|
Thread Rules
1. This is not a "do my homework for me" thread. If you have specific questions, ask, but don't post an assignment or homework problem and expect an exact solution.
2. No recruiting for your cockamamie projects (you won't replace facebook with 3 dudes you found on the internet and $20)
3. If you can't articulate why a language is bad, don't start slinging shit about it. Just remember that nothing is worse than making CSS IE6 compatible.
4. Use [code] tags to format code blocks. |
On March 15 2013 11:18 Orome wrote:Hey guys, small C++ question that I hope someone can help me with. Assuming we have a std::stack called someStack consisting of SomeClass objects, why does this sort of code work fine? SomeClass *next = someStack.top();
stack.pop();
next.printWhatever(); Since next is a pointer variable pointing to someStack.top(), shouldn't it point to a null object once that object is destroyed through pop?
Pop does not destroy the object, it simply de-references it from the stack (otherwise, using a stack as a container wouldn't be very useful would it!). Since next* contains the address to the object, the garbage collector has no reason to clean out that object from memory.
|
On March 15 2013 11:18 Orome wrote:Hey guys, small C++ question that I hope someone can help me with. Assuming we have a std::stack called someStack consisting of SomeClass objects, why does this sort of code work fine? SomeClass *next = someStack.top();
stack.pop();
next.printWhatever(); Since next is a pointer variable pointing to someStack.top(), shouldn't it point to a null object once that object is destroyed through pop?
i don't think pop() invokes the destructor of top() when called.
pop_back (same as pop()):
"Removes the last element of the container.
No iterators or references are invalidated."
|
On March 15 2013 06:49 Morfildur wrote:Show nested quote +On March 15 2013 03:53 alwinuz wrote:
I don't think speed matters, it's more a question of what is more readable and maintainable code.
Write easy code first, optimize for speed later (and measure first!).
But is this more readable than the original code? Not sure... personal preference I think. I think it's definitely more readable and maintainable. There is just one place you have to look at to know where every level begins and ends, you don't have to go through each "else if" to find all the values. If you need to adjust it you instantly know where you have to do it, no need to search through potentially dozens of cases. If you have to add a case, no need to search through all branches to know where it belongs, you can see it in the short list. Sure, in this case it's a small amount of branches but i had a similar situation with price groups where certain video lengths resulted in different prices and there were twice as many potential values and the previous "if/else if" stuff was far too long. After i changed it to something similar to my code above (in PHP), it was so easy that i was able to copy-paste the list into an email to the product manager and he understood it easily. I could probably also have written it like this, which is easier to the untrained eye:
var boundaries = new Array();
boundaries["39"] = "";
boundaries["49"] = "third level";
...
Hey I'm with you  I write code like this all the time. But it can look difficult for someone who is unfamiliar with this type of coding. Clever code is not always the easiest code to read, so at least a comment would be great for my coworkers.
On March 15 2013 04:17 RoyGBiv_13 wrote:Show nested quote +On March 15 2013 03:53 alwinuz wrote:
I don't think speed matters, it's more a question of what is more readable and maintainable code.
Write easy code first, optimize for speed later (and measure first!).
But is this more readable than the original code? Not sure... personal preference I think. I don't really want to start a big huff over nothing, but there is a misconception I don't want spread. The idea that an easily maintainable design is a trade off with with code speed is absurd. The better the design, the smaller the code base, more maintainable, readable, and faster it will be. If you find that your code is running slow, then optimizing it will only cause problems in your understanding of the code. This is often why people rewrite programs from scratch, because they tried to optimize something, realized they did the design wrong, and start over. Optimization should be done out the gate, so that your design pattern will reflect the quickest way to solve the problem, and be easier to understand because of it.
You don't want to start a big huf, then what did you just do? 
I'll just dump a relevant story: http://www.codinghorror.com/blog/2009/01/the-sad-tragedy-of-micro-optimization-theater.html
And you give the answer yourself: there's a difference between micro-optimization at code level, and optimizing the design of the software as a whole. + what Tobberoth said
On March 15 2013 06:44 Blisse wrote:Show nested quote +On March 15 2013 06:37 Tobberoth wrote:On March 15 2013 06:34 Blisse wrote:On March 14 2013 23:25 FFGenerations wrote:Using javascript & html, i have a button on a page and a text box. User inputs a grade number into the text box and presses the button. An alert pops up based on the input, using a simple if statement, telling the user if the entered grade is a merit, pass etc. + Show Spoiler +
if(grade!="")
{
if(grade>-1&&grade<40)
{
alert("This is below threshold.")
}
else if(grade>39&&grade<50)
{
alert("This is a third class degree.")
}
else if(grade>49&&grade<60)
{
alert("This is a lower second degree.")
}
else if(grade>59&&grade<70)
{
alert("This is an upper second degree.")
}
else if(grade>69&&grade<101)
{
alert("This is a first class degree.")
}
else
{
alert("Please enter a grade from 0 to 100.")
}
}
else
{
alert("Please enter a grade from 0 to 100.")
}
}
The question is, why do i need to lay out everything within a second if statement? The first time i wrote it, i had similar to the following: + Show Spoiler +
if(grade="")
{
alert("You need to enter a grade")
}
else if(grade>-1&&grade<40)
{
alert("This is below threshold.")
}
else if(grade>39&&grade<50)
{
alert("This is a third class degree.")
}
else if(grade>49&&grade<60)
{
alert("This is a lower second degree.")
}
else if(grade>59&&grade<70)
{
alert("This is an upper second degree.")
}
else if(grade>69&&grade<101)
{
alert("This is a first class degree.")
}
else
{
alert("Please enter a grade from 0 to 100.")
}
however, this kept displaying "This is below threshold." instead of running the if(grade="") part. My tutor said it might be something to do with the if statements switching from string ("") to int (grade>-1&&grade<40) which is somehow messing it up. Many thanks Your code seems to be able to compare integers and strings without regard for the data type. I think you have to apply a parseInt(grade) after you've validated the grade is a number, before you can perform all these checks. That's what I think I did when I did JS some years ago. There's no problem comparing strings and numbers in javascript. Try something like: alert("20" > 5) and you'll get true. Because "20" as an ASCII value should be greater than 5, at least it would make sense if it were. "4" > 5 doesn't seem like it would return true, but it probably does here. Again, not a JS person. And anyways, I don't believe you're suggesting one should compare strings with integers like that. Best practice should always be string with string and int with int.
"4" > 5 actually returns false! (just tested) I didn't know that because I always follow your best practice.
On March 14 2013 23:25 FFGenerations wrote:Using javascript & html, i have a button on a page and a text box. User inputs a grade number into the text box and presses the button. An alert pops up based on the input, using a simple if statement, telling the user if the entered grade is a merit, pass etc. + Show Spoiler +
if(grade!="")
{
if(grade>-1&&grade<40)
{
alert("This is below threshold.")
}
else if(grade>39&&grade<50)
{
alert("This is a third class degree.")
}
else if(grade>49&&grade<60)
{
alert("This is a lower second degree.")
}
else if(grade>59&&grade<70)
{
alert("This is an upper second degree.")
}
else if(grade>69&&grade<101)
{
alert("This is a first class degree.")
}
else
{
alert("Please enter a grade from 0 to 100.")
}
}
else
{
alert("Please enter a grade from 0 to 100.")
}
}
The question is, why do i need to lay out everything within a second if statement? The first time i wrote it, i had similar to the following: + Show Spoiler +
if(grade="")
{
alert("You need to enter a grade")
}
else if(grade>-1&&grade<40)
{
alert("This is below threshold.")
}
else if(grade>39&&grade<50)
{
alert("This is a third class degree.")
}
else if(grade>49&&grade<60)
{
alert("This is a lower second degree.")
}
else if(grade>59&&grade<70)
{
alert("This is an upper second degree.")
}
else if(grade>69&&grade<101)
{
alert("This is a first class degree.")
}
else
{
alert("Please enter a grade from 0 to 100.")
}
however, this kept displaying "This is below threshold." instead of running the if(grade="") part. My tutor said it might be something to do with the if statements switching from string ("") to int (grade>-1&&grade<40) which is somehow messing it up. Many thanks
Not sure if you know this already, but have you tried debugging? In Google Chrome press F12, or use Firebug addon in Firefox. You can set a breakpoint, step through your code, inspect the value of 'grade' etc. Sorry no actual answer to your question, but maybe you can find the answer yourself!
|
What are you storing in your std::stack? Pointers(stack<SomeClass*>) or Objects(stack<SomeClass>)?
Judging from the way you use you wrote the code, it's pointers. Remember that pointers are merely addresses, and that they point to some location in memory. If you created the object using new (ie. on the heap), it won't be destroyed until you delete the pointer. stack.pop() simply removes the topmost element in the stack. If it's of this type: stack<SomeClass>, then it'll call the destructor, else for stack<SomeClass*> it simply removes the topmost address
|
Damn that was quick, thanks to everyone for pitching in. ddengster nailed what I was confused about, I really was storing pointers to the objects, not the objects themselves. Makes perfect sense now.
|
On March 15 2013 11:57 Orome wrote:Damn that was quick, thanks to everyone for pitching in. ddengster nailed what I was confused about, I really was storing pointers to the objects, not the objects themselves. Makes perfect sense now.
Even if it did get destroyed upon popping, pointers don't just reassign themselves when the thing they point to is deallocated, and the memory doesn't just disappear. I would expect that the function call would still usually work even if the object was destroyed immediately before.
|
What's even more sadder that I have read the article and other people did as well and there are walls of comments, that there is even discussion about it, when this should be obvious to everyone with a working brain ;;
|
On March 15 2013 11:48 alwinuz wrote:Show nested quote +On March 15 2013 04:17 RoyGBiv_13 wrote:On March 15 2013 03:53 alwinuz wrote:
I don't think speed matters, it's more a question of what is more readable and maintainable code.
Write easy code first, optimize for speed later (and measure first!).
But is this more readable than the original code? Not sure... personal preference I think. I don't really want to start a big huff over nothing, but there is a misconception I don't want spread. The idea that an easily maintainable design is a trade off with with code speed is absurd. The better the design, the smaller the code base, more maintainable, readable, and faster it will be. If you find that your code is running slow, then optimizing it will only cause problems in your understanding of the code. This is often why people rewrite programs from scratch, because they tried to optimize something, realized they did the design wrong, and start over. Optimization should be done out the gate, so that your design pattern will reflect the quickest way to solve the problem, and be easier to understand because of it. You don't want to start a big huf, then what did you just do? I'll just dump a relevant story: http://www.codinghorror.com/blog/2009/01/the-sad-tragedy-of-micro-optimization-theater.htmlAnd you give the answer yourself: there's a difference between micro-optimization at code level, and optimizing the design of the software as a whole. + what Tobberoth said Show nested quote +On March 15 2013 06:44 Blisse wrote:On March 15 2013 06:37 Tobberoth wrote:On March 15 2013 06:34 Blisse wrote:On March 14 2013 23:25 FFGenerations wrote:Using javascript & html, i have a button on a page and a text box. User inputs a grade number into the text box and presses the button. An alert pops up based on the input, using a simple if statement, telling the user if the entered grade is a merit, pass etc. + Show Spoiler +
if(grade!="")
{
if(grade>-1&&grade<40)
{
alert("This is below threshold.")
}
else if(grade>39&&grade<50)
{
alert("This is a third class degree.")
}
else if(grade>49&&grade<60)
{
alert("This is a lower second degree.")
}
else if(grade>59&&grade<70)
{
alert("This is an upper second degree.")
}
else if(grade>69&&grade<101)
{
alert("This is a first class degree.")
}
else
{
alert("Please enter a grade from 0 to 100.")
}
}
else
{
alert("Please enter a grade from 0 to 100.")
}
}
The question is, why do i need to lay out everything within a second if statement? The first time i wrote it, i had similar to the following: + Show Spoiler +
if(grade="")
{
alert("You need to enter a grade")
}
else if(grade>-1&&grade<40)
{
alert("This is below threshold.")
}
else if(grade>39&&grade<50)
{
alert("This is a third class degree.")
}
else if(grade>49&&grade<60)
{
alert("This is a lower second degree.")
}
else if(grade>59&&grade<70)
{
alert("This is an upper second degree.")
}
else if(grade>69&&grade<101)
{
alert("This is a first class degree.")
}
else
{
alert("Please enter a grade from 0 to 100.")
}
however, this kept displaying "This is below threshold." instead of running the if(grade="") part. My tutor said it might be something to do with the if statements switching from string ("") to int (grade>-1&&grade<40) which is somehow messing it up. Many thanks Your code seems to be able to compare integers and strings without regard for the data type. I think you have to apply a parseInt(grade) after you've validated the grade is a number, before you can perform all these checks. That's what I think I did when I did JS some years ago. There's no problem comparing strings and numbers in javascript. Try something like: alert("20" > 5) and you'll get true. Because "20" as an ASCII value should be greater than 5, at least it would make sense if it were. "4" > 5 doesn't seem like it would return true, but it probably does here. Again, not a JS person. And anyways, I don't believe you're suggesting one should compare strings with integers like that. Best practice should always be string with string and int with int. "4" > 5 actually returns false! (just tested) I didn't know that because I always follow your best practice.
JS, what are you high on. Cool to know, but -.-
|
JS is just very loose with the types, if you compare a string to an int the string will get converted to a number.
|
Types in JS are a bloody nightmare.
typeof(NaN) => "Number"
Yay Javascript.
|
On March 15 2013 06:44 Blisse wrote:Show nested quote +On March 15 2013 06:37 Tobberoth wrote:On March 15 2013 06:34 Blisse wrote:On March 14 2013 23:25 FFGenerations wrote:Using javascript & html, i have a button on a page and a text box. User inputs a grade number into the text box and presses the button. An alert pops up based on the input, using a simple if statement, telling the user if the entered grade is a merit, pass etc. + Show Spoiler +
if(grade!="")
{
if(grade>-1&&grade<40)
{
alert("This is below threshold.")
}
else if(grade>39&&grade<50)
{
alert("This is a third class degree.")
}
else if(grade>49&&grade<60)
{
alert("This is a lower second degree.")
}
else if(grade>59&&grade<70)
{
alert("This is an upper second degree.")
}
else if(grade>69&&grade<101)
{
alert("This is a first class degree.")
}
else
{
alert("Please enter a grade from 0 to 100.")
}
}
else
{
alert("Please enter a grade from 0 to 100.")
}
}
The question is, why do i need to lay out everything within a second if statement? The first time i wrote it, i had similar to the following: + Show Spoiler +
if(grade="")
{
alert("You need to enter a grade")
}
else if(grade>-1&&grade<40)
{
alert("This is below threshold.")
}
else if(grade>39&&grade<50)
{
alert("This is a third class degree.")
}
else if(grade>49&&grade<60)
{
alert("This is a lower second degree.")
}
else if(grade>59&&grade<70)
{
alert("This is an upper second degree.")
}
else if(grade>69&&grade<101)
{
alert("This is a first class degree.")
}
else
{
alert("Please enter a grade from 0 to 100.")
}
however, this kept displaying "This is below threshold." instead of running the if(grade="") part. My tutor said it might be something to do with the if statements switching from string ("") to int (grade>-1&&grade<40) which is somehow messing it up. Many thanks Your code seems to be able to compare integers and strings without regard for the data type. I think you have to apply a parseInt(grade) after you've validated the grade is a number, before you can perform all these checks. That's what I think I did when I did JS some years ago. There's no problem comparing strings and numbers in javascript. Try something like: alert("20" > 5) and you'll get true. Because "20" as an ASCII value should be greater than 5, at least it would make sense if it were. "4" > 5 doesn't seem like it would return true, but it probably does here. Again, not a JS person. And anyways, I don't believe you're suggesting one should compare strings with integers like that. Best practice should always be string with string and int with int.
Best practice is one thing, the mechanics of a language something else. JS is very weakly and dynamically typed. If you see this as a weakness of the language, of course you should avoid comparision between types, but the language does support it just fine. I find it's less an issue of "You shouldn't compare strings and integers" because it's fine to do so in JS, it's more along the lines of "What's the reason behind trying to compare a string and an integer, why aren't both integers to begin with?" If there's a good reason for one of the inputs being a string, and there's previous validation in place to make sure it's a number, I see no reason to write extra code to convert it when JS doesn't need you to.
|
On March 16 2013 00:26 Tobberoth wrote:Show nested quote +On March 15 2013 06:44 Blisse wrote:On March 15 2013 06:37 Tobberoth wrote:On March 15 2013 06:34 Blisse wrote:On March 14 2013 23:25 FFGenerations wrote:Using javascript & html, i have a button on a page and a text box. User inputs a grade number into the text box and presses the button. An alert pops up based on the input, using a simple if statement, telling the user if the entered grade is a merit, pass etc. + Show Spoiler +
if(grade!="")
{
if(grade>-1&&grade<40)
{
alert("This is below threshold.")
}
else if(grade>39&&grade<50)
{
alert("This is a third class degree.")
}
else if(grade>49&&grade<60)
{
alert("This is a lower second degree.")
}
else if(grade>59&&grade<70)
{
alert("This is an upper second degree.")
}
else if(grade>69&&grade<101)
{
alert("This is a first class degree.")
}
else
{
alert("Please enter a grade from 0 to 100.")
}
}
else
{
alert("Please enter a grade from 0 to 100.")
}
}
The question is, why do i need to lay out everything within a second if statement? The first time i wrote it, i had similar to the following: + Show Spoiler +
if(grade="")
{
alert("You need to enter a grade")
}
else if(grade>-1&&grade<40)
{
alert("This is below threshold.")
}
else if(grade>39&&grade<50)
{
alert("This is a third class degree.")
}
else if(grade>49&&grade<60)
{
alert("This is a lower second degree.")
}
else if(grade>59&&grade<70)
{
alert("This is an upper second degree.")
}
else if(grade>69&&grade<101)
{
alert("This is a first class degree.")
}
else
{
alert("Please enter a grade from 0 to 100.")
}
however, this kept displaying "This is below threshold." instead of running the if(grade="") part. My tutor said it might be something to do with the if statements switching from string ("") to int (grade>-1&&grade<40) which is somehow messing it up. Many thanks Your code seems to be able to compare integers and strings without regard for the data type. I think you have to apply a parseInt(grade) after you've validated the grade is a number, before you can perform all these checks. That's what I think I did when I did JS some years ago. There's no problem comparing strings and numbers in javascript. Try something like: alert("20" > 5) and you'll get true. Because "20" as an ASCII value should be greater than 5, at least it would make sense if it were. "4" > 5 doesn't seem like it would return true, but it probably does here. Again, not a JS person. And anyways, I don't believe you're suggesting one should compare strings with integers like that. Best practice should always be string with string and int with int. Best practice is one thing, the mechanics of a language something else. JS is very weakly and dynamically typed. If you see this as a weakness of the language, of course you should avoid comparision between types, but the language does support it just fine. I find it's less an issue of "You shouldn't compare strings and integers" because it's fine to do so in JS, it's more along the lines of "What's the reason behind trying to compare a string and an integer, why aren't both integers to begin with?" If there's a good reason for one of the inputs being a string, and there's previous validation in place to make sure it's a number, I see no reason to write extra code to convert it when JS doesn't need you to.
If readability/understanding isn't an issue, then your reasoning sounds fine to me.
|
Czech Republic4646 Posts
On March 15 2013 06:49 Morfildur wrote:Show nested quote +On March 15 2013 03:53 alwinuz wrote:
I don't think speed matters, it's more a question of what is more readable and maintainable code.
Write easy code first, optimize for speed later (and measure first!).
But is this more readable than the original code? Not sure... personal preference I think. I think it's definitely more readable and maintainable. There is just one place you have to look at to know where every level begins and ends, you don't have to go through each "else if" to find all the values. If you need to adjust it you instantly know where you have to do it, no need to search through potentially dozens of cases. If you have to add a case, no need to search through all branches to know where it belongs, you can see it in the short list. Sure, in this case it's a small amount of branches but i had a similar situation with price groups where certain video lengths resulted in different prices and there were twice as many potential values and the previous "if/else if" stuff was far too long. After i changed it to something similar to my code above (in PHP), it was so easy that i was able to copy-paste the list into an email to the product manager and he understood it easily. I could probably also have written it like this, which is easier to the untrained eye:
var boundaries = new Array();
boundaries["39"] = "";
boundaries["49"] = "third level";
...
Should be in DB anyway
EDIT: by that I mean I agree with your way of coding it. I was more commenting on your story, that it should have been stored in DB and loaded into the structures like the above. Then easy select to show manager the data, and you can change stuff without deploy. But I am exaggerating a bit, in smaller programs/programs not dependent on DB the trade-off is often not worth it.
|
Czech Republic4646 Posts
On March 16 2013 00:04 AmericanUmlaut wrote:
Types in JS are a bloody nightmare.
typeof(NaN) => "Number"
Yay Javascript.
Yep, JS type design is terrible shit. Empty string is equivalent to false, who knew  Caused me some wasted time yesterday.
|
Anybody here have any experience with Objective-C? I'm trying to do some C work in it (don't ask me why, it's long winded), and my C is a bit rusty. I can't seem to figure out what I'm doing wrong.
Basically, I have a char array and I'm trying to replace characters. Here's an example:
@implementation BlahViewController {
char *_test;
}
- (void)viewDidLoad {
char test[3] = "ab";
_test = test;
NSString *withoutZ = [NSString stringWithCString:_test encoding:NSASCIIStringEncoding];
NSLog(@"%@", withoutZ);
_test[0] = "z";
NSString *withZ = [NSString stringWithCString:_test encoding:NSASCIIStringEncoding];
NSLog(@"%@", withZ);
}
The result is that withoutZ = "ab", but withZ = "hb". Sometimes the first character in withZ is different from one build to the next. I'm sure it's something simple I'm doing wrong. Can anyone help me out?
Edit: Nevermind, I figured it out. I forgot to put the pointer.
_test[0] = *"z";
But now I'm not sure how to implement that in the way I need to. The above was just an example test for me to make sure I understood how it works, but when I try to implement it, it's not working quite right. Here's what I'm trying to do, simplified:
@implementation CSTGameBoard {
char *_boardKey;
char *_charForNumber;
}
- (id)init {
self = [super init];
if (self) {
char boardKey[33] = "00000000000000000000000000000000";
_boardKey = boardKey;
char charForNumber[11] = "0123456789";
_charForNumber = charForNumber;
}
return self;
}
- (void)updateBoardKeyWithInt:(int)number atCoefficient:(int)coefficient {
_boardKey[coefficient] = _charForNumber[number];
}
I'm writing up my own AI for one of my favorite board games, so optimization is paramount. For a given board position, I need to generate a unique board key that represents the current state of the board. I then convert the key to an NSString object that I use as a key in an NSDictionary to see if I've already evaluated this position before.
Because optimization is so important, I figured it would be much more efficient to use a char index rather than using NSString. Allocating new objects hundreds of thousands of times is not very efficient, right? This way I can limit the times I need to allocate a new NSString object to only when I need it for the dictionary reference.
So I use the charForNumber as a reference to replace characters in the boardKey at any given position with a character that represents a number. I don't do any conversions and it's one simple step. But I can't get it to work. The boardKey is fine when it's initialized, but as soon as I start replacing characters using updateBoardKeyWithInt, it goes wacky and spits out gobbley gunk. I tried to put a pointer before the _charForNumber[number], but then the project fails to compile. Any ideas?
|
On March 14 2013 06:07 supereddie wrote:
Unit tests shouldn't be dependant on eachother - problem solved.
Theoretically yes. But with web pages it's sometimes unavoidable. You can't test stuff in a website if your login fails, or if the server goes down. Test suite runs on a VM for hours and sometime days and all your test fails, but it tells you nothing. Obviously can just let it finish but it's just a waste of resources / time (since email notification is not sent until all tests are completed).
Also it's not always realistic or practical to have complete dependency for integration tests. If you want to test delete an order, then needs to be an order there for you to delete. While you can put the create order code inside delete order test, but in real life it's much easier just have the delete test dependent on create test.
|
On March 16 2013 02:16 ragz_gt wrote:Show nested quote +On March 14 2013 06:07 supereddie wrote:
Unit tests shouldn't be dependant on eachother - problem solved. Theoretically yes. But with web pages it's sometimes unavoidable. You can't test stuff in a website if your login fails, or if the server goes down. Test suite runs on a VM for hours and sometime days and all your test fails, but it tells you nothing. Obviously can just let it finish but it's just a waste of resources / time (since email notification is not sent until all tests are completed). Also it's not always realistic or practical to have complete dependency for integration tests. If you want to test delete an order, then needs to be an order there for you to delete. While you can put the create order code inside delete order test, but in real life it's much easier just have the delete test dependent on create test.
You don't need to log in to be able to do a unit test. You don't need a server up for a unit test. That is the entire point of a unit test: you're testing a very small part of functionality of one class. If you've modelled your application correctly, it shouldn't be any problem to have lots of unit tests.
If needed you can mock/stub various dependencies of the class you're testing.
I can perfectly test a delete without a database; usually there will be a lot of functionality before the actual delete (like validation etc.), which can all be done by unit tests. For the actual deletion you can just use a test-database that you will be re-populating / restoring every time the test runs.
You should also test what happens when you try to delete an order if there is none in the datase. Or if you pass an invalid id.
Remember I'm talking about unit test, not integration tests. Unit tests should be stand-alone, independant tests.
|
I get the impression he is not talking about unit tests but (UI) functional tests implemented in JUnit. (It is a different matter that he should have a thin UI layer and rely on API testing instead of UI testing.)
Unfortunately functional tests often require special enviroment and application state. That means a programmer has three options when it comes to functional tests:
- Build his application from the start with startup time in mind. Hot config only, no expensive initialization, etc. This is not always achievable due to external components, compatibility reasons, size of data, or simply because hot config would make the application too complex.
- Group his tests together that require the same kind of application state. Contexts should set up and tear down this state. Tests should restore this state after their run. Unfortunately sometimes they don't and thus can cause test interference, both within and outside the same group.
- Just ignore the entire issue, let his tests run forever and watch the astronomical costs.
|
On March 16 2013 04:08 Frigo wrote:
I get the impression he is not talking about unit tests but (UI) functional tests implemented in JUnit. (It is a different matter that he should have a thin UI layer and rely on API testing instead of UI testing.)
Unfortunately functional tests often require special enviroment and application state. That means a programmer has three options when it comes to functional tests:
- Build his application from the start with startup time in mind. Hot config only, no expensive initialization, etc. This is not always achievable due to external components, compatibility reasons, size of data, or simply because hot config would make the application too complex.
- Group his tests together that require the same kind of application state. Contexts should set up and tear down this state. Tests should restore this state after their run. Unfortunately sometimes they don't and thus can cause test interference, both within and outside the same group.
- Just ignore the entire issue, let his tests run forever and watch the astronomical costs.
I think functional tests are supposed to be independent too but in my experience they are not. Functional testing deletes in a database without a database does not mimic live environments. So in this sense you are right.
So you get tests that depend on each other.
I do want to note that the email at the end can still be sent though. You just have to make it send its results at a time when you think your functional tests would have already run.
|
On March 16 2013 01:46 mcc wrote:Show nested quote +On March 16 2013 00:04 AmericanUmlaut wrote:
Types in JS are a bloody nightmare.
typeof(NaN) => "Number"
Yay Javascript. Yep, JS type design is terrible shit. Empty string is equivalent to false, who knew Caused me some wasted time yesterday.
Empty string does not equal false, it is merely a falsy value. Falsy values are very useful, and greatly increase the expressiveness of the language imo. I've programmed in JS a lot, and I can tell you that if you're actually checking if a variable equals or does not equal a falsy value, you are generally not doing the right thing anyway. But just in case, I'd recommend running your code through http://jshint.com (there's a command-line version as well if you'd like to lint through your editor/console) and it'll catch cases where you compare against falsy values without using !== or ===.
JS has some weird idiosyncracies, but its type system isn't one of them and makes for a very expressive and useful language.
|
|
|
|
|
|
EPT
