|
The problem with the simple example is that the catch could blow up throwing you out of your process completely. (The passed "string" is null. The Info function should throw because it was passed null...) What if the string is a few thousand characters? I wouldn't use data passed to the process in the message either.
Probably something like
Log.Info("Using default value for \"ConvertToInt\" process because of msg:" + e.Message); That isolates what caused the problem from the log process. And may actually tell the user what went wrong and why.
Your user guide would, of course , tell the user what ConvertToInt does, how they might encounter it and what the default value would be.
I can certainly understand the orig. poster's frustration, he has a friend who's being thick-headed in one area. He used a humorous way to illistrate that no message could be nearly as damaging as intentionally sending a misleading message.
|
|
|
|
|
1. A programming language that would even let you do this is broken. I feel so unclean just having read this.
2. Type that into the codebase, and it's only a matter of time before you are the victim; either of forgetting to take it out, or having it used against you.
3. Pranks. Oy! I've worked in organizations where pranks got to be the accepted norm. They are sick and unprofessional organizations. Generally what happens is somebody goes too far. Maybe there is an injury, or a customer gets pranked by accident. Then there is a short, grim meeting where your boss or your division manager or somebody from Personnel lays it on the line saying, "Next prank gets you fired."
A better approach is to make the dumb behavior (not logging) an issue with your colleague, and if he doesn't make an attempt to improve, with your team or your manager. Because your colleague can certainly make the pranking an issue with your manager. Then he will have the moral high ground and you will lose traction on getting the undesirable behavior fixed. If you act unprofessionally because your teammate did, that fixes nothing.
|
|
|
|
|
Quote: "A programming language that would even let you do this is broken." -SeattleC++
.
Have you ever actually used C++, at all? I'm guessing no, otherwise you'd realize that it's quite capable of doing truly horrible things...
|
|
|
|
|
Joe_Dert wrote: Have you ever actually used C++, at all? I'm guessing no, otherwise you'd realize that it's quite capable of doing truly horrible things...
This is hardly to the point. And yeah, I've used C++. Since you were in middle school, I'm guessing.
It may well be true that it's a poor workman who blames his tools. But it's also true that it's a deranged colleague who thinks up ways to deliberately damage shared tools and then checks their perverse thoughts in. And then brags about it.
|
|
|
|
|
You're dodging the issue, he did something devious, yes, but, that wasn't my point.
Pay attention, you said that any language that allowed him to do what he did was broken, to which I responded(more or less), that C++ is far worse in that regard.
(ie: By your reasoning, C++ is broken, because it has many mechanisms which can be abused to do devious things.)
Furthermore, I'd suggest that you take some of your "assumed" wisdom of years, and at least try act the part of the wise older man if you want to play the: "I've been coding since you were in diapers" card.
Let's be frank, I had quoted you, so, assuming you speak English, you know what I meant, so, either you're a troll, illiterate, or perhaps just senile?
Best case scenario, I'm willing to conclude there was some sort of misunderstanding.
modified 22-Oct-12 2:44am.
|
|
|
|
|
I'm going to have to join the crowd who say that: yes, this is funny, and devious, but if that made it to the source control for an actual release then you're a bad developer too.
Also I'm pretty sure you'll get a compiler warning for hiding inherited method GetType which should be a bit of a giveaway.
And honestly if it was only 15 minutes that you wasted tracking it down then it's not worth getting annoyed by.
|
|
|
|
|
BobJanova wrote: but if that made it to the source control for an actual release then you're a bad developer too
No, he and I have a branch especially made for stuff that's either highly unstable, proofs of concept or simply little experiments.
BobJanova wrote: Also I'm pretty sure you'll get a compiler warning for hiding inherited method GetType which should be a bit of a giveaway.
My little shenanigan wasn't meant to stand anything but superficial scrutiny anyway, which I mentioned in the original post. Any debug through the code would've revealed it immediately.
BobJanova wrote: And honestly if it was only 15 minutes that you wasted tracking it down then it's not worth getting annoyed by.
Yes, I did get pretty carried away. I don't normally do this, nor do I intend to do it anytime soon, but this time I slipped.
Full-fledged Java/.NET lover, full-fledged PHP hater.
Full-fledged Google/Microsoft lover, full-fledged Apple hater.
Full-fledged Skype lover, full-fledged YM hater.
|
|
|
|
|
Although i love to play pranks, i consider that if this have the remote possibility to reach production, then you're fried (and fired) and as a matter of fact, if your buddy doesn't log errors, i think very unlikely that he sees the log.
|
|
|
|
|
Yes, both your points are valid, and I already replied to them in other posts above.
Full-fledged Java/.NET lover, full-fledged PHP hater.
Full-fledged Google/Microsoft lover, full-fledged Apple hater.
Full-fledged Skype lover, full-fledged YM hater.
|
|
|
|
|
maybe iam missing something, but if you log e.ToString(), then you have both class name and method name plus line of the code the exception happend.In my helper i just log e.ToString.Any reason?
Behzad
|
|
|
|
|
No, we are logging class name, method name, and exception message. We also sometimes log stack traces for errors that tend to get funky, but usually just logging the exception message is enough for us.
Codebases are not that great (we do lots of small applications instead of a large one that does everything), and it's pretty easy to figure what went wrong and why, without explicitly needing a StackTrace for it. That's exactly why it's optional, for cases where it's not immediately clear what goes wrong. As you said, we could log the whole error, but for most cases it would be overkill.
Full-fledged Java/.NET lover, full-fledged PHP hater.
Full-fledged Google/Microsoft lover, full-fledged Apple hater.
Full-fledged Skype lover, full-fledged YM hater.
|
|
|
|
|
Andrei Straut wrote: That should teach them to log stuff properly, or use the debugger
Unless of course it is you that encounters it 6 months from now after you forgot about it.
|
|
|
|
|
I suppose there is the possibility that a potentially exception throwing routine simply must be called in a destructor, and that logging or otherwise reporting the exception is not acceptable because allocating the necessary resources is potentailly unsafe, in which case your colleagues code would be quite acceptable. One might find a similar argument in a C++ library with an extern "C" interface.
Reporting or logging errors isn't always an option. Sometimes you just have to use other methods. Besides, further compounding the problem by obfuscating the errors is a suspect solution. Did you push your changes?
|
|
|
|
|
Time to point out that this is C# code.
|
|
|
|
|
...you mean to tell me that C# won't run on my doorbell?
|
|
|
|
|
I'd like to go a bit further, as a rule of thumb everytime time you go into the more general exception (as in your case) you MUST quit the application, of course you can log it but it's mandatory to quit if your program stayed in a compromised situation, otherwise you are able to inform your users about the problem (which should be very specific or maybe catch a more specific exception and solve it).
Many programmers do what your colleague did because they think it's not professional to quit that abruptly, wrong!, it's not professional not to accept the problem.
Carlos Alberto
|
|
|
|
|
Hah! You think that's bad?! Some time ago, I inherited some code which had about 70 data classes, each containing three or four methods along the lines of:
public override void CreateEntity()
{
try
{
base._database.ExecuteNonQuery(storedProcCommandWrapper);
int parameterValue = (int) storedProcCommandWrapper.GetParameterValue("@RETURN_VALUE");
if (parameterValue == 0)
{
throw new Exception("Create failed.");
}
}
catch (Exception exception)
{
exception.Message.ToString();
throw new Exception("Exception occured", exception);
}
}
Needless to say, all of the code which called these methods looked like:
try
{
customer.CreateEntity();
}
catch (Exception ex)
{
MessageBox.Show(ex.Message);
}
No logging. No access to the inner exception. No way to diagnose why one button was displaying the really helpful "Exception occured" error every time the user clicked it.
"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer
|
|
|
|
|
If this is to teach him lesson then I agree with you, considering the fun.
But for deployment, it may cause serious problem.
|
|
|
|
|
Once, as a prank, a guy put an ASP.NET error message as data in the database (with the HTML formatting). Then he told another developer to look into a problem where in a HTML table on of the cells would display the ASP error message.
|
|
|
|
|
I could see him getting spoken to to correct his coding and you looking to explain to a new employer why you got sacked for sabataging the codebase.
|
|
|
|
|
FAIL.
Sorry to be a party pooper, but your evil intentions would not be fulfilled by that. You haven't overridden the GetType method. You've hidden it. And that means the compiler will only bind to the implementation if invoked on a reference of type TheClass. And it's unlikely anyone would call GetType on the thing if it's already declared to be TheClass (although the run-time type of course could be a descendant).
So, since GetType isn't virtual it'd still return "TheClass" and your colleague would go to the code and see that you've attempted to fool him. (If you guys have decent source control anyway.)
I do sympathize with your sentiments though. But frankly it is often the project managers who are at fault, because rather than foster a culture where mistakes are openly discussed, not for the purpose of punishment but for education, they tend to reward people who "are positive". In my experience that's often people who aren't competent enough to see the myriad things that are usually F-ed up in... well, any codebase I've come across.
|
|
|
|
|
Andrei Straut wrote: Yes, I'm a devious SOB, but hey, if they don't care, why should I?
If you didnt, then you and that person do not have/make/create any difference. Its just like watching a crime happening and ignoring it than atleast reporting.
|
|
|
|
|
hahaha lol... clever idea
============================================
The grass is always greener on the other side of the fence
|
|
|
|
|
So my boss is like a rabbit. He just loves his greens.
Unclear naming should stay unclear by commenting it (helper was never used, but that's beside the point).
Dim helper As Boolean Of course we could make clear comments in favor of properly naming our variables.
s.DoSomething()
p.DoSomethingElse The obvious also needs extra explanation.
i = nX + nY In case we ever want to compile our code into a beginners how-to-program book.
i = 10 We just can't have enough comments!
RefreshScreen Keep an active log inside your code, you never know when you need it (SVN alone isn't good enough)... (date, initials, ticket number).
The fun thing about this is that when a fix is larger than a few code files it does not have to be commented.
If it is only a few code files chances are this piece of commentary is copied to each of them, sometimes in multiple places in the same file.
CallSomeMethod Some of this guys methods are commented line by line with these kinds of comments...
I tried to talk to him about it, but he's convinced code is unreadable and can never be understood easily by just reading it. Comments are absolutely necessary if you, or another programmer, ever want to quickly make changes to the code. This is the same guy who thinks making a seperate class for almost every method (and randomly make it non-shared/non-static or shared/static) makes code more clear and readable (and single responsibility principle is a bonus!) and who just doesn't use access modifiers out of laziness or indifference...
Of course he is my boss and has eight years of experience where I have only two (how often I had to hear that) so when I disagree with him I'm rude and inexperienced.
I guess someday I'll laugh about it
It's an OO world.
public class Naerling : Lazy<Person>{
public void DoWork(){ throw new NotImplementedException(); }
}
|
|
|
|
|
Naerling wrote: Keep an active log inside your code
Ugh. I hate that. At my last job, they were all about that. Half a million lines of SQL, and none of it in version control (it was all nearly lost once, except I and another person had a personal backup).
Naerling wrote: I guess someday I'll laugh about it
When the company collapses and you find a new job with reasonable code practices.
|
|
|
|
|