1,0)] = 0 print L L[(1,1)] = 0 print L #this doesn't work L[(1,)] = 0 print L if __name__=='__main__': test()

5 8

Re: Question about accessing class-attributes.
by mis6＠pitt.edu May 2, 2003

May 2, 2003

Alex Martelli <aleax(a)aleax.it> wrote in message news:<mailman.1051689777.10470.python-list(a)python.org>... > <snip> Indeed if we were able to > classify the current behavior as "a bug" in some respect, AND give > a patch that breaks no currently designed-in behavior, yet "fixes the > bug" and makes super more useful to you, then that's the kind of > fix that COULD go in 2.3 beta 2 (new functionality won't, but in > some sense it might be considered "a bug" if super violates the > normal expectations of clearly-advanced users such as you). > <snip> > Maybe we should break this thread in two -- one about metaclasses > and one about super -- with suitable subjects... I do not understand > where the 'bugs' of the two may be related nor actually what they > really have to do with "accessing class-attributes" except in the > vaguest, broadest sense. > <snip> > Yes, maybe we should separate super-issues and metaclass-issues > and find a sharp example in each case of where reasonably expected > behavior from docs, and observed behavior in 2.3 beta 1, differ. I > think it does not need to be a "reasonable use case": that would be > needed if we were arguing for EXTRA functionality of course -- but > then it would go in 2.4 so there's no hurry, to say the least. If we're > arguing there are BUGS, so the fixes could go in 2.3, then just a > way to show up each bug clearly should suffice (hopefully). > > > Alex I do agree with you. Speaking about wanted functionality (maybe for 2.4) I would like very much a __superclass__ attribute. Example: now I can do (NOTICE: in p23b) class B(object): pass class C(B): pass c=C() super(C,c).__thisclass__() #=> C super(C,c).__self_class__() #=> C I would like to have an additional attribute (not a method) returning the superclass of C, in this case B: super(C,c).__superclass__ #=> B For instance I could do b=super(C,c).__superclass__() returning an instance of B. I think Bjorn would we happy with this, too. Certainly the documentation in http://www.python.org/dev/doc/devel/lib/built-in-funcs.html has to be changed. It says: """ super(type[object-or-type]) Return the superclass of type. If the second argument is omitted the super object returned is unbound. If the second argument is an object, isinstance(obj, type) must be true. If the second argument is a type, issubclass(type2, type) must be true. """ But it DOES not return the superclass! I admit I myself at the beginning was confused. Michele

2 7

Textbooks?
by Steven Hodgen May 2, 2003

May 2, 2003

Hello, I'm currently teaching a high-school introduction to programming course. I'm currently teaching in Java, but am seriously considering changing to Python. The problem is that I cannot seem to find any good textbooks, much less ones for students with no prior programming experience. This is the one area in which Java excels. I'm using an awesome programming environment called "BlueJ", and there are lots of textbooks for students at all levels. However, Java is not a great language for beginners. Python, on the other hand, is ideal. If I can find a textbook that is even ok, I may consider changing, since the language is so much better for students at this level. Doesn anyone know of a book like this? Ideally, it would have examples and excercises, and perhaps even a web-site with other instructor and student resources. Thanks! --Steven steven(a)twitch.net

21 24

win32 adding a tcpip alias to an interface
by David Bear May 2, 2003

May 2, 2003

I've never used the win32 extensions to python. Most of my python tinkering has been around unix. However, I am contemplating a script that would add a second ip address to an interface. Here's the general idea: 1) when a windows box starts, it gets a dhcp lease on a non-routable IP address, web browsing is handled through NAT 2) if the users needs a fully routable ip address, he authenticates to something -- this could be Active Directory, Samba, Kerberos -- something else. 3) after successful authentication the user is given a fully routable ip address So, the python script would need to handle 2 and 3. There are probably classes someone has written to do authentication (haven't looked yet) But, I'm worried about how one might add an ip address to a windows tcpip stack. Anyone have any ideas how to do this? tia

4 4

Python 2.3b1 and _tkinter.Tcl_Obj
by Stephen D Evans May 2, 2003

May 2, 2003

Is this Python or Tk/Tcl? Is it a bug or a feature? There are some inconsistencies in the types returned by some Tkinter widgets when using cget() or configure() e.g. widget.cget('background') Menu widgets return a Tcl_Obj (typename='border'), Buttons return a string. The following code demonstrates using configure() rather than cget() # Python 2.3b1 (#40, Apr 25 2003, 19:06:24) [MSC v.1200 32 bit (Intel)] on win32 # import Tkinter import _tkinter import types # workaround #Tkinter.wantobjects = 0 for name in Tkinter.__dict__: c = getattr(Tkinter, name) if isinstance(c, types.ClassType) and Tkinter.Widget in c.__bases__: widget = c() tup = widget.configure('background')[3:] # last two print c, type(tup[0]), type(tup[1]) assert 2 == len(tup) for i in tup: assert isinstance(i, (types.StringType, _tkinter.Tcl_Obj)) assert not isinstance(i, _tkinter.Tcl_Obj) or i.typename == 'border' #----------------------- I found this while fiddling with the Pmw megawidget (http://pmw.sf.net) tests and demos under Python 2.3b1 Stephen D Evans

2 5

2.3b1 PackageManager fails to initialize, Mac OS X
by David Lewis May 2, 2003

May 2, 2003

I've built Python 2.3b1 from the tarball. I've got WASTE installed, so it built the PythonIDE, etc. If I select PackageManager from the PythonIDE file menu, I get a traceback from urllib2.py, when it tries to get 'http://homepages.cwi.nl/~jack/pimp/pimp-darwin-6.5-Power_Macintosh.plis t'. I can see that page in a browser, so that's not the problem. I can manually execute the following code in the PythonIDE interactive window: import urllib2 f = urllib2.urlopen( 'http://homepages.cwi.nl/~jack/pimp/pimp-darwin-6.5-Power_Macintosh.plis t') for line in f: print line and that all works. Any ideas? David

2 2

POD vs. reST for standalone writing?
by eichin＠metacarta.com May 1, 2003

May 1, 2003

So, after a recent bad experience with LyX and version skew, I was looking at alternatives for software end-user and administrator documentation. The constraints are pretty simple: * produce good html and PDF ** via linux tools, and highly preferably free ones * be editable on multiple platforms ** linux first, but also, osx, windows * support "literal computer output" and screen shots DocBook was under consideration (and was used for some man pages and standalone docs) but has a high learning curve, and none of the windows dtd-aware or docbook-aware editing tools actually seemed to *work* well enough to hand them to non-techies (the techies, on linux, are already happy with emacs and vi :) Also, the subject of the documentation changes rapidly enough that having document specialists just isn't acceptable, everyone involved needs to be able to update things. Anyway, POD came up as a documentation format in a related context (in particular, "I'm going on vacation, here's a walk through of how that code works, don't call me" :) and it was very effective for output that was dominated by verbatim code snippets interspersed with text; in fact *easier* to write in than raw HTML would have been, though probably no more expressive. PDF output was not unreasonable (via html or more directly via latex, so I haven't had to dig further.) Now I start seeing mention of reStructuredText (especially in comments about pyCon.) reST files look very "ascii art" on the input side, but after a little experiment, that seems to make them *harder* to write, if somewhat (and I think only a little) easier to read. Am I missing something? Is there actually a reST clear short hand, and a tool that expands it to full-blown ascii-art? Or is at least fuzzy in the counting of dashes and other punctuation for marking different levels? (POD is perhaps a little too simple, but a first cut POD Parser class in python, with a derived pod2html was the work of a recent cross-country plane flight, so that's not all bad... and I'm not sure that a pod2reST would be that hard either...)

10 12

Some oddities in 2.3b1, esp. with Tkinter
by Russell E. Owen May 1, 2003

May 1, 2003

I finally got my app almost running in 2.3b1. I've run into the following: 1) A filter to remove fonts with non-ascii names (used to filter out stuff that I could not display in a pop-up menu) broke. try: unicode(fontName, 'ascii', 'strict') return True except UnicodeError: return False Python 2.2 throws UnicodeError. Python 2.3b1 throws TypeError. The fix is trivial (e.g. "except (UnicodeError, TypeError)". The filter is probably not needed anymore anyway, since Tk 8.4.2 can apparently display menu items with non-ascii characters. 2) Tk file events are no longer supported: RuntimeError: _tkinter.createfilehandler not supported for threaded Tcl This is painful since it's hard to see how to write a proper networking GUI with Tkinter without file events for reading data. Perhaps I can get them back again by disabling the new threaded tcl support. However, I suspect it's time to try out the Twisted framework. 3) Tk returning things as objects is nice, but I had this break at least one thing of mine and it will break others (see 4 below): I was assuming that the values in the dictionary returned by tkFont.Font.config() were strings. Fortunately this was easy to fix. A bit disconcerting -- I wonder how many any other little pitfalls like this still buried in my code -- but well worth the cost of the change. 4) Tk Variables are trickier to use and the change will break some existing code. Tkinter used to return a variable name if the variable of a widget was requested. Not as convient as returning the original variable, but once one found the magic setvar and getvar functions (which work on variable names) things were fine. Now Tk returns a _tkinter.Tcl_Obj instead, and and setvar and getvar don't seem to work on that. (I'm relieved that I never actually used setvar and getvar in my final code; despite many temptations I was always uncomfortable with it). Example: v = StringVar() v is a <Tkinter.StringVar instance at 0x2c98a0> l = Label(textvariable=v) x = l["textvariable"] x is reported as a <namespaceVarName object at 0x0168b048> but is a _tkinter.Tcl_Obj based on the following traceback: root.getvar(x) Traceback... TypeError: getvar() argument 1 must be string, not _tkinter.Tcl_Obj Note that x.get() doesn't work either -- i.e. the object does not mimic a Tk Variable. I'm guessing there is some way to manage this, but have no idea what it might be. 5) Note that Tkinter's handling of font objects does not seem to have changed. I was hoping it would start returning font objects instead of strings, or offer some way to convert the string to a font object, but apparently not: f = tkFont.Font(font="system") l = Label(font=f) l["font"] returns: 'font2922816' -- Russell

4 6

problems using magic "<broadcast>" socket address on different platforms
by Irmen de Jong May 1, 2003

May 1, 2003

Hello I'm trying to use UDP broadcasting but there are some weird things going on regarding Python's magic "<broadcast>" address (which is the INADDR_BROADCAST address). Using the following piece of Python code to send UDP broadcast messages: ---- from socket import * sock = socket(AF_INET, SOCK_DGRAM, IPPROTO_UDP) sock.setsockopt(SOL_SOCKET, SO_BROADCAST, 1) sock.sendto("Het Bericht", ("<broadcast>", 2000) ) print "done" ---- This works fine. The problems are in the server code: ---- from socket import * import select import sys sock = socket(AF_INET, SOCK_DGRAM, IPPROTO_UDP) sock.setsockopt(SOL_SOCKET, SO_BROADCAST, 1) sock.bind( ("<broadcast>", 2000) ) while 1: (rs,ws,es)=select.select([sock],[],[],1) if sock in rs: (data, addr) = sock.recvfrom(9999) print "Got data: {%s}" % data print "from:",addr else: print ".", sys.stdout.flush() ---- When running this on Linux, everything works as expected. When running the server code on Windows, Python fails to bind the socket on the magic "<broadcast>" address: socket.error: (10049, "Can't assign requested address") I've had reports of users that the same problem is present on Mac OS X. Because I don't have access to other platforms than Windows (NT) and Linux, I am wondering what the results are on other systems such as Solaris, *BSD... --Irmen de Jong PS The really weird part is that a piece of C code that directly uses the BSD socket API *also* fails on windows with the same error (bind: Cannot assign requested address, when trying to bind the socket on the INADDR_BROADCAST address). The exact same piece of C code runs fine on Linux. Used GCC in both cases. If you're interested in this code let me know, I thought it was too big to post here. PPS To get my code to work on windows, I had to use the magic "" address (INADDR_ANY). But I don't think that's right, "<broadcast>" should work, shouldn't it?

2 2