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

RE: list index question
by Delaney, Timothy C (Timothy) April 1, 2003

April 1, 2003

> From: Matt Singer [mailto:mrsinger@usa.net] > > Can I wildcard? I don't know the values in the second column. I want > to find the first entry what has a particular value in the first > column. That's not at all what you said. Next time, please ask the question you want an answer to. Anna's response was perfectly correct for the question you asked. http://www.catb.org/~esr/faqs/smart-questions.html No, you cannot wildcard - it doesn't make sense for this. You need to define you problem further. 1. Do you want the *index* of the first tuple which matches or 2. Do you want the first tuple that matches itself. 3. Is the data in the sequence sorted? If so, how? In any case, since you already have the data in a sequence, looping is probably the best option here - it has the advantage of being O(N). If the data is already sorted in a way that you want, you may be able to use a more efficient search algorithm to find *one* of the entries that matches, then search backwards until you find one that doesn't match. Tim Delaney

1 0

Re: ANNOUNCE: SCons 0.13 (build tool in Python) is now available
by Syver Enstad April 1, 2003

April 1, 2003

I get an error when trying to download the 0.13 version. Sourceforge complains that the file doesn't exist. -- Vennlig hilsen Syver Enstad

1 0

Re: When Python outruns C++ ...
by tweedgeezer＠hotmail.com April 1, 2003

April 1, 2003

Jp Calderone <exarkun(a)intarweb.us> wrote in message news:<mailman.1049181423.2644.python-list(a)python.org>... > It looks like this could easily be improved with the application of a > slightly smarter searching algorithm, such as Rabin-Karp search or > Boyer-Moore find. These searches depend on a fixed needle to search for in the haystack. They can't be applied to a regular expression search. Jeremy

2 1

Re: Terminal-based text editor with color syntax hilighting
by Dave Benjamin April 1, 2003

April 1, 2003

In article <mailman.1049223987.18476.python-list(a)python.org>, Lulu of the Lotus-Eaters wrote: > Dave Benjamin <ramen(a)lackingtalent.com> wrote previously: >|Are there any others? What about Fte, I wonder... is that editor/IDE still >|being actively maintained? According to their website it supports Python >|color hilighting, but I'm not sure if the s-lang version does. I'll check >|into it. > > Unfortunately, AFAIK FTE only works in Unix-like systems over XTerm. A > console session doesn't do FTE (under Linux, OS/2 has both console and > PM versions of FTE). As with the OP, this rules out using FTE on most > of my connections--my web host doesn't run an X client, my OS/2, Windows > and MacOSX machines don't run an X server). Actually, believe it or not, I just looked into this and there are three versions of FTE, one for X, one for a terminal (like SSH) and one for console (I imagine they do direct screen-writes, since it seems to require root privileges). I was able to get color hilighting and pulldown menus working with a fresh install. So, this is a viable alternative. > IMO, the best option for the rather common scenario of remote SSH > connection is Jed. But I would welcome learning of others. I *have* > been using vim on one host recently, and have kinda-sorta learned basic > navigation, but it won't stick in my head after a month (I know from > experience). Likewise with emacs. That's been my experience also, more or less. I really like the speed and familiarity of Jed. It gets a little screwy about Python triple-quotes sometimes, depending on what version you have, but overall its auto-indentation works great for Python. > Jed has this wonderful feature of having topbar menus that actually let > you visually select the most common actions (without needing to remember > a cryptic key-sequence). This fabulous design had only been in DOS > text editors since 1982, so it's nice that Linux is starting to catch up > with that. Hehe... if you like that, I'd say FTE is even more in that direction. It reminds me of editors like Boxer and the TurboC++ environment. It's a bit more of an IDE than a plain-vanilla text editor though. <ot> DOS had the advantage of being relatively predictable, at least as far as text mode was concerned. Running full-screen text applications remotely was another story entirely (anyone remember Doorway?) because of the direct screen writes... anyhow, I'd imagine developing full-screen text apps for *nixes (and being compatible with the various kinds of terminals) is a more difficult endeavor, even with the help of ncurses/s-lang/etc. </ot> Thanks, Dave

1 0

Re: Implementing a "classmethod" in an extension type
by Thomas Heller April 1, 2003

April 1, 2003

"Mark Franklin" <mfranklin(a)tcsi.com> writes: > Thanks Thomas, > > The key was the DoClassMethods() call at the end of your code. Once > I'd copied that over, it worked perfectly:) > It seems I forgot to include the code for DoClassMethods() in my reply, sorry for that. But you found it anyway ;-) Thomas

1 0

Re: PEP 313 - Roman numeral literals
by Peter Hansen April 1, 2003

April 1, 2003

Dave Brueck wrote: > > On Tue, 1 Apr 2003, Mike Meyer wrote: > > > PEP: 313 > > Title: Adding Roman Numeral Literals to Python > > Version: $Revision: 1.2 $ > > Last-Modified: $Date: 2003/04/01 17:41:34 $ > > Author: Mike Meyer <mwm(a)mired.org> > > Status: Draft > > Type: Standards Track > > Content-Type: text/plain > > Created: 01-Apr-2003 > > Python-Version: 2.4 > > Post-History: > > +1, well done! > > Before we submit this to the BDFL, I propose that we wait a short while to > give the community an opportunity to debate its merits, followed by a > vote, just so Guido can get idea of where the community stands. Further, I > propose that Aahz be the official vote counter. Does anybody have any > preferences as to the voting method we should use? Wouldn't "we who are about to die salute you" somehow be appropriate here? SPQR!

1 0

Anything like Mailman's LockFile.py in std lib?
by Gregory (Grisha) Trubetskoy April 1, 2003

April 1, 2003

Is there anything being planned for the standard lib that provides locking a la LockFile.py in Mailman? I find it extremely useful, but just recently came across a project where it can't be used becuase it's GNU licensed - thus the question. Grisha

2 1

RE: Precidence of unary plus/minus relative to exponentiation
by Batista, Facundo April 1, 2003

April 1, 2003

#- >>> -2.0**2.0 #- -4.0 #- #- Python's operator predicence is causing this to be interpreted as: #- #- >>> -(2.0**2.0) It's OK, because the minus and the plus are the last verified signs, in Python and everywhere else, as far I know. #- rather than the normal mathematical precidence, which would give #- #- >>> (-2.0)**2.0 No, no, no. In "everyday math" if you want to elevate the -2 to 2, you need to put the brackets explicitly. .elm.

1 0

RE: Implementing a "classmethod" in an extension type
by Mark Franklin April 1, 2003

April 1, 2003

Thanks Thomas, The key was the DoClassMethods() call at the end of your code. Once I'd copied that over, it worked perfectly:) void DoClassMethods(PyTypeObject *type) { PyObject *func; PyObject *meth; PyMethodDef *ml = type->tp_methods; for (; ml->ml_name; ++ml) { if ((ml->ml_flags & METH_CLASS) == 0) continue; ml->ml_flags &= ~METH_CLASS; func = PyCFunction_New(ml, NULL); if (!func) return; meth = PyObject_CallFunctionObjArgs( (PyObject *)&PyClassMethod_Type, func, NULL); if (!meth) return; if (-1 == PyDict_SetItemString(type->tp_dict, ml->ml_name, meth)) return; } } Mark > -----Original Message----- > From: Thomas Heller [mailto:theller@python.net] > Sent: 01 April 2003 16:10 > To: python-list(a)python.org > Subject: Re: Implementing a "classmethod" in an extension type > > > "Mark Franklin" <mfranklin(a)tcsi.com> writes: > > > Hi, > > > > I've got an built-in extension type, implemented in C. I > want to be able to implement the equivalent of a > "classmethod" in that type. > > > > I thought I might be able to do something like this, which > fairly closely mimics what you do in Python: > > > > PyTypeObject myType = { > > // initialised in the usual way. It's > tp_methods field includes a method definition for "MyMethod"....... > > }; > > > > PyObject *method = PyObject_GetAttrString(myType, "MyMethod"); > > // method is a PyMethodDescrObject > > PyObject *classMethod = PyClassMethod_New(method); > > PyObject_SetAttrString(myType, "MyMethod", classMethod); > > Py_DECREF(method); > > > > PyType_Ready(myType); > > > > > > However, when I do this I get a TypeError exception from > the PyObject_SetAttrString call: "can't set attributes of > built-in/extension type 'MyType' ". > > > > So, how do I get the class method into the type object? > > > > > > Thanks, > > Mark > > In Python 2.3, there is a METH_CLASS flag. Unfortunately not in 2.2. > Here are the relevant snippets that I use - it is from > http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/ctypes/ctypes/s > ource/_ctypes.c > and it compiles with Python 2.2 as well as 2.3. > > HTH, > > Thomas > > > #include "Python.h" > > #ifndef METH_CLASS > #define METH_CLASS 0x0010 /* from Python 2.3 */ > #define NO_METH_CLASS > #endif > > ... > > /* > * If this is a real C classmethod (Python 2.3 and later), > * it has METH_O style. So the type is in the first argument, > * and the arg in the second. > * > * We fake this in Python 2.2, where METH_CLASS is not present, > * and 'args' is a tuple containing the type as first member. > */ > static PyObject * > CString_from_param(PyObject *cls, PyObject *args) > { > PyObject *value; > > #ifdef NO_METH_CLASS > PyObject *ignore; > > if (!PyArg_ParseTuple(args, "OO", &ignore, &value)) > return NULL; > #else > value = args; > #endif > ... > } > > > static PyMethodDef CString_methods[] = { > #ifdef NO_METH_CLASS /* Python 2.2 */ > { "from_param", CString_from_param, METH_VARARGS | METH_CLASS, > #else > { "from_param", CString_from_param, METH_O | METH_CLASS, > #endif > from_param_doc }, > { NULL, NULL }, > }; > > > ... > > if (PyType_Ready(&CString_Type) < 0) > return; > #ifdef NO_METH_CLASS > DoClassMethods(&CString_Type); > #endif > PyModule_AddObject(m, "c_string", (PyObject *)&CString_Type); > > -- > http://mail.python.org/mailman/listinfo/python-list >

1 0

RE: PEP 313 - Roman numeral literals
by Batista, Facundo April 1, 2003

April 1, 2003

#- #- In addition, any calculations involving the Roman calendar #- (cf http://www.clubs.psu.edu/aegsa/rome/romec.html) would #- clearly benefit #- from such a module. #- #- My kalend.py module, currently running at 87,543 lines of code, could #- be shrunk substantially if it could do its calculations using the #- new roman type. *Thank you*, Mike, for starting the process #- of getting #- this valuable feature into Python. I think Mike is proposing a built-in function. If this is the case, I'll like these two: literal IntToRoman(int) int RomanToInt(literal) But if we're thinking in a module, I'll like to have an object called Roman (or so) where I could make something like this: myRoman1 = Roman() myRoman1.fromInt(15) myRoman2 = Roman() myRoman2.fromString("X") myRoman3 = Roman() myRoman3 = myRoman1 + myRoman2 print myRoman3.toInt() 25 print myRoman3.toString() "XXV" A nice function in the module could be too: Roman.verify(literal) returning True if the string is a "well written" roman number (after, if we call Roman.fromString is allright, othwerwise it will raise an error) I salute you. .elm.

1 0