Example: Why does .NET promote bad programming practices?

Ray Cassick

Although I tend to try to stay away form case statements that are on anything other than numbers (or enums as the case may also be) I can see where that would have it usefulness. It is difficult, as it is in any syntactical/grammatical base question, to say outright if it is 'wrong' without some surrounding context. I will admit that I have done somewhat similar things in cases where I asked the user to verify with an exact text string that they want to do something (IE: delete a record form a database). Of course once again you have to put it into context. My app was not going to be used outside the company so I had no language/culture problems to deal with that would have made string comparisons a problem. It was a simple solution to a very narrow problem.

My Blog[^]
FFRF[^]

Clickok

Marc Clifton wrote:

Now, I personally don't feel a language should (or even can) enforce good programming practices. But a language that supports a bad practice certainly doesn't help the situation.

Yeah, but I really like of the freedom of choice.

Marc Clifton wrote:

Anyone want to contribute other examples or thoughts?

Once I have suggested in MSDN Feedback Center this language improvement (in bold):

for(int i=0; i<10; i++) as externalFor
{
for(int j=0; j<10; j++) as middleFor
{
for(int k=0; j<10; k++) as innerFor
{
if (someCondition)
{
break middleFor;
}
if (anotherCondition)
{
continue externalFor;
}
}
}
}

and they told me to use goto statements... X|

Engaged in learning of English grammar ;)
For God so loved the world, that he gave his only begotten Son, that whosoever believeth in him should not perish, but have everlasting life.(John 3:16) :badger:

DavidNohejl

Phrase "astronaut architects" is not known to be, but I think I agree :) Using TryParse when dealing with user input is usually sufficient. If TryParse methods are not fast enough, by all means implement yours. edit: oops, didn't see you responded to original post and not TryParse debate above. Well, surprisingly my post didn't lost its meaning. :->

"Throughout human history, we have been dependent on machines to survive. Fate, it seems, is not without a sense of irony. " - Morpheus

DavidNohejl

Interesting, this is what my Reflector shows: private static bool TryStringToNumber(string str, NumberStyles options, ref Number.NumberBuffer number, NumberFormatInfo numfmt, bool parseDecimal);

  if (!Number.ParseNumber(ref chPtr2, options, ref number, numfmt, parseDecimal) 
        || ((((long) ((chPtr2 - chPtr1) / 2)) < str.Length)
        && !Number.TrailingZeros(str, (int) ((long) ((chPtr2 - chPtr1) / 2)))))
            {
                  return false;
            }

private static void StringToNumber(string str, NumberStyles options, ref Number.NumberBuffer number, NumberFormatInfo info, bool parseDecimal);

      if (!Number.ParseNumber(ref chPtr2, options, ref number, info, parseDecimal) 
       || ((((long) ((chPtr2 - chPtr1) / 2)) < str.Length) 
       && !Number.TrailingZeros(str, (int) ((long) ((chPtr2 - chPtr1) / 2)))))
            {
                  throw new FormatException(Environment.GetResourceString("Format_InvalidString"));
            }

"Throughout human history, we have been dependent on machines to survive. Fate, it seems, is not without a sense of irony. " - Morpheus

realJSOP

Of course, if case isn't an issue, you could just do the equivalent of MFC's MakeUpper() or MakeLower() and then run through your switch statement. However, that does not address the issue of language unless you can put the strings into a resource file or something, but then you'd be able to run your switch on integers instead of text. Still, having a switch statement for text is the next best thing to stupid I've seen in a long time.

"Why don't you tie a kerosene-soaked rag around your ankles so the ants won't climb up and eat your candy ass..." - Dale Earnhardt, 1997
-----
"...the staggering layers of obscenity in your statement make it a work of art on so many levels." - Jason Jystad, 10/26/2001

Chris Losinger

John Simmons / outlaw programmer wrote:

unless you can put the strings into a resource file or something, but then you'd be able to run your switch on integers instead of text.

not if, for example, those strings are coming from database records, and there's no numeric equivalent in the record (so you can't just use the number instead of the string), and there's no way you can change the DB schema (because too many things would break). you can do a string compare on each record and generate a numeric code to carry around along with the recordset, to use that in your switch statements. but that just moves the string compare - it doesn't eliminate it. sometimes the best you can do is Make It Work.

image processing toolkits | batch image processing | blogging

jpg 0

Since strings in .NET are unicode, the string "Hello" is basically a byte[10] array with fixed, compile-time known value. This is not a language issue or compiler issue at all because instead of 'case 0: case 1: case 234:', you have a string representative of the underlying values, which is much more readable. In terms of case-sensitive and culture issue, it really depends on where the given string comes from and what does it represent. For instance, in a localized app, regardless of the culture, you will need a culture-unspecific id to identify a control/component. That id could be a GUID, an integer, or a string. From your code sample, I will presume that the given 'myString' is properly normalized with correct cap and trimming. If this isn't the case, then it is the programmer's problem, never the tool.

Tomas Petricek

It is very difficult to find the ballance between enforcing good practices and giving developers more control. The example you posted can be dangerous, but it is perfectly correct in certain situations (as Rocky Moore mentioned). Interesting extension to the swith control structure are Active patterns available in F#: http://blogs.msdn.com/dsyme/archive/2006/08/16/ActivePatterns.aspx[^]. In simple words, active patterns make it is possible to write patterns (rules) that are executed when "case" statement is reached. This rule can than decide whether case statement should be executed. Of course this can be misused, but I think that it would be interesting extension to main-stream programming languages.

// You could then use something like this..:
switch(str)
{
case CaseSensitiveMatch("something"):
... break;
case CaseInsensitiveMatch("SomeThinG"):
... break;
case CaseInsensitiveContains("thing"):
... break;
}

Actually working with strings directly is always potential source of problems (with localization), but I think that active patterns could be useful in many other sitations.

Homepage: TomasP.net | Photo of the month: Calendar | C# and LINQ, F#, Phalanger: My Blog
Latest article: Phalanger, PHP for .NET: Introduction for .NET developers

Ed Poore

Hmm, must have been mistaken then, where did I see that code then... :suss:

tgrt

I remember seeing it in Reflector too, so don't feel bad. Maybe it changed with SP1?

Stuart Dootson

And then the "quick job" gets used a lot and built upon and (before you know it) that quick hack is the corner stone of a dreadful big monolithic application...

Paul Watson

I first heard about it from Joel[^].

regards, Paul Watson Ireland & South Africa

Shog9 wrote:

And with that, Paul closed his browser, sipped his herbal tea, fixed the flower in his hair, and smiled brightly at the multitude of cute, furry animals flocking around the grassy hillside where he sat coding Ruby on his Mac...

PIEBALDconsult

Switching on strings is a good feature, but to be used sparingly. Many years ago I had need of reading in some reports and processing the data, having only Vax-C as a tool. I found that because a character constant could actually hold two characters I could switch on the first two characters of each line, something like:

FILE* file ;
char* buffer ;
char firsttwocharacters ;

...

readaline ( file , buffer ) ;

firsttwocharacters = *buffer ;

switch ( firsttwocharacters )
{
case 'AB' : ...
case 'CD' : ...
...
}
...

And I don't use TryParse, I've read articles like this one: http://www.codeproject.com/dotnet/ExceptionPerformance.asp[^]

Paul Brower

First thing one learns in a Test Driven Development environment is: switch statements = bad bad bad.

Dan Neely

I don't have VS05, what's inside Number.ParseNumber?

-- Rules of thumb should not be taken for the whole hand.

Paul Watson

Well obviously that shouldn't happen. People just need to use their heads :)

regards, Paul Watson Ireland & South Africa

Shog9 wrote:

And with that, Paul closed his browser, sipped his herbal tea, fixed the flower in his hair, and smiled brightly at the multitude of cute, furry animals flocking around the grassy hillside where he sat coding Ruby on his Mac...

Chris Maunder

Paul Watson wrote:

premature optimisation

You can get medication for that these days.

cheers, Chris Maunder

CodeProject.com : C++ MVP

Paul Watson

Sadly it can make you go blind.

regards, Paul Watson Ireland & South Africa

Shog9 wrote:

And with that, Paul closed his browser, sipped his herbal tea, fixed the flower in his hair, and smiled brightly at the multitude of cute, furry animals flocking around the grassy hillside where he sat coding Ruby on his Mac...

DavidNohejl

You don't need VS2005, but Reflector[^] and .NET framework assemblies, obviously. Well, in Number.ParseNumber is lot of code, but no exception throwing.

"Throughout human history, we have been dependent on machines to survive. Fate, it seems, is not without a sense of irony. " - Morpheus

Dan Neely

True, but since I'm only doing 1.1 dev work, I don't have, or need, the 2.0 framework installed.

-- Rules of thumb should not be taken for the whole hand.