Code review - hilarious code snippets

[doodab]

unhandled exception

Oh for ****s flippin' sake. <mod. you meant that didn't you?>

I'm not suggesting you don't handle the ******* exception. I'm comparing try catch, with an exception raised when an enclosed method fails, to calling a method and checking the result to see if it worked or not.

Anyway as i said, semantics is key. If the method is returning null because of an error it should be raising an exception. This may well need to be propogated, and propgating a null value up the call stack is error prone and you lose context so you can't write a stack trace etc. If it's returning null because the answer is null then return null.

[doodab]

Exceptions should only be used for situations that you cannot foresee i.e. A external resource not being available, not for an empty search result from a query, the clue is in the name.

Exactly. But the reverse is that returning null should only be used to indicate a null result, not to indicate that everything is ****ed because of something that happened or didn't happen 300 lines of code back when you were parseing a configuration file.

[eek]

Exceptions should only be used for situations that you cannot foresee i.e. A external resource not being available, not for an empty search result from a query, the clue is in the name.

Russell, I was simply querying Suity's (in)ability to read rather than the question itself. suity did his usual job of half reading something and making a comment based on his assumption (in this case that doodab suggested a try block with no exception handling).

[russell]

Exactly. But the reverse is that returning null should only be used to indicate a null result, not to indicate that everything is ****ed because of something that happened or didn't happen 300 lines of code back when you were parseing a configuration file.

Yeah that is true. If you come from a language that has no exceptions then returning null rather than a pointer for eg is valid, then let the caller assert that or handle it.

[Sysman]

If it's a valid result then fair enough but if you can't initialize an object properly when the constructor is called throw an exception in the constructor. Don't return normally then expect people to call isValid() on your object, check the result of every method invocation to see if it's null, or even worse just throw unchecked exceptions all over the shop because the object was never initialised properly.

Especially don't do the latter two when it's even more of your tulipty code that is calling methods on the improperly constructed object.

That reminds me. Why do so many C/C++ programmers never bother checking error returns?

[TheFaQQer]

And again in English

Well, eek said

I don't think anyone has suggested using a try block with no exception (until you arrived).

You said

Post 44, paragraph 1, sentence 2, clause 1.

HTH BIDI

Looking at Post 44, paragraph 1, sentence 2, clause 1, it says

Testing for a null result is more expensive than a try block if no exception is raised

That isn't having an exception without a try block, it's talking about not including a try block, but having to test afterwards whether the result is null or not.

Until you posted in the thread, no-one suggested having a try block without an exception. Maybe you should stick to Oracle?

[russell]

That reminds me. Why do so many C/C++ programmers never bother checking error returns?

Because the're bad coders and they write buggy, flaky software?

With discipline and using tools like valgrind you can write robust C code that runs for years.

[Sysman]

I can't remember the last time any of my apps crashed and took down the system... my C++ apps crash just the same way as my Java apps crash Don't MacOSX, Linux and modern versions of Windows protect against this?

Not so sure about that. In an early attempt to do something with Objective-C strings I created a memory leak in a tight processig loop. That brought the machine to its knees - the swapfile hit 3GB on a 640 MB system, and was growing

A reboot was the only way out.

[suityou01]

Well, eek said


You said


Looking at Post 44, paragraph 1, sentence 2, clause 1, it says

That isn't having an exception without a try block, it's talking about not including a try block, but having to test afterwards whether the result is null or not.

Until you posted in the thread, no-one suggested having a try block without an exception. Maybe you should stick to Oracle?

Absolutely. The original statement was about having a try block without raising an exception.

[Sysman]

Was reviewing some code the other day with the main developer for the site in question. In a comment written a couple of years ago:

/* Here be dragons, proceed with caution */

No clue as to exactly why the dragons existed there or what we should be cautious about

I've often wondered about that one myself. Apparently to do with ancient maps:

origin of phrase: Here be dragons; or Here there be dragons