[Snippet] Constant Timer Loop 32

[Bribe]

You have not "completely shortened" the modules, because you still have those useless "return false" things in there.

Keep in mind that "returns nothing" no longer crashes Mac computers (tested on OS X Snow Leopard). This "returns nothing" fix probably happened when Macs started using Intel chipset.

Also keep in mind that a huge amount of Mac users have switched to Lion now, and WarCraft III doesn't even run on Lion.

"returns nothing" is perfectly fine to use with boolexprs.

Sources:

http://www.wc3c.net/showthread.php?t=110483
http://forums.battle.net/thread.html?topicId=27813861503&sid=3000

 

[Nestharus]

Bribe, I prefer to keep with better coding practices. A boolean expression should always return a boolean. Who knows what kind of nonsense is going on behind the scenes with functions that don't return anything :\.

 

[Bribe]

I prefer to keep with better coding practices

You can't honestly use that argument, unless you want to go back to using names for your variables, and stop caring about meaningless efficiency things like "0 == i".

It's been tested pretty extensively as well. So far, it just acts the exact same as "return false". The only thing left to do is the benchmark.

 

[Nestharus]

It's been tested pretty extensively as well. So far, it just acts the exact same as "return false". The only thing left to do is the benchmark.

Yea, but I don't know what it's doing behind the scenes... when something is supposed to return something and it doesn't, I don't know how wc3 handles that. It could leak, I don't know >.>.


The main issue here isn't speed or anything, it's the fact that Conditions, Filters, and etc are supposed to return booleans and I don't know how wc3 handles them if they don't return booleans. The only way I'd be willing to change it is if you could, without a doubt, tell me how it handled them.

 

[Bribe]

True, wc3 is mightily unpredictable. It has been working no problem for me as I've been using "return nothing" boolexprs for my WarChasers revision and all the results have happened just as predicted. I will run some tests later that compare the FPS and the RAM usage of "return nothing" versus "return false".

 

[Nestharus]

This is going to be deprecated with the next version of Timer Tools. Time to graveyard this ; ).

 

[Bribe]

Adding backwards compatibility?

 

[Nestharus]

Well.. I could update this to just run completely off of Timer Tools, but that'd be semi silly ^)^.

It'd just be CTM module with a constant timeout, lol... the new version I'm working on is very good..

Also, I found out that TimerUtils is crap. I already made a much faster version for 1 shot timers.. TimerUtils is slower than standard timers >.>., it's just bad.

edit
yea, this just needs to be gy'd.

 

[Bribe]

Why should this be graveyarded? The user would have to import your huge Timer Tools library just if they want this little snippet.

This has a purpose because it is a lightweight version.

 

[Magtheridon96]

I have an idea, keep this library for backwards compatibility and use static ifs to check if TimerTools is in the map.
If it is, you would generate wrappers called CTL, CTLExpire, CTLNull, CTLEnd, CT32 and CT32End that would wrap the TimerTools module implementations, else, you'd use those modules to wrap the CTM code.

This shouldn't be graveyarded, I already have approximately 80 libraries in my map that use it, and a bunch of public resources that use it too D:

But if you really want to graveyard it, you should ask all the people who uploaded spells and systems that use CTL prior to August 2011 to make updates.

Which solution sounds more tempting to you? Backwards compatibility, or a spells/systems section full of spells that no longer compile?

 

[Nestharus]

I have an idea, keep this library for backwards compatibility and use static ifs to check if TimerTools is in the map.
If it is, you would generate wrappers called CTL, CTLExpire, CTLNull, CTLEnd, CT32 and CT32End that would wrap the TimerTools module implementations, else, you'd use those modules to wrap the CTM code.

This shouldn't be graveyarded, I already have approximately 80 libraries in my map that use it, and a bunch of public resources that use it too D:

But if you really want to graveyard it, you should ask all the people who uploaded spells and systems that use CTL prior to August 2011 to make updates.

Which solution sounds more tempting to you? Backwards compatibility, or a spells/systems section full of spells that no longer compile?

thing is, this is written incorrectly atm... if I were to correct it, it would be almost an exact clone of Timer Tools.

 

[Luorax]

I've also written a couple of things based on this resource, so I wouldn't like to go through the whole map and rewrite things again.

 

[Nestharus]

Ok then, how about marking this as deprecated rather than gying it?

That means it's still up, but no longer supported as better things have been done.

 

[Luorax]

Yea, maybe - but I'd vote for Maggy's suggestion (if possible)

 

[Magtheridon96]

I guess I'll update my public resources to use TimerTools and you can mark this as Deprected until we can confirm that no one's using it

There are probably 5-10 spells/systems other than mine that use CTL (Hehe)

 

[Nestharus]

Yea, maybe - but I'd vote for Maggy's suggestion (if possible)

Updating this is pointless work that I'm not going to do ^)^

edit
What I can do is add a constant merge module that specially works with constant low period timeouts ^)^. I can also add a constant timeout module. These will increase the speed of creation/destruction a little bit.

 

[Luorax]

Well, looks like I'll have to give that monster, on thread below a try. Fortunately it's just a few hundreds of triggers I'll have to go through and modify if I want to gain extra performance.

 

[Nestharus]

Well, looks like I'll have to give that monster, on thread below a try. Fortunately it's just a few hundreds of triggers I'll have to go through and modify if I want to gain extra performance.

don't modify if you don't want to >.>.


This still works I believe, so you can just use this if you like.


edit
Yea, this has faster expiration, creation, and destruction and still works just fine, so just use this. No reason to change if ur already using this.

 

[Bribe]

Glad to see that we're all on the same page. if this is faster then there's no need to deprecate it

 

[Nestharus]

Well... actually with the latest version of Timer Tools, the speed gain on everything is negligible on this. CTTC is just about as fast and can merge with CTM and CTT timers.

 

[Bribe]

the way I see it: this supports low timeouts (or ctl does if you use it already) and then a non- preloaded stack of timers for the high timeouts or one shot timers

 

[Nestharus]

And CTTC from Timer Tools supports exactly what this does with very minor loss in speed from create/destroy and very minor loss in speed in expiration (negligible). Furthermore, CTTC merges with timers from anywhere else in the system (CTM timer with a .1 timeout would merge with CTTC module with .1 timeout). Therefore, the CTTC module beats this module out :\, hence why I say it's deprecated ;D. Furthermore, CTTC supports any timeouts that will have a constant merge, not just .03125 ; P.

I'm not saying to graveyard this at this point, I'm just saying that this should be marked as deprecated so that people don't use it in the future ; ).


I obviously don't recommend its use over the CTTC module from Timer Tools ; P.

If you want methods where you can create your own loops, Timer Tools has the functions to do that.

 

[Magtheridon96]

http://www.hiveworkshop.com/forums/triggers-scripts-269/spell-bug-213357/

Care to explain? O-o

edit
If you delete line 140, you can fix this resource.

 

[Nestharus]

Commented out line 140 by popular request ;p

 

[Bribe]

What a sneaky bug.

 

[Nestharus]

There are other bugs as well, but w/e.

The entire design is a little bit flawed ; )

 

[Bribe]

What other bugs?

edit: delete this line private timer t32r = CreateTimer()

 

[Nestharus]

Deleted

the entire design is a little flawed. Compare it to the design of Timer Tools and you'll see what I mean ;o

 

[mckill2009]

does this support arrayed members?, coz in my code i tried to put arrayed units and reals, like...

private struct SE
   unit array bombs[500]
   real array gap[500]

and distributed it this way in the creation method...

loop
   exitwhen i==maxbombs
   set .bombs[i] = CreateUnit(p,BOMB_ID,0,0,0)
   call SetUnitFlyHeight(.bombs[i], 0,0)
   call UnitAddAbility(.bombs[i], DUMMY_SPELL_ID)
   call SetUnitAbilityLevel(.bombs[i],DUMMY_SPELL_ID, lvl)
   set .gap[i] = g
   set g = g + gapdist
   set i = i + 1
endloop

only the first cast does well, the rest not, meaning NOT MUI, on the other hand, I've converted it to T32 and works fine...why is this happening?...

 

[Magtheridon96]

Because arrayed members are compiled into shit.
It's not Nestharus' fault.

By the way, using a size of 500 makes it only work for 16 casts total.

The solution is to use a Table.

 

[Luorax]

Or use an array struct and 2D static arrays. A Table would be definitely faster and safer, but the 2D array workaround can be perfectly fine, and it's faster as far as I know.

 

[mckill2009]

Well that proves your theory that arrays compiled shit, coz I've never encountered such...
as I said, T32 works just fine, and I have a version of it using hashtables as well, it's finr also, but it's slow...
Im thinking of TableArrays but seems T32 is faster to load and easier...

btw 500 is just a test mags...

 

[msongyboi]

this is so confusing, I don't get what all this does... can someone help me in breaking it down, line by line?

 

[edo494]

why do you have //set ct[r] = null line when you have it commented, you could just remove it

Also why you keep writing 0 == ... and not ... == 0 ?

 

[Bribe]

He claims to have tested 0== as an atom on a planet faster and thus changed all of his scripts therein and thereafter to exploit this enormous real world dvantage.

 

[Magtheridon96]

0 == ... is only faster because of the cache line and the alignment. (I don't know if they're variable struct has padding)

 

[edo494]

lol one must be mad when doing 0 == ... instead of ... == 0 when thinking that will give any performence gain

 

[Nestharus]

Apparently someone found a bug with this, so I'll update it today

 

[Luorax]

Yup, "someone".

 

[Magtheridon96]

EUPHEMISMS. Everywhere.

 

[Luorax]

Wasn't sure if you had told him that it was me, I just wanted him to know that I was that faggot that bugged his snippet. Plus I wanted to re-subscribe, maybe there's more that I didn't notice.

 

[Nestharus]

Fixed

Also improved efficiency a bit and behavior on the flags.

 

[Luorax]

Cool, I'll update, don't really like modified stuff.

ty for the bug report, but next time just post on the thread =)

Sorry about that, I knew you prefered TT over this, and you said there were other bugs, thought you were aware of it. I will go directly to you next time then

 

[Luorax]

Oh, one more thing: why is "start" in the CT32End module? Sometimes (quite often) the loop stops itself, so putting it to CT32 instead of CT32End can help you avoid those trigger evaluations.

 

[Bribe]

Oh, one more thing: why is "start" in the CT32End module? Sometimes (quite often) the loop stops itself, so putting it to CT32 instead of CT32End can help you avoid those trigger evaluations.

Good point. He can move start and stop into the CT32 model.

 

[Nestharus]

They were in the End so that they could see the expiration function, but I could do an onInit and a global code var :\

 

[Bribe]

They were in the End so that they could see the expiration function, but I could do an onInit and a global code var :\

Well the OnInit method doesn't have to move, and Stop and Start don't depend on the placement whether they are top or bottom.

 

[Nestharus]

right you are

 

[Luorax]

Yea, I was kinda confused too, but I couldn't reply when I saw your post. >.>

So simply moving stop (and possibly start too) is enough. I've actually done it when the bug was fixed, but only now crossed my mind to actually post the suggestion here.

 

[Nestharus]

Why can't you all make helpful suggestions for Timer Tools too ;(