|
try
{
int i = 1;
int j = 2;
return i;
}
finally
{
i += 3;
}
The code in the finally statement will be executed ? when? before return or after return ?
|
|
|
|
|
What happened when you tried?
|
|
|
|
|
Well, that's a dumb question. If there's no other code, then before or after the return are basically the same thing, excepting that when the code returns, i will no longer exist. In fact, I don't think this will compile, because the scope of i is the try block, I think you need to define it outside the try for the catch to see it.
Christian Graus
Driven to the arms of OSX by Vista.
"! i don't exactly like or do programming and it only gives me a headache." - spotted in VB forums.
I can do things with my brain that I can't even google. I can flex the front part of my brain instantly anytime I want. It can be exhausting and it even causes me vision problems for some reason. - CaptainSeeSharp
|
|
|
|
|
Well that is a dumb answer because it's totally unhelpful!
You're right that it won't compile, but surely you can see that the question works by simply declaring i at the method level instead. And there is in fact a subtle effect of the try-finally.
int foo()
{
int i = 1;
try { return i; }
finally { i = 5; }
}
This code compiles, but what does foo() return? I think quite a lot of people would expect it to return 5, since the finally block is executed before the method returns. But the return value is *evaluated* before the finally block runs and the method returns 1.
Now what would happen if we boxed the int in a class of our own?
class BoxedInt { public int Value; }
BoxedInt bar()
{
BoxedInt i = new BoxedInt() { Value = 1 };
try { return i; }
finally { i.Value = 5; }
}
While foo() returns 1, bar() returns a BoxedInt whose Value is 5. That's because BoxedInt is a class and thus a reference type, so it does not matter if the return value is evaluated before or after the finally block executes. Change BoxedInt to a struct, and bar will return a BoxedInt with Value equal to 1 as well.
I'd say there's some subtlety here, so calling that a stupid question seems to me rather a stretch.
|
|
|
|
|
So... you did try it, but asked the question anyway? 
|
|
|
|
|
Now that got me nicely confused.
You replied to my message so presumably "you" means me. But what "it" and "the question" refers to is less clear. I tried many things including compiling and running the various code snippets I've put up on this thread. I may have asked questions too, but I am not the OP so if you were referring to the original message that might explain the confusion.
If you did mean I tried something (and consequently found out what it does) and still asked what it does, let me know what it was and I will attempt to communicate whatever it was I was wondering about more clearly.
|
|
|
|
|
Whoops, I meant to reply to the OP.
|
|
|
|
|
I think it was an interesting question... thou I don't see why you need to ask when you can just try yourself. Thou as Christian said, you variable in the finally block is not in scope.
The answer would appear that the finally code is executed after the return thou I would say that the function is not complete until after the finally block. That means whenever return is called then that is the value that will be return regardless of what is done in the finally block.
I wonder what would happen if you put a return in the finally block too....
...oh, you cant return from a finally block.
Life goes very fast. Tomorrow, today is already yesterday.
|
|
|
|
|
The answer is in fact more subtle: The finally block executes before the method returns, but the return value is *evaluated* before the finally block runs. Because his code works with a value type (int) the increment he did in the finally is not reflected in the return value of the method - it's as if the compiler transformed return i; to int ret = i; i += 3; return ret;, executing the finally block after evaluating the return expression, but before returning. If it had been an object (reference type) and the finally modified the state of that object, the caller would "see" the changes made by the finally block. See my replies above if you want more details.
|
|
|
|
|
That's the same thing I said just using different words, so maybe before correcting me you should have spent time understanding what I said and not just repeating me and thinking that you have given a different answer.
Life goes very fast. Tomorrow, today is already yesterday.
|
|
|
|
|
I'll try to remember you're an ungrateful little brat. Actually you didn't say half of what I did, although you did say something vaguely similar. (Don't reply just to repeat a claim - readers can just look at the thread and decide for themselves.)
|
|
|
|
|
You know, going round insulting the regulars isn't a good way to endear yourself to a community.
|
|
|
|
|
And you think I deserved the snotty reply he gave me? I think I contributed a very clear explanation of what exactly happens and provided more information than he had done (his explanation was onto something, but didn't point out how reference versus value types would affect the outcome - so I had something to add).
It's up to you whether you rate contributions according to how regularly the author posts messages. I'd rather rate them for quality and value. And I reserve the right to return rudeness to anyone, regular or not.
|
|
|
|
|
Ungrateful for what? you didn't post anything that was designed to help me.
I said that the return is executed before the 'finally code' but the function is not complete (returned) until the finally code is processed.
dojohansen wrote: Actually you didn't say half of what I did
I was not trying to claim to have said what you did - I said you said what I said.
Anyway, my point is I don't appreciate something thinking there correcting me by basically saying the same thing.
If I said: "1 + 1 = 2"
and then someone said: "No, Actually, one add one equals 2"
then that person would be considered some sort of idiot, don't you think?
Life goes very fast. Tomorrow, today is already yesterday.
|
|
|
|
|
Just for any who are interested then dojohansen sent me an email with the following content...
Hi,
I don't want the thread to degenerate into a petty fight between you and me about whether or not we contributed something. But I do want to replay to *you* - hence this private message.
musefan wrote:Ungrateful for what? you didn't post anything that was designed to help me.
I said that the return is executed before the 'finally code' but the function is not complete (returned) until the finally code is processed.
I definitely said something that was meant to help, although it was not meant for you specifically. A reply in a forum is a reply to a message, not to the author of the message - if I wanted to address you I'd send a private message like this one.
What you said (as you accurately repeat in the above quote) is not wrong, but nor is it very precise. There's no mention of the crucial fact that the return value is evaluated before the finally executes[I said that the 'finally block' executes after the return - which to me is the same as saying "the return value is evaluated before the finally executes"]. There is also nothing that can help your reader understand that it would make a difference if the method worked with a reference type rather than a value type[personally I see the OP as simply asking if finally code is executed before or after the return statement. Not "how do I add to a variable before returning the value?"].
So although I feel people ought to be very happy to be corrected, and I try to be so myself when I am, this message was not even an attempt to correct you. I merely tried to clarify further and add some information regarding the subtleties that arise from the fact that the return value is evaluated before the finally block runs and returned afterwards.
To this, you gave a reply that was less than useless. Having first misinterpreted my message as a correction you decided to give me a snotty reply accusing me of trying to score brownie points by repeating you[I did not accuse you of 'stealing' my answer]. Frankly I feel that if you thought that was what I was doing you should be kind enough to the rest of the community to either just forget about it, or alternatively bring it up in a private message. [the impression I got (intended or not) was that you were correcting me and then saying it was the same answer - to me that warrants a telling off]
musefan wrote:Anyway, my point is I don't appreciate something thinking there correcting me by basically saying the same thing.
You appear very insecure to be so touchy about what you perceive as being corrected[insecurity is irrelevant here and therefore I can only assume it is meant as a dig]. Perhaps you should consider, if someone does some day write the same thing as you, the possibility that what you said wasn't very clear, or not complete, or that maybe the person had misunderstood your message. Immediately going for the knee-jerk reaction and attacking people for "copying" you makes you look fairly stupid, especially if the people you accuse of doing so are making contributions of superior quality to your own.[yeah maybe I could have handled it differently but you clearly misunderstood my answer and therefore instead of 'correcting' it then you should say you don't understand]
All of this having been said, I hope we can both contribute constructively to the forums and the community and not have to waste time on silly ego-driven [I am on an internet forum that does not display my real name - my ego is not effected either way by any post from anyone] activities that just pollute the threads for anyone interested in the *issue* we're supposed to discuss[people don't have to read what they don't think is worth reading].
Have a nice day,
Dag Johansen
...I don't have a problem being corrected when I am wrong (I am also a user of this forum to learn in addition to help) But there are some comments in this email that I don't see to make sense (which lead me to believe my original 'answer' requires more intelligence to interpret than I had hoped) but maybe that's just my opinion. everyone else if welcome to form there own. My comments on the email are in square brackets [ ] BTW
Life goes very fast. Tomorrow, today is already yesterday.
|
|
|
|
|
He's obviously a self riteous idiot who likes the sound of his own voice (or rather the look of his own typing). Ive not even seen the name here before today, but in the last few hours he's answering question after question in the most verbose way possible.. often replying to other satisfactory answers (like yours above) just to add his 2pence worth.
Ignore him. You do a great job of answering questions here.
|
|
|
|
|
Thanks for your appreciation - I will take your advise and ignore any more responses 
Life goes very fast. Tomorrow, today is already yesterday.
|
|
|
|
|
At least I'm a "self-rietous idiot" who can spell reasonably well.
I've only posted a little over 200 messages since I became a member of CodeProject in 2006, partly because I often have too much work to do to fool around here, partly because I usually take some time to craft responses the way I think people ought to respond. I don't dictate how much detail should be provided in messages or how much care should go into things like clear explanations, correct English grammar and spelling, or code examples that provide clear illustration. But just like you guys presumably do, I post the sort of messages I think the forums need. And you know what? I've had quite a lot of people who appreciated it. Every one of those is worth about a hundred of those smallminded responses I sometimes get that seem motivated by a need to point out that I am no better than them. If I never said I was better and the only reason people react this way is because I provide a decent answer to a question, that's fine by me.
It will be interesting to see if you take your own advice and just ignore me.
|
|
|
|
|
I tried to take this off the thread because in my opinion personal differences do not belong to a thread where someone asks about a try-finally block, nor do I think it is the place to discuss forum etiquette if that is what you attempt to make it look like you're doing.
Since you insist on polluting with this, I shall offer you one more insult for your fragile ego to consume today: You really are the biggest loser I've yet come across.
|
|
|
|
|
Since you're being such a pain in the back about this, I'll just keep it up.
musefan wrote: I said that the 'finally block' executes after the return - which to me is the same as saying "the return value is evaluated before the finally executes"
Well at this point I cannot just add to what you're saying, because the finally block of course does NOT execute "after the return" - that would be impossible. No part of a method executes after it has returned, and that certainly includes the finally. And saying that is the same "to you" doesn't make it right. If you say "that's what I *meant* to say" I'll just take your word for it, mainly because I don't really think it matters if it is true or not - my goal in contributing in forums is to put some stuff in here for the community to learn about, not to craft replies for you personally. (However, when you take it totally off the thread topic, I prefer to switch to personal messages unless my opponent is so rude as to force me to publicly defend myself.)
musefan wrote: I did not accuse you of 'stealing' my answer
Read your own reply that started this whole thing. Unless it was precisely that accusation, what content is there in the message? Are you trying to claim you just wanted me to know that you had already said the same thing I said? If so, why did you want me to have this piece of information?
musefan wrote: yeah maybe I could have handled it differently but you clearly misunderstood my answer and therefore instead of 'correcting' it then you should say you don't understand
First, I didn't correct you, I merely gave what I see as a better explanation. The closest I got to hinting you might have said something less than accurate was to say it was "a bit more subtle". To my mind, this was a way of acknowledging that my additional comment concerned minute details, but it seems to have upset you instead. Second, who are you to tell me what I should and should not post? I am in fact a big fan of saying "I don't understand", but not when I do.
|
|
|
|
|
Hi,
Your code wouldn't actually compile because you've scoped i to the try block but use it also in the finally block.
That said, any code of a method that executes at all *necessarily* does so *before* the method returns. BUT... your method, if you move the int i = 1; line before the try-block (so it can compile), will nonetheless return 1.
Logically you can think of it as transforming the code into something like this:
int i = 1;
try
{
int returnValue = i;
i += 3;
return returnValue;
}
catch
{
i += 3;
throw;
}
In other words, the finally code executes prior to returning, but everything else, including evaluating the return value, happens before. You can see this by taking your try-finally code, set a breakpoint at entry, and step through it using "step into" (F11 in the English Visual Studio).
However, while it is *possible* to write code that is a bit confusing using try-finally or try-catch-finally, in practice there is no real problem. Catch blocks are for error handling and finally blocks are for freeing up resources. The classic example is databases and transactions: Let's say your method needs to open a connection and start a transaction to insert a record into two tables. If either insert fails you want both to fail, ensuring the consistency of your database, and regardless of how the inserts go you must always close the connection before returning from the method. In other words, error handling consists of rolling back a failed transaction, and resource cleanup consists of closing (and thus freeing up) the connection. The code thus would be something like this:
SqlConnection cnx = new SqlConnection("Data Source=.; Initial Catalog=Northwind; Integrated Security=true");
cnx.Open();
SqlTransaction tx = null;
try
{
tx = cnx.BeginTransaction();
insertFirst();
insertSecond();
tx.Commit();
}
catch
{
if (tx != null) tx.Rollback();
throw;
}
finally
{
cnx.Close();
}
If any statement in the try-block fails, control passes to the catch block. If BeginTransaction() failed tx is null and no rollback is attempted; if anything else fails we have a transaction and it is rolled back. Whether or not the operation fails, the connection is closed before the method either returns normally or re-throws the exception that occured during the try-block.
Hope this helps!
|
|
|
|
|
Thank you for everyone's replies.
I saw one question that if there is the return statement in the try block, and when the finally is to be executed.
I thought that when return, the finally statement will not be executed. otherwise ,if input one return statement in the finally or change the variable value in the finally block, the result will be changed.
The code in the subject can not be compiled, because I define the variable i and j of the int type in the try block.
Now I modify them.
Following is the new code:
using System;
using System.Collections.Generic;
using System.Text;
namespace Finally
{
class Program
{
static void Main(string[] args)
{
int z = func1();
}
static public int func1()
{
int i = 1;
int j = 2;
try
{
i = 110;
j = 119;
return i;
}
finally
{
i += 3;
}
}
}
}
I find that the return statement can not be typed in the finally block.
From the asm code, we can find that the function save the return value and then jump to the finally block to execute. so the result will not be influenced.
Following is the asm code.
using System;
using System.Collections.Generic;
using System.Text;
namespace Finally
{
class Program
{
static void Main(string[] args)
{
int z = func1();
}
static public int func1()
{
00000000 push ebp
00000001 mov ebp,esp
00000003 push edi
00000004 push esi
00000005 push ebx
00000006 sub esp,38h
00000009 xor eax,eax
0000000b mov dword ptr [ebp-10h],eax
0000000e xor eax,eax
00000010 mov dword ptr [ebp-1Ch],eax
00000013 cmp dword ptr ds:[00AE8364h],0
0000001a je 00000021
0000001c call 792611C6
00000021 xor edx,edx
00000023 mov dword ptr [ebp-3Ch],edx
00000026 xor edx,edx
00000028 mov dword ptr [ebp-40h],edx
0000002b xor edx,edx
0000002d mov dword ptr [ebp-44h],edx
00000030 nop
int i = 1;
00000031 mov dword ptr [ebp-3Ch],1
int j = 2;
00000038 mov dword ptr [ebp-40h],2
try
{
0000003f nop
i = 110;
00000040 mov dword ptr [ebp-3Ch],6Eh
j = 119;
00000047 mov dword ptr [ebp-40h],77h
return i;
0000004e mov eax,dword ptr [ebp-3Ch]
00000051 mov dword ptr [ebp-44h],eax //save the return value
00000054 nop
00000055 mov dword ptr [ebp-20h],0
0000005c mov dword ptr [ebp-1Ch],0FCh
00000063 push 0E311B7h
00000068 jmp 0000006A
}
finally
{
0000006a nop
i += 3;
0000006b add dword ptr [ebp-3Ch],3
}
0000006f nop
00000070 pop eax
00000071 jmp eax
00000073 nop
}
00000074 mov eax,dword ptr [ebp-44h]
00000077 lea esp,[ebp-0Ch]
0000007a pop ebx
0000007b pop esi
0000007c pop edi
0000007d pop ebp
0000007e ret
0000007f mov dword ptr [ebp-1Ch],0
00000086 jmp 00000073
|
|
|
|
|
Indeed; this is what I explained, but thanks for the disassembly to prove it.
Just one tip: Wrap your code in <pre></pre> tags and it will make a world of difference for readability!
|
|
|
|
|
|
I have the following code set up to iterate through an array making an http get request on each loop. The trouble I am having is that after about the 9th queue I start to get 503 : Errors from the server. Am I doing something wrong? This currently runs on a timer than runs every 5 seconds. I did alot of reading on this over the last couple days and it appears that I am properly reusing the objects, doing a netstat -an from my machine shows only one connection to the server, but I am essentially doing 17 http Gets every 5 seconds.
foreach (Queue queue in Queues)
{
try
{
string uri = "myProvider/com.broadsoft.xsi-actions/v1.0/user/" + queue.Number + "/queue";
XmlDocument xmlDoc = new XmlDocument();
HttpWebRequest webRequest = (HttpWebRequest)WebRequest.Create(uri);
webRequest.Credentials = new NetworkCredential("MyUser", "MyPass");
webRequest.Timeout = 5000;
webRequest.UserAgent = ".NET Framework Test Client";
HttpWebResponse wResponse = (HttpWebResponse)webRequest.GetResponse();
using (StreamReader responseStream = new StreamReader(wResponse.GetResponseStream()))
{
xmlDoc.Load(responseStream);
StringBuilder aHttpCallXmlOutL = new StringBuilder();
string tempString;
while ((tempString = responseStream.ReadLine()) != null) { aHttpCallXmlOutL.Append(tempString + "\r\n"); }
responseStream.Close();
wResponse.Close();
}
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
