[taz-me]

Hi,

It seems that CUpDownClient::SetDownloadState can be called with new download states not covered by main case (just for example : DS_ERROR).

I suggesting (might be there's a better or more accurate ) the following code change :

@CUpDownClient::SetDownloadState
...
		switch( nNewState )
		{
			case DS_CONNECTING:
	            m_dwLastTriedToConnect = ::GetTickCount();
				break;
			case DS_TOOMANYCONNSKAD:
				//This client had already been set to DS_CONNECTING.
				//So we reset this time so it isn't stuck at TOOMANYCONNS for 20mins.
				m_dwLastTriedToConnect = ::GetTickCount()-20*60*1000;
				break;
			case DS_WAITCALLBACKKAD:
			case DS_WAITCALLBACK:
				break;
            case DS_NONEEDEDPARTS:
                // Since tcp asks never sets reask time if the result is DS_NONEEDEDPARTS
                // If we set this, we will not reask for that file until some time has passed.
                SetLastAskedTime();
                //DontSwapTo(reqfile);

// By Taz - fix
				break;
			default:
				break;
// <------- fix

			/*default:
				switch( m_nDownloadState )
				{
					case DS_WAITCALLBACK:
					case DS_WAITCALLBACKKAD:
						break;
					default:
						m_dwLastTriedToConnect = ::GetTickCount()-20*60*1000;
						break;
				}
				break;*/
		}
...

Tests held by some SharkX beta testers and myself showed a noticeable decrease in network related exceptions after the change (there might be a difference in actual code behavior since SharkX exe is build using VS2008)...

This post has been edited by taz-me: 19 November 2009 - 12:46 PM

[Some Support]

What do you think your addition is going to change in the codeflow? It looks pretty much semantically idententical to me.

[taz-me]

Some Support, on 19 November 2009 - 05:22 PM, said:

What do you think your addition is going to change in the codeflow? It looks pretty much semantically idententical to me.

Switch sentence won't fall short due to "unmatched" value.

[Some Support]

Fall short? Not sure what you mean. Just to quote msdn:

Quote

Switch Statement behavior

Condition -> Action

• Converted value matches that of the promoted controlling expression. -> Control is transferred to the statement following that label.
  
• None of the constants match the constants in the case labels; a default label is present. -> Control is transferred to the default label.
  
• None of the constants match the constants in the case labels; default label is not present. -> Control is transferred to the statement after the switch statement.

The break statement is used to stop execution and transfer control to the statement after the switch statement.

[taz-me]

I'll check the "default" label via parsing code to assembly (it will take me some time) ...

Regardless, if the quote covers the whole concept of taken action, the break at the end of DS_NONEEDEDPARTS is still needed - since otherwise one is relying on code (each label) being parsed at the order they appear. This is surely wrong since it is legitimate to place a "default" label as 1'st label and yet other labels code is reached ... (no compiler should add such a missing break on his own)

This post has been edited by taz-me: 19 November 2009 - 05:41 PM

[Stulle]

nope, there is no need for a break in the last case and there is no need for a default. a default will be called if no case matches the input. that makes sense for debugging or if you got 8 cases and at this point in code only two different measures are taken such as this:

int iColor = randomColorCode(); // 8 colors represented by 8 integer numbers
switch(iColor)
{
case 1: // white
ColorTheFool();
break;

default:
PraiseTheFool();
break;
}

note, the break at the end is not needed but for this see below.

now, on the usage of break in switch constructions... break is only needed if other executions are not needed. this way you can link up various cases if needed. here's an example:

int iMissingClothing = CurrentlyMissingClothing(); // 1 - boxer shorts missing, 2 - pants missing, 3 - shirt missing, 4 - hat missing

switch(iMissingClothing)
{
case 1:
SlipIntoBoxerShorts();
// no break here because if we miss the shorts we miss the pants!
case 2:
SlipIntoPants();
break;
case 3:
SlideOnShirt();
break; // again, this is not really needed
}

now, if we got no shorts, we got no pants - unless we are some kind of filthy underwear missing redneck - so when we get to case 1 we also go to case 2. if we only miss the pants we just slip those on. if we got no shirt we slide on a shirt and are done for it. if we miss the hat nobody cares.

so, we could of course make a default and/or a case 4 but they will not benefit any useful code addition because all they will do is call "break;" to terminate the switch construct which is not benefiting in the slightest...

i took my time explaining this to you but it is yet again another of those fixes you present that is basically trivial and brings no gain whatsoever even if it does not corrupt the working code... follow SS's lead and use MSDN already!

This post has been edited by Stulle: 19 November 2009 - 06:07 PM

[taz-me]

Stulle, on 19 November 2009 - 08:03 PM, said:

nope, there is no need for a break in the last case and there is no need for a default. a default will be called if no case matches the input. that makes sense for debugging or if you got 8 cases and at this point in code only two different measures are taken such as this:

int iColor = randomColorCode(); // 8 colors represented by 8 integer numbers
switch(iColor)
{
case 1: // white
ColorTheFool();
break;

default:
PraiseTheFool();
break;
}

note, the break at the end is not needed but for this see below.

- just slow down right here, what are you basing your claim that last break is not needed upon ...
- and what if the labels were with reversed order, would you claim the same (I'm refering for not needing the break on last label) ???

(your second example is trivial, yet making a code somewhat less clear for readers)

Quote

i took my time explaining this to you but it is yet again another of those fixes you present that is basically trivial and brings no gain whatsoever even if it does not corrupt the working code... follow SS's lead and use MSDN already!

Don't waste your time on educing me if your argument (claims) are based on nothing but thin air (I'm not saying your statements are faulty - just that they do not rely on anything you posted ...)

[Stulle]

you still have not understood the concept of the switch statement. there obviously is a difference when you change the order. okay, here another example that will show how a switch can be rewritten using the goto command (which should not be used)...

switch statement code:

int iMissingClothing = CurrentlyMissingClothing(); // 1 - boxer shorts missing, 2 - pants missing, 3 - shirt missing, 4 - hat missing

switch(iMissingClothing)
{
case 1:
SlipIntoBoxerShorts();
// no break here because if we miss the shorts we miss the pants!
case 2:
SlipIntoPants();
break;
case 3:
SlideOnShirt();
break; // again, this is not really needed
} // closing the switch statement execution block here

transcribed code:

int iMissingClothing = CurrentlyMissingClothing(); // 1 - boxer shorts missing, 2 - pants missing, 3 - shirt missing, 4 - hat missing

if(iMissingClothing == 1) goto c1;
else if (iMissingClothing == 2) goto c2;
else if (iMissingClothing == 3) goto c3;
else goto end;

c1: SlipIntoBoxerShorts();

c2: SlipIntoPants();

goto end; // this represents the break

c3: SlideOnShirt();

end: return; // this represents } closing the switch statments execution block

it is apparent that we do not need an extra "goto end;" after finishing c3. also, since the else makes us directly go to "end: return" which represents the end of the switch execution block it is apparent that we do not need a c4 and cDefault that trigger a "goto end;". this is all most basic C++!

as for my second example from above, it may be trivial - i made both of them this way so you'd understand... my bad, assuming you would, here... O_O - but it's not making anything harder to read, in fact it simplifies a whole lot. if you'd make a break; statement after the execution of case 1: you'd have to add the instructions from case 2, to case 1 as well. now, it might not be as bad in my example but imagine you got 30 or 50 lines of complex code there you add twice just because you are desperately trying to close ANY case statement with a break. code should be short and readable. the one i posted above is readable if you do not lack sufficient knowledge of the C++ language...

you know, i got no problem with teaching people stuff about c++ after i spent learning it for some years now. i made many mistakes when i started and i still do some but there is one thing i never tried and that was prove people with apparently superior knowledge in these matters wrong in basics of the language. no problem with you still learning but you honestly need to accept knowledge that we present you. this is not philosophy, this is computer science and if two people and a world renowned software manufacturer (MS) agree upon the terms of usage of switch, case and break they might all actually tell you what the ISO/IEC 14882:2003 set into stone in 2003.

i would have thought better of a linux sys admin... but probably it requires different skill than basic C++ knowledge...

This post has been edited by Stulle: 19 November 2009 - 09:33 PM

[Freak51]

taz-me, on 19 November 2009 - 06:21 , said:

This is surely wrong since it is legitimate to place a "default" label as 1'st label and yet other labels code is reached ...

If I remember correctly You cannot place default as the 1st action because default will always be executed (unless you break before, of course), making any following case useless..

So default has to be the very last case

[fox88]

Freak51, on 20 November 2009 - 12:42 AM, said:

taz-me, on 19 November 2009 - 06:21 , said:

This is surely wrong since it is legitimate to place a "default" label as 1'st label and yet other labels code is reached ...

If I remember correctly You cannot place default as the 1st action because default will always be executed (unless you break before, of course), making any following case useless..

So default has to be the very last case

No, it has not.
Default is commonly placed last only because switch statemnt looks more logical that way.

[Stulle]

yep, i thought the same and reading the MSDN article on switch this assumption is correct. default may be placed anywhere.

speaking of which, the MSDN article has an example that uses a default without break at the end. it is of course a different matter if you start rearranging the cases but if you do that just the need for break shifts. so if case 1 needs it and case 2 needs it but default does not need it and you move default to the top there will obviously be a need for break after default and no need for break after case 2 (now being the last case). it is also notable that the default statment is not just otional but can also be entirely disabled!

This post has been edited by Stulle: 20 November 2009 - 08:38 AM

[taz-me]

Stulle, on 20 November 2009 - 10:37 AM, said:

yep, i thought the same and reading the MSDN article on switch this assumption is correct. default may be placed anywhere.

... lucky me

Let us all do prcatice togather :

int iColor = randomColorCode(); // 8 colors represented by 8 integer numbers
switch(iColor)
{
default:
ColorTheFool();
break;

case 1: // white
PraiseTheFool();
break;
}

1'st example labels switched.

int iColor = randomColorCode(); // 8 colors represented by 8 integer numbers
switch(iColor)
{
default:
ColorTheFool();
break;

case 1: // white
PraiseTheFool();
}

Someone claimed break is not needed on last label. Is it ???

That someone claimed / guessed (choose since edited) : switch is translated to if then else if ...

Instead of guessing, placing a break can eliminate the doubt as well as the risk of compilers interpeting (VC++ is just one of many C++ compiler and MS didn't invent C++) how to parse last label code without a break statement.
Further more, if due to compiler optimization labels code order is shuffeled - should one rely on a break statement added for him ...

[Stulle]

YOU ARE STUPID! you got no idea and are ignorant to learn. just STFU and GTFO, it is unbearable to even read this. i have presented you simple examples, explanation and a link to MSDN, if you are incapable to understand those just shut it and begone.

did you even try to imagine that optimization may be aware of what the coder intended? it is optimizing the code and not fucking up the code. you might code an optimizer function that ignores what the coder wrote but no sane professional software engineer would. so even if the cases are being rearranged they would make sure that just the statments from the original code would be executed and not all following statements as if the written code was rearranged.

this got nothing to do with guessing, the only person guessing and arguing on a standard is you! besides, while MS did not invent C++ it sure as hell left a mark on its compiler and thus the dialect in which C++ needs to be written to work with their compiler. are you even aware that compilers do not compile all every code identically? while the most compilers do not allow array of unknown size (int iArray[]) at least one other compiler (the name slipped me just now) allows this.

what is your fucking problem, anyway? SS told you it is unneeded, i tried to make you understand why it is unneeded and posted the link to the MSDN article and yet you insist that you are right? how can one single being be so ignorant and self-centered to admit he actually learned something? must be because you are stupid and feel free to laugh because you managed to enrage me once again. i may be a fool to get enraged but you are a so much bigger fool to display your lack of ability like this.

disclaimer: strong language was needed or my eyes would have bled from this for nothing...

edit: on your example... you just switched the label which ruins the codes functionality. you are just proving you understand nothing.

This post has been edited by Stulle: 20 November 2009 - 10:12 AM

[fox88]

Stulle, on 20 November 2009 - 11:37 AM, said:

it is also notable that the default statment is not just otional but can also be entirely disabled!

The default case in switch statement is part of C/C++ ANSI standard, therefore I do not see how it could be disabled.
In MSDN article you mentioned the word disable refers only to MS extensions for minimum ANSI requirements about the number of cases in switch statement.

taz-me, on 20 November 2009 - 12:48 PM, said:

Someone claimed break is not needed on last label. Is it ???

In general it depends on your intentions - break is necessary if you want to interrupt linear (as the code is written; and standard specifies that) execution of code inside the switch statement.

taz-me, on 20 November 2009 - 12:48 PM, said:

That someone claimed / guessed (choose since edited) : switch is translated to if then else if ...

As far as I know, it is not if-then-else in VS, but that fact is of no importance as long as switch statement is interpreted according to standards.

taz-me, on 20 November 2009 - 12:48 PM, said:

Instead of guessing, placing a break can eliminate the doubt as well as the risk of compilers interpeting (VC++ is just one of many C++ compiler and MS didn't invent C++) how to parse last label code without a break statement.

There is no guessing: after the last case in a switch statement break statement would be superfluous.

[Some Support]

Remember the forum rules, be nice!

[Freak51]

fox88, on 20 November 2009 - 09:04 , said:

Freak51, on 20 November 2009 - 12:42 AM, said:

taz-me, on 19 November 2009 - 06:21 , said:

This is surely wrong since it is legitimate to place a "default" label as 1'st label and yet other labels code is reached ...

If I remember correctly You cannot place default as the 1st action because default will always be executed (unless you break before, of course), making any following case useless..

So default has to be the very last case

No, it has not.
Default is commonly placed last only because switch statemnt looks more logical that way.

Ooopppss I see

Apart from that, in the code above one could remove/comment

 case DS_WAITCALLBACKKAD:
 case DS_WAITCALLBACK:
 break;

since there is no default statement

[taz-me]

Stulle, on 20 November 2009 - 12:04 PM, said:

YOU ARE STUPID! you got no idea and are ignorant to learn. just STFU and GTFO, it is unbearable to even read this. i have presented you simple examples, explanation and a link to MSDN, if you are incapable to understand those just shut it and begone.

did you even try to imagine that optimization may be aware of what the coder intended? it is optimizing the code and not fucking up the code. you might code an optimizer function that ignores what the coder wrote but no sane professional software engineer would. so even if the cases are being rearranged they would make sure that just the statments from the original code would be executed and not all following statements as if the written code was rearranged.

this got nothing to do with guessing, the only person guessing and arguing on a standard is you! besides, while MS did not invent C++ it sure as hell left a mark on its compiler and thus the dialect in which C++ needs to be written to work with their compiler. are you even aware that compilers do not compile all every code identically? while the most compilers do not allow array of unknown size (int iArray[]) at least one other compiler (the name slipped me just now) allows this.

what is your fucking problem, anyway? SS told you it is unneeded, i tried to make you understand why it is unneeded and posted the link to the MSDN article and yet you insist that you are right? how can one single being be so ignorant and self-centered to admit he actually learned something? must be because you are stupid and feel free to laugh because you managed to enrage me once again. i may be a fool to get enraged but you are a so much bigger fool to display your lack of ability like this.

disclaimer: strong language was needed or my eyes would have bled from this for nothing...

edit: on your example... you just switched the label which ruins the codes functionality. you are just proving you understand nothing.

I've taken some of your advices : the one to keep myself away from threads and forums you have moderation power on, however by using such language I'm not sure, such forums and threads should exists ...

Based on past (not that far past - some of the stuff is still there) misses and mistakes I wouldn't call names or judge others if I were you.

And yes this is personal - you have an habbit of making it so. You knew what I was about to post here - few days a head, however you chosed to hush. Besides if you think Some Support (like JvA and might be more so - I can't recall) is being anoyyed by me, let him answer ...

I wasn't guessing stating that labels order (including default) shouldn't affect the outcome, however others were (yourself included with specultion about how switch sentence it parsed : once you claimed if then else if notation, later you changed it to goto ...) - so don't blame others of your own doing.

I've switched the labels on your 1'st example in order to point out the issue of break on last label. I didn't try to preserve outcome - just to point out that if default label is not the last one, if then else if parsing is not accurate and thus break on last label might be needed (surely won't hurt).

As for need of "break" on last label : as any programming language C++ have sytax and quoting examples to prove a point here is useless unless it is coupled with the parsed code or the affect of the code.

[Stulle]

just three small things. firstly, if you change an examples outcome it proves nothing anymore and it gets redundant. secondly, i never changed any of my examples when editing, i only edited text. this does also mean that my attempt to make you understand how the switch works transcribing it with an if/else and goto structure was like it is posted from the start! thirdly, i have never misused my power for personal gain, i only used it to stop endless and pointless discussion. as long as i am in charge of some maintenance around the Xtreme mod i got a vote in that matter and i am not ashamed to tell anyone who asks that i do not think i am a perfect moderator but at the time i got moderator of the Xtreme subforum i was the only regular who still cared. in fact, when i asked SomeSupport or Birk (can't actually remember just now) about this i stated that i am rather reluctant for that post. this is the exact reason why Andu became moderator for the Morph subforum and not me and this is the exact reason why i am no moderator on eMule-web.de board. you know what, despite my frequent usage of strong language you are no better than i am because you just go by personal feelings when judging on my work around the community. just take the latest DLP thing, rather than giving up on your idea when i labled it unsuitable for DLP you moved to another place to avoid "censorship" i never even thought about and went on about how much of a censor i am and how much i do not aprove of the idea because it is you. just ask some of the guys i am working with what i tell them when i don't like the idea. with you i've just been frank because i did not want to make a big thing out of it but you insisted. know what? why don't you just quit on arguing points set into stone? that includes basic C++ rules and projects i am involved in. that would help us both a whole bit and you can probably spend more time coding or at least learning to code...

fox88: ahh, you are right there... i must admit, i just skim-read the last few lines of the article because it was just some extra bit of information that does not actually effect the general way the statemnet is used.

edit: small extra on my side, admittedly the last bit of my first post was not too nice but then, if i would not have wanted to help i would not have gone the lengths to make up examples and write all that stuff in the first place... so you can't just say i was mearly trying to have a fight... think about it.

This post has been edited by Stulle: 20 November 2009 - 09:22 PM

[taz-me]

Stulle, on 20 November 2009 - 11:19 PM, said:

just three small things. firstly, if you change an examples outcome it proves nothing anymore and it gets redundant. secondly, i never changed any of my examples when editing, i only edited text. this does also mean that my attempt to make you understand how the switch works transcribing it with an if/else and goto structure was like it is posted from the start! thirdly, i have never misused my power for personal gain, i only used it to stop endless and pointless discussion. as long as i am in charge of some maintenance around the Xtreme mod i got a vote in that matter and i am not ashamed to tell anyone who asks that i do not think i am a perfect moderator but at the time i got moderator of the Xtreme subforum i was the only regular who still cared. in fact, when i asked SomeSupport or Birk (can't actually remember just now) about this i stated that i am rather reluctant for that post. this is the exact reason why Andu became moderator for the Morph subforum and not me and this is the exact reason why i am no moderator on eMule-web.de board. you know what, despite my frequent usage of strong language you are no better than i am because you just go by personal feelings when judging on my work around the community. just take the latest DLP thing, rather than giving up on your idea when i labled it unsuitable for DLP you moved to another place to avoid "censorship" i never even thought about and went on about how much of a censor i am and how much i do not aprove of the idea because it is you. just ask some of the guys i am working with what i tell them when i don't like the idea. with you i've just been frank because i did not want to make a big thing out of it but you insisted. know what? why don't you just quit on arguing points set into stone? that includes basic C++ rules and projects i am involved in. that would help us both a whole bit and you can probably spend more time coding or at least learning to code...

1. Using the language you did, got nothing to do with misusing moderation power. It has to do with manners and rules obeying - one that doesn't qualify by those should be provoked of moderation power.
2. You can not stand CA be mentioned on DLP thread, and for that matter any perssonal critisism as well (dispite the fact it's not stopping you from using it) - that's why I shifted discussion.
3. My coding skills are not bothering you, it's seeing my name on change log on things that might be valued (this is why you wouldn't change the faulty weight for preview chunks on ICS and remove the bypass which is not needed after the fix).
4. If you would spend more time on coding it will probably be more useful - so take the path that makes more sense ...

5. For any moderator here : Can you start a thread "Stulle is writing about taz codding skills" thread and shift all off topic discussions there.

[Stulle]

1. you said i was misusing my moderation powers. that is what you claimed. now you say i am unsuitable for my strong language but if i do not misuse my moderation power after all, maybe i can distinguish between moderator job and personal stuff...
2. yes, i do not like the CA to be mentioned in the DLP thread because it ain't got nothing to do with the DLP and i consider it repsonsible to stop old wounds and old discussions from being reopend. anyway, i am not posting in your thread to advertise my mods so don't you advertise the CA in my threads. simlpe, eh? oh and as for critcism, i can take criticism but it needs to be justified and constructive. when people come around "you mod is so unk00l" or "crashed. utter crap!" i don't consider it constructive in the least and I will reply accordingly.
3. no, what i don't like is trying to gain credits for trivial or even faulty changes. about ICS, your claim just shows you did not understand why Morph kept both versions of the packet request function. ICS is choosable in Morph so there is no loss when using the official for getting the first few chunks rather than the last one. besides, never change a running system they say, eh? that means to say i am reluctant to change code that works well because it might just as well introduce new bugs!
4. oh, be assured, i spent plenty hours coding in the last few weeks. my mod release cycles may not be as short as yours but i rather produce something that is well tested and has a some more extras than you probably do. anyway, this is none of your business because generally my users are satisfied.
5. wow, now that was mature! are you even seeing i am trying to make a step towards you and you just stay right where you are. if you intent to make the fighting stop you are best advised to learn that it takes not one but two for that.