Keith Devens .com

Dumb substring behavior in C# (and Java)

Substring in C# (and Java) throw an exception if you take a substring and give it a length (or a starting position) that puts you after the end of the string. That's basically to ensure that you always get a string that's exactly the length you want? Rather than just being able to take a substring and not having to worry about it, you have to include code like the following around every single substring you ever take:

if (str.Length > 30){

    str = str.Substring(0, 30);

}

Typically when you take a substring you want to ensure that your string is no longer than a certain number of characters. I can't think of a situation that would it a good idea for the language to enforce that you can't get a string that's less than the maximum length you want instead of exactly the length you want.

On the other hand, most of the dynamic languages (PPPR) let you substring off the end of the string without worrying about it.

Permalink | 2 Comments | Tags: C#, Java, Perl, PHP, Python, Ruby | id 8114

13:31 (utc-5)

Thursday, June 29, 2006

Monday, April 3, 2006

Thursday, March 2, 2006

Sunday, February 26, 2006

Tuesday, February 21, 2006

Monday, February 20, 2006

Monday, January 16, 2006

Tuesday, January 10, 2006

Saturday, December 31, 2005

Wednesday, December 21, 2005

Wednesday, December 14, 2005

Python's metaclasses

Dev Shed has an excellent introduction to Python's metaclasses (via DPU). I get it now:

Python's metaclasses are just normal classes that inherit from type
When you instantiate a metaclass the object returned is itself a class
To declare that a class has a given metaclass, set the __metaclass__ data member to the metaclass when declaring the class.
The metaclass' __init__ method is called when a class that has it as its metaclass is declared (__init__ for metaclasses takes (klass, klass_name, base_classes, dict_of_attributes) as its parameters).

And that's it (according to the article). So, it seems that Python very nearly has a prototype-based object system, but not quite. (And I ❤ prototype-based object systems.) With a prototype-based object system the distinction between metaclasses and classes (heck, and objects) goes away, so it seems to me that you could do the same thing as what Python calls metaclasses simply by overriding a particular object's clone method. Am I wrong?

Permalink | 2 Comments | Tags: Metaclasses, Programming, Prototype OO, Python | id 7727

02:25 (utc-5)
edited: 02:34 (utc-5)

Friday, December 9, 2005

Thursday, December 8, 2005

Tuesday, December 6, 2005

Friday, December 2, 2005

Unusual Python list comprehensions

I needed to write a little code that would grab the filenames of all zip files out of a directory hierarchy. I immediately tried to write it as a big list comprehension and completely failed. So, I backed off a step to the following (some code in this post refomatted to not break my layout):

finalfiles = []

for (dir, subdirs, files) in os.walk("c:/temp/"):

    finalfiles.extend([

        os.path.join(dir, file)

        for file in files if file.lower().endswith(".zip")

    ])

I took a look at the list comprehension examples in the Python quick reference to figure out why I was having a problem. Here is their reference code (the two blocks are equivalent):

result = [expression for item1 in sequence1  if condition1

             for item2 in sequence2 ... for itemN in sequenceN

         ]



result = []

for item1 in sequence1:

    for item2 in sequence2:

        ...

        for itemN in sequenceN:

            if (condition1) and further conditions:

               result.append(expression)

So, first I converted my finalfiles.extend() to an append() form:

finalfiles = []

for (dir, subdirs, files) in os.walk("c:/temp/"):

    for file in files:

        if file.lower().endswith(".zip"):

            finalfiles.append(os.path.join(dir,file))

so that I could directly translate from the nested loop version to the list comprehension version, essentially doing textual substitution of item1, expression, and so on.

After doing the substitution, this is what you wind up with:

finalfiles = [

    os.path.join(dir,file)

    for (dir, subdirs, files) in os.walk("c:/temp/")

    for file in files if file.lower().endswith(".zip")

]

That seems backwards to my brain, which is why I didn't get it right the first time. I think this is really a case where doing a list comprehension is not clearer than doing at least part of it in a normal loop fashion.

Note: my brain feels like it should be this way:

finalfiles = [

    os.path.join(dir,file)

    for file in files if file.lower().endswith(".zip")

    for (dir, subdirs, files) in os.walk("c:/temp/")

]

which corresponds rather to:

result = [expression for item2 in sequence2

    for item1 in sequence1

]

I think that makes more sense. Which fits your brain?

Permalink | 2 Comments | Tags: Programming, Python | id 7662

04:09 (utc-5)
edited: 05:06 (utc-5)

Thursday, November 24, 2005

PLNews: IronPython Demo Video Available

PLNews: IronPython Demo Video Available. To watch. Argh, you need Internet Explorer to play it. Bah.

Oh my goodness. First time through the video IE blocked the popup the video requires half-way through. Then I temporarily enabled popups in IE, and this time around the Google toolbar bundled with my computer (never turned it off since I've only used IE once to download Firefox) blocked it. Argh! And I can't even get to the Google toolbar for this window since the window has no toolbars. They could have just done the normal thing and opened up a Windows Media Player instance and been done with it.

Hey, neat presentation so far. But why does everyone who works for Microsoft sound like they've drunk the Kool-aid?

Whoa, I didn't know you could do the following in the Python interactive shell:

>>> foo = ["foo", "bar", "baz"]

>>> foo

['foo', 'bar', 'baz']

>>> bar = _

>>> bar

['foo', 'bar', 'baz']

Very cool when he debugged Python, going through stack frames back into the C# code that called it.

Permalink | Comments? | Tags: IronPython, Programming, Python | id 7624

00:07 (utc-5)
edited: 00:38 (utc-5)

Tag: Python

Thursday, April 24, 2008

Wednesday, November 8, 2006

Tuesday, August 1, 2006

Thursday, July 27, 2006