open All Channels
seplocked EVE Information Portal
blankseplocked New Dev Blog: ing Lag: Module Lag - Why Not All Bugfixes Are A Good
 
This thread is older than 90 days and has been locked due to inactivity.


 
Pages: 1 2 3 4 5 [6] 7

Author Topic

Ban Doga
Posted - 2010.08.25 05:39:00 - [151]
 

Edited by: Ban Doga on 25/08/2010 05:39:46
Originally by: Ranger 1
What I find surprising about these blogs is that they had to be written in the first place.

What has been lined out is simply a brief glimpse into the methodology of any company in the gaming industry (although CCP has several unique, and stellar concepts that outclass their competition).

For some reason the EVE forum community feels it is their right to have the senior management and development team come and personally hold their hand, explaining every last detail of what they are doing and why (which any idiot should already have a general grasp of), and then argue with them in a egocentric attempt to show them the "right way to do it".

Rolling Eyes

Throw in a liberal helping of "we know you said this, but we all agree you really meant that" nonsense and you end up with the most myopic, rabble rousing, self centered collection of misfits it has been my displeasure to interact with.

The only positive thing to come out of this sorry mess is that I was pleasantly surprised that CCP took you seriously enough to give you more insight into their development process than any other gaming company (or any type of software company) has to my knowledge. Of course, they were met with condescension and general idiocy but hey, that appears to be the norm on these boards.

Now if you can raise the general level of response to something higher than you would expect from your average 13 year old muppet you might actually deserve the respect you have been given.



No one is forcing you to read the dev blog threads or the player comments.
If you find this to be such a displeasure you can just steer clear of them.

I assume you are old enough to find out what you like and dislike and provide yourself with an environment that suits you.
Viewer discretion is adviced...

Miklas Laces
A.N.A.R.C.H.I.C.A
Posted - 2010.08.25 06:47:00 - [152]
 

Quote:
At the Council of Stellar Management (CSM) summit in June, the CSM made an excellent presentation of the issues around fleet-fight lag, specifically about the issues of modules becoming ‘stuck' and the workarounds used by players. This provided useful insight into a problem which, until recently, we'd never been able to reproduce in a development environment.


Are you serious ? You didn't know how to reproduce this without that video ?

HAHAHAHAHAHAHAHAHA



Miklas Laces
A.N.A.R.C.H.I.C.A
Posted - 2010.08.25 07:06:00 - [153]
 

Originally by: CCP Oveur
Edited by: CCP Oveur on 24/08/2010 10:07:51
No. What you might have missed in previous blogs and might not be as clear here is, that as big as this sounds, we have months of investigation, implementing fixes and testing left in the war against lag. This was a big low hanging fruit we saw. Like a pineapple. The size of the moon.


"Need for speed" started years ago, but it failed, lag is worse now then when the program begun, despite many lag-generator systems beeing changed (number of drones nerf, scanner nerf, warp-to-zero, fleet health bars removed, and so on).

The only time when it was looking to get a bit better was in 2009, but then Dominion came..

I understand that there were 10k concurrent players in 2005 while now you have 50k, but hardware is also more powerful now and you have way more resources then 5 years ago.

And it's not only fleet fights, the entire game is way more laggy and slower. Look at a video from 3 years ago and see how fast it was to undock from a station or jump through a gate.

p.s.
Also very old issues are still around: if you undock with a new ship it takes twice the time compared to undocking with a ship you have already undocked previously in that system. Is this by design ? It's a bug ? Will we know one day ?

Sedilis
Lead Farmers
Kill It With Fire
Posted - 2010.08.25 07:56:00 - [154]
 

Best dev blog ever... very interesting.

But essentially you're saying it was fine without drones... soooo if we get rid of drones it will work! Aw sucks to you be you Gellente Laughing

Vuk Lau
4S Corporation
Morsus Mihi
Posted - 2010.08.25 10:46:00 - [155]
 

Originally by: CCP Explorer
My dear Vuk Lau, you were present at both those CSM meetings, with CCP Oveur and then with CCP Atlas, and you have the full context of this remark and the full details. This disappoints me. Sad

For everyone else: At the meeting with CCP Oveur then a claim was made that EVE was in its worst state ever. I rejected that such a general statement could be made and in the second meeting presented to the CSM the overall, general health of EVE. We emphasized at the second meeting that although overall general health was in good shape then we acknowledged that there were problems in 0.0 fleet fights. We had already been working on those issues at the time of the CSM meeting and have continued that work with reinforced vigour since then.


My dear CCP Explorer. 1st to make things clear, I am almost always adding a pinch of troll in my posts esp. on Mondays, but I really found your statement back then lets say...strange, cause even if it is true (and it was from your standpoint) game was never in worse condition (if we put aside week after deploying every patch) for me and for the people I play and communicate with. That just shows how detached your metrics are with actual User experience which further leads to the conclusion that metrics are not.....erm...that good in judging the state of your product. I hope you didnt took this personally, and you will not remove me as friend on facebook because then I will not be able to stalk you like I stalk CCP Oveur.


Camios
Minmatar
Sebiestor Tribe
Posted - 2010.08.25 10:48:00 - [156]
 

Originally by: Miklas Laces
Quote:
At the Council of Stellar Management (CSM) summit in June, the CSM made an excellent presentation of the issues around fleet-fight lag, specifically about the issues of modules becoming ‘stuck' and the workarounds used by players. This provided useful insight into a problem which, until recently, we'd never been able to reproduce in a development environment.


Are you serious ? You didn't know how to reproduce this without that video ?

HAHAHAHAHAHAHAHAHA



QFT. It may mean that despite this thread, people at CCP don't join fleet battles, or the ones that do are not listened to.

The knowledge in that video is pretty much the basis of fleet warfare, something that every 0.0 BS pilot know.


I would like to think that CCP is just very polite and they know what is going on, but they want to encourage player participation in solving this problem and for that reason they said that the presentation is interesting.

Originally by: Miklas Laces

Also very old issues are still around: if you undock with a new ship it takes twice the time compared to undocking with a ship you have already undocked previously in that system. Is this by design ? It's a bug ? Will we know one day ?



May you refer to the session change timer due to ship change? It adds 30 seconds to wait before undocking.

Trebor Daehdoow
Gallente
Sane Industries Inc.
Posted - 2010.08.25 11:11:00 - [157]
 

Originally by: CCP Atropos
That said, we're always after more people bug reporting, simply so we can track how prevalent a problem is.

IMHO the current bug-reporting system actively discourages people from reporting bugs, because the process is so time-consuming and frustrating. Furthermore, the non-public nature of the bug-reporting system causes massive duplication of effort by players and extra, pointless clerical work by the BH team. I have no doubt that many bug reports that contain some crucial nugget of information get lost in the pile because they didn't get past the "send us logs" and "could not replicate" filters.

Just as sunlight is the best disinfectant, it is also the best insecticide. EVE should have a public bug-reporting system, implemented using an open-source tool like Bugzilla, with an option for private reporting of exploits. That way players can check to see if someone else has reported a bug, and add information when they find edge cases and replication methods -- not to mention append screenshots and videos.

Having a public bug-reporting system will occasionally cause embarrassment. But it will pay off many times over in reduced bug-hunting effort.

Make bug-hunting fearless, transparent, excellent and a united effort between players and CCP.

Lev Aeris
United Amarr Templar Legion
Fidelas Constans
Posted - 2010.08.25 12:14:00 - [158]
 

Awesome to see somebody doing some quality work on the lag. I'd love to see an entire expansion of this nature.

EVE: FIXANIS

Get all those incarna devs to stop playing with dolls and hammer out all the existing content. ;)

This List isn't getting any shorter.

Ollodem
Posted - 2010.08.25 12:42:00 - [159]
 

Originally by: Trebor Daehdoow
Originally by: CCP Atropos
That said, we're always after more people bug reporting, simply so we can track how prevalent a problem is.

IMHO the current bug-reporting system actively discourages people from reporting bugs, because the process is so time-consuming and frustrating. Furthermore, the non-public nature of the bug-reporting system causes massive duplication of effort by players and extra, pointless clerical work by the BH team. I have no doubt that many bug reports that contain some crucial nugget of information get lost in the pile because they didn't get past the "send us logs" and "could not replicate" filters.

Just as sunlight is the best disinfectant, it is also the best insecticide. EVE should have a public bug-reporting system, implemented using an open-source tool like Bugzilla, with an option for private reporting of exploits. That way players can check to see if someone else has reported a bug, and add information when they find edge cases and replication methods -- not to mention append screenshots and videos.

Having a public bug-reporting system will occasionally cause embarrassment. But it will pay off many times over in reduced bug-hunting effort.

Make bug-hunting fearless, transparent, excellent and a united effort between players and CCP.


I strongly agree with this.
I reported 3 bugs in my eve-time, 2 said yeah we know already. And in both cases i couldnt see what exactly had been reported before and if i could maybe add anything helpfull.

But on an unrelated note:
I allways thought about posting this in one of these threads nut allways way too afraid as i actually dont really understand what is going on and i cant imagine that my uneducated idea might actually be one of the problems leading to lag but whatever, in one of the last devblogs it was said that the problem might as well be something stupid.

As i understand it, lag has allways been worse in fw and general lowsec (maybe only shining in fw as lowsec seldom has huge fleetfights otherwise).
And it got worse in 0.0 with Dominion right?
What i allways wondered, how is in lowsec checked for sec standing increases or decreases? As i would imagine that some system that checked for every other shot a player fires wether this shot was legal or not and wether it decreases his sec standing or nets him a GCC, might scale horrible. Especially as the targets sec status might change in between as well because it fires too. To me who only writes small simulation programs in MatLab, this sound really scary.

And if this might really be a small problem, could this somehow happen in 0.0 too? Might the system node have to check for each fighting player in the system wether it will increase the systems millitary or whatever index? Probably not, but, yeah, i allways wanted to point this out. ^^

Cor Aidan
Shore Leave
Posted - 2010.08.25 12:50:00 - [160]
 

Very interesting and informative blog. It's quite interesting to get a small peek into the architecture of the code, and while I could jump on the opinion bandwagon about one technology versus another, instead I'll just say this to all the aspiring engineers of any type out there:

Remember kids, decisions that seem very reasonable under certain sets of assumptions often have "interesting" side effects or consequences when those assumptions are violated. Also remember that assumptions are always violated during the course of development on any project ("Hey, this will be enough paint to paint this room...oops!")

In this specific case, choice of programming tools (interpreted vs compiled) and general task architecture (cooperative vs preemptive vs real-time) all have tradeoffs: team skill, time to develop, ease of maintenance, scalability, performance, and even things like toolchain costs must all be considered - not just "hey I heard this technology is cool and awesome let's use that!".

Kudos all around for the increased communication and also to the folks that are allowing those pesky engineers to have the correct tools for the job!

Raimo
Genos Occidere
HYDRA RELOADED
Posted - 2010.08.25 13:05:00 - [161]
 

Edited by: Raimo on 25/08/2010 13:04:59
Originally by: Trebor Daehdoow
Originally by: CCP Atropos
That said, we're always after more people bug reporting, simply so we can track how prevalent a problem is.

IMHO the current bug-reporting system actively discourages people from reporting bugs, because the process is so time-consuming and frustrating. Furthermore, the non-public nature of the bug-reporting system causes massive duplication of effort by players and extra, pointless clerical work by the BH team. I have no doubt that many bug reports that contain some crucial nugget of information get lost in the pile because they didn't get past the "send us logs" and "could not replicate" filters.

Just as sunlight is the best disinfectant, it is also the best insecticide. EVE should have a public bug-reporting system, implemented using an open-source tool like Bugzilla, with an option for private reporting of exploits. That way players can check to see if someone else has reported a bug, and add information when they find edge cases and replication methods -- not to mention append screenshots and videos.

Having a public bug-reporting system will occasionally cause embarrassment. But it will pay off many times over in reduced bug-hunting effort.

Make bug-hunting fearless, transparent, excellent and a united effort between players and CCP.


Agreeing 100%. This would have so many benefits.

Yeay Fritg
Caldari
Confrerie de Kaedri
Cluster Of Rebirth
Posted - 2010.08.25 13:37:00 - [162]
 

Hello,

When a real Dev Blog about 'features bugs' not lag effect ?

When will you solve the bugs you have identified in the Dominion features e.g. Infrastructure Hub it also impact all the 0.0 ?

Why not answering this point ?

Good to know you worked on the Lag but it's not a Bug, it's an effect.

When a Backlog CCP Dev Blog or Position ?

Yeay

Ranger 1
Amarr
Ranger Corp
Posted - 2010.08.25 14:43:00 - [163]
 

Edited by: Ranger 1 on 25/08/2010 14:43:15
Originally by: Ban Doga
Edited by: Ban Doga on 25/08/2010 05:39:46
Originally by: Ranger 1
What I find surprising about these blogs is that they had to be written in the first place.

What has been lined out is simply a brief glimpse into the methodology of any company in the gaming industry (although CCP has several unique, and stellar concepts that outclass their competition).

For some reason the EVE forum community feels it is their right to have the senior management and development team come and personally hold their hand, explaining every last detail of what they are doing and why (which any idiot should already have a general grasp of), and then argue with them in a egocentric attempt to show them the "right way to do it".

Rolling Eyes

Throw in a liberal helping of "we know you said this, but we all agree you really meant that" nonsense and you end up with the most myopic, rabble rousing, self centered collection of misfits it has been my displeasure to interact with.

The only positive thing to come out of this sorry mess is that I was pleasantly surprised that CCP took you seriously enough to give you more insight into their development process than any other gaming company (or any type of software company) has to my knowledge. Of course, they were met with condescension and general idiocy but hey, that appears to be the norm on these boards.

Now if you can raise the general level of response to something higher than you would expect from your average 13 year old muppet you might actually deserve the respect you have been given.



No one is forcing you to read the dev blog threads or the player comments.
If you find this to be such a displeasure you can just steer clear of them.

I assume you are old enough to find out what you like and dislike and provide yourself with an environment that suits you.
Viewer discretion is adviced...


Thanks for the advice Ban, but I'm right where I think I need to be.

I know the truth hurts, but that doesn't change the fact that you need to hear it occasionally. Wink

You may now go back to sticking your fingers in your ears and singing "lalalalala".

Miklas Laces
A.N.A.R.C.H.I.C.A
Posted - 2010.08.25 14:46:00 - [164]
 

Originally by: Camios
May you refer to the session change timer due to ship change? It adds 30 seconds to wait before undocking.


No. Do this:

Log in, click "undock", in whatever ship you are, take note of how much time it takes to actually undock and load (you will be surprised how much time passes from the moment you click undock to the moment you are in control of your ship in space).

Then dock and undock a second time, without changing ship. Take note of how much time it takes to undock and you will see it's much shorter.

Dock again, wait 30 seconds, change ship, undock, take note of the time it takes, it's same as first undock again.

CCP GingerDude

Posted - 2010.08.25 15:43:00 - [165]
 

Check out this guys work. It's awesome:

http://knol.google.com/k/upsideyourhead/chapter-i-ship-motion-in-eve-online/2mdavnicxps8v/4

Originally by: Axemaster
Originally by: CCP Veritas
Originally by: Axemaster
Btw, I don't know if the devs are still reading this thread, but I was still hoping for an answer to my question back at post 38.


Yup, we are. I thought I handled post 38 in post 48. Did I miss something?


My bad, I meant post 56. Embarassed

Solra Wolfe
GunStars
Posted - 2010.08.25 17:12:00 - [166]
 

Edited by: Solra Wolfe on 25/08/2010 17:13:13

Originally by: CCP Atropos
Originally by: Caius Sivaris
What bothers me immensely is that it took the CSM showing you a video for the specific problem (weapon cycling issues, not lag in general in all its manifestation) being worked on. The workaround for this specific bug (manual weapon cycling) has been discussed openly on forums since at least 2008, so the hint was right there. I would however bet good isks that the issue never made it to a dev because the bug reports were relentlessly filtered, because yes, a reproduction case was hard to impossible to find.

The bug hunters discarding bug report are shielding the devs from the truth and doing the game a disservice. I'm afraid it's the balls of destiny being discovered by a player, discarded and rediscovered independently by a dev all over again.

You would be surprised at how many bug reports have nothing more for reproduction steps than "just hold a fleet fight and have some guys jump through a gate". I've seen people posting on the forums claiming that they've just submitted a bug report with pages of information on how a bug occurs and the cure for world hunger, only to look at the bug report and find it contains nothing more than the example I gave above.

The CSM's video however works wonderfully; it clearly shows a bunch of things, with a wealth of client activity and the effects that are seen when a particular bug presents itself, including all the little things that people forget to mention in bug reports that are crucial to fixing the problem! I would even go as far as to suggest all bug reports have handy videos attached detailing the ingame effect of a bug; they've proven immensely useful.

That said, we're always after more people bug reporting, simply so we can track how prevalent a problem is.


The problem with bug reporting is that often it's impossible to reproduce the bug, so it then becomes pretty much useless from a bug hunter's point of view. In the case of lag, it is in fact possible to reproduce the problem by saying "jump in a fleet of 100 ships and start fighting". I think what Caius was trying to say is why did it take a video from the CSM about manual cycling for you guys to understand the problem, when this has been a known work around for a long time. Does no one at CCP play the game at the fleet combat level? One would think that somebody does, and would have been able to do a video of their own to show to their devs wtf is going on with this ****.... Rolling Eyes

CCP Explorer

Posted - 2010.08.25 19:58:00 - [167]
 

Originally by: Vuk Lau
Originally by: CCP Explorer
My dear Vuk Lau, you were present at both those CSM meetings, with CCP Oveur and then with CCP Atlas, and you have the full context of this remark and the full details. This disappoints me. Sad

For everyone else: At the meeting with CCP Oveur then a claim was made that EVE was in its worst state ever. I rejected that such a general statement could be made and in the second meeting presented to the CSM the overall, general health of EVE. We emphasized at the second meeting that although overall general health was in good shape then we acknowledged that there were problems in 0.0 fleet fights. We had already been working on those issues at the time of the CSM meeting and have continued that work with reinforced vigour since then.
My dear CCP Explorer. 1st to make things clear, I am almost always adding a pinch of troll in my posts esp. on Mondays, but I really found your statement back then lets say...strange, cause even if it is true (and it was from your standpoint) game was never in worse condition (if we put aside week after deploying every patch) for me and for the people I play and communicate with. That just shows how detached your metrics are with actual User experience which further leads to the conclusion that metrics are not.....erm...that good in judging the state of your product.
I think there is a perfect match there. I rejected the general statement and provided the context and background of my argument, while at the same admitting that there were specific problems in 0.0 fleet fights, which matches your statement above about user experience. The truth is complicated, it has layers and depth.

Dr Lebroi
Posted - 2010.08.26 00:27:00 - [168]
 

Originally by: Sedilis
But essentially you're saying it was fine without drones... soooo if we get rid of drones it will work! Aw sucks to you be you Gellente Laughing


Pesky drones.... Time for broad strokes methinks, cut out drones and drone stuff, make isk and skill point reimbursement as necessary, you can do this while killing off learning skills and reimbursing those points. No drones, then Gallente suck ass (moar ass) for PVP, missions etc - solution? All Gallente die in Gallente only space disease - Gallente get to re-roll a new character with their SP, you won't even have to fix blasters......

Megan Maynard
Minmatar
Navigators of the Abyss
Posted - 2010.08.26 00:56:00 - [169]
 

Originally by: CCP Veritas
Originally by: Ix Forres
Great facial expression, good dev blog. How many major fixes have come about since the introduction of the thin client platform, and when was this made internally available? What were your methods of debugging EVE before it was available?


Thanks~

The thin clients matured enough for usage roughly a month ago, and while it took a fair bit of time to get them to do useful things, much has come from them already. It's hard to say how many major fixes have come from their usage so far, since that's dependent on your definition of major. Certainly most of what I talk about in this blog was made possible in the timeframe it happened via thin clients, and there are some nice optimizations coming down the pipe as well that originated in controlled thin client setups.

Before the thin clients we did load testing by getting as many warm bodies to log in to test servers, both internally and with the public mass tests. This type of testing is fantastic for finding things you didn't think to test, but not very useful for the kind of specific drill-down testing that I've been talking about above. Removing the need to use mass tests for that purpose has been very liberating, both in that we can do more of it and that we can focus on higher level information during mass tests.


It's nice to see you guys getting into the realm of what us engineers do quite often. (Use smaller simulations to gauge performance of a larger system.)

A couple questions:
1. Do you use thin clients on tranquility before/after patches to verify function? If not, look into it. It's a great way to run thousands of tests without someone needing to be there. (Because you can simply look at the results as you have time to catch the little things.)

2. Are there plans to incorporate a gui specifically for running multiple thin clients all at once? It sounds weird, having a gui to run another set of code that controls essentially another gui, but I like to think of it as the "turbo" button on my old N64 3rd party controller. Very Happy

CCP Veritas

Posted - 2010.08.26 09:55:00 - [170]
 

Originally by: Megan Maynard
1. Do you use thin clients on tranquility before/after patches to verify function? If not, look into it. It's a great way to run thousands of tests without someone needing to be there. (Because you can simply look at the results as you have time to catch the little things.)


We don't yet have that kind of regression testing framework for them yet, but it's absolutely on the table. Of course, thousands of tests don't write themselves, so it will take time.

Originally by: Megan Maynard
2. Are there plans to incorporate a gui specifically for running multiple thin clients all at once? It sounds weird, having a gui to run another set of code that controls essentially another gui, but I like to think of it as the "turbo" button on my old N64 3rd party controller. Very Happy


I regularly command hundreds of them at the same time; I'd go mad if I had to execute commands on them separately =)

Yeay Fritg
Caldari
Confrerie de Kaedri
Cluster Of Rebirth
Posted - 2010.08.26 12:46:00 - [171]
 

Sorry,

But CCP I just want a BackLog Bug Dev Blog !

Do you read the forums or will you moderate all our quality request if not in line with your marketing plan ?

Yeay

Leena Raphael
Posted - 2010.08.26 14:36:00 - [172]
 

Originally by: Yeay Fritg
Sorry,

But CCP I just want a BackLog Bug Dev Blog !

Do you read the forums or will you moderate all our quality request if not in line with your marketing plan ?

Yeay



Oh, just shut up.


Jees... Rolling Eyes

Yeay Fritg
Caldari
Confrerie de Kaedri
Cluster Of Rebirth
Posted - 2010.08.26 14:59:00 - [173]
 

Originally by: Leena Raphael
Originally by: Yeay Fritg
Sorry,

But CCP I just want a BackLog Bug Dev Blog !

Do you read the forums or will you moderate all our quality request if not in line with your marketing plan ?

Yeay



Oh, just shut up.


Jees... Rolling Eyes


lol

Fantome
Sniggerdly
Pandemic Legion
Posted - 2010.08.26 19:24:00 - [174]
 

Originally by: Miklas Laces
Quote:
At the Council of Stellar Management (CSM) summit in June, the CSM made an excellent presentation of the issues around fleet-fight lag, specifically about the issues of modules becoming ‘stuck' and the workarounds used by players. This provided useful insight into a problem which, until recently, we'd never been able to reproduce in a development environment.


Are you serious ? You didn't know how to reproduce this without that video ?

HAHAHAHAHAHAHAHAHA





I must say than i got the same reaction than Miklas (o/mate !). We are using it (disable autorepeat / cycling guns) in fleet since years.. have you never played your own game on tranquility ?? Rolling Eyes

Anyway nice blog and good to see we could get some nice improvement.. soon (tm) Razz

Samuel Miner
Caldari
Perilous Expedition
Posted - 2010.08.28 16:24:00 - [175]
 

Edited by: Samuel Miner on 28/08/2010 16:25:15

After reading this particular blog I am a little concerned, and late in reading it because of uni starting.

Concerned though because it seems the system, even if completely fixed, bug free with optimum yields in place for priority tasks to get processed, will still only be able to scr4pe by in a heavy load situation.

My question is if you had the chance to rewrite all the functions for combat (or maybe even deeper in the core systems) from the ground up, couldn't you design it more efficiently. It seems, once again, you are searching for just enough of a fix to squeak by until the next round of rage lag threads.

ceaon
Posted - 2010.08.29 17:13:00 - [176]
 

Originally by: CCP Oveur
Dunbar's number

since you bring this up and there are higth chances that you have read The Tipping Point what you can comment about the Broken windows theory because you get this thing on EvE and because few broken windows the :eve is dying: have reached a tipping point.


did you talk whit Bill Gore how they scale their company around dungar's number ? also is this the reason why you open offices on new locations ?

also i fixed ur linkRazz

Pellegro
Posted - 2010.08.31 04:58:00 - [177]
 

Edited by: Pellegro on 02/09/2010 00:38:06
Originally by: CCP Oveur
we believe in Dunbar's number


Curious. A guy named David Wong wrote an interesting little bit about Dunbar's number, calling it "The Monkeysphere". That was back in 2007.

Then we had the player, "The Monkeysphere" whose exploits of being able to fly in local undetected are now infamous. While some may have thought being caught using such a hack/exploit would result in a ban, "The Monkeysphere" is back (although his/her killboard is decidedly less impressive now).

Obvious alt is obvious.

Yehat Quan
Gallente
Posted - 2010.09.01 12:21:00 - [178]
 

Hello there,

awesome blog. However, I seriously disagree with your assesment about Width vs. Depth, aka, "I have to fix the algorithm before I change anything".

a) The algorithm, and the bug, results from using cooperative multitasking. It would probably be a completely different algorithm if it wasn't for Python.

b) from the tiny amount of physics you are doing, and the fairly small amount of players per node your nodes should not be CPU-bound. If anything, an MMO node that is not doing serious physics (and you don't) should be network bandwidth - bound. (For comparison: A server written in Javascript, (of all the languages!) (Node.js) has been shown to serve about 6500 requests a second. You have a maximum of 1500 simultaneous users in Jita. And I don't think you run your calculations every second for every ship, do you? )

b): C*O(n) in reality (as opposed to theoretical informatics) _can_ be larger then c*O(n^2) for appropriate C and n. Now, we are (hopefully!!!) not talking O(n-square) levels of suck here, and our n is -- if we take Jita as the worst case -- only 1500. If you can halve the work by changing the algorithm, but cut the execution time by a factor of 3 by simply switching to a different programming language, what do you think should have a higher priority? And we aren't talking about anything as little as factor 3... not nearly. But I'll say something about that later.

Now I understand why you guys use python and have a hard time switching:

Originally by: Cor Aidan
In this specific case, choice of programming tools (interpreted vs compiled) and general task architecture (cooperative vs preemptive vs real-time) all have tradeoffs: team skill, time to develop, ease of maintenance, scalability, performance, and even things like toolchain costs must all be considered


However, let me tell you a little story.

2005 we had an assignment in a bioinformatics class. It was "align two DNA sequences of a certain length a thousand times and calculate X". (The specifics evade me right now).
Quote:

Our tutor: "It takes about ten minutes to write the code. After that, you can go and have a coffee. Or a dinner."
My partner in crime (who is a mathematician): "How long does it run, exactly?"
Our tutor: "About 2 hours".
Me: "What programming language?"
Our tutor: "Python."


Me and my study partner look at each other, sit down and write the code in C. It takes us half an hour (three times longer then it took our tutor to write his.)

Prise question: how long does it run?

(drumroll)

12 seconds.

- No, I am NOT kidding
- No, I didnīt mean to write 12 minutes.
- No, we didnīt change the algorithm. It was EXACTLY the same.
- No, it didnīt thread or use multiple cores. We only had one. It was 2005.
- No, it didnīt leak memory (since it was pretty straight-forward and I am an industry programmer who knows his ****.)

(And if you are wondering, yes, it did produce the same result as the python code.)

What did I learn from that? Transporting an algorithm 1:1 from one language to another can give you a time gain of (2 hrs / 12 seconds) = 120*60/12=600x It can take 3x longer to write. In pure numbers, that adds up to a 200x gain. Multiply this by a factor of your choice reflecting how much you value developer time versus hardware costs.

Conclusion:
Of course, NPC behaviour, GUI interaction, etc, are all fine and good in python, since a graphics- or a story designer might look at it and understand it.
However, unless the python interpreter has been massively rewritten since 2005, Python is a toy language. Writing serious server-side _processing_ code in python is like trying to build a lego robot out of Barby parts. And the whole GIL thing doesnīt exactly make this any better. I would ask you to consider how much easier your life would become if you didnīt have to control the yielding behavior of your code by hand, versus how much time you might spend learning a new language (which, in itself, can be fun).

Qujulome
Amarr
Posted - 2010.09.02 17:03:00 - [179]
 

Originally by: Yehat Quan
Hello there

-snip-




CCP, hire this guy. Seriously.

CCP Veritas

Posted - 2010.09.03 17:35:00 - [180]
 

Originally by: Yehat Quan
<stuff>


Hi there.

I at first felt like going point by point with you, but I think there's a deeper point I can touch on that covers most everything.

I've been a professional C/C++ programmer for the better part of a decade, and was raised on it early. I know the performance difference between Python and C. I know, deeply and exactly, why that difference exists.

I've also done game development in C++, and am very familiar with the maintenance nightmare that such a codebase can become after many years of disjoint development by a shifting team with shifting focus. The tradeoff between speed of execution and ease of maintenance and extension is not as simple as you make it out to be.

There's a time for reworking code from a higher level language to a lower level language. It used to be dropping into ASM from C, now it's droping into C from something higher level. That time has always been after you've cut away the unnecessary computation and poor design models. From what I've seen so far with Eve, we're not at that point yet. If we had one little tight loop that was consuming most of the time, you'd have to hold me back from pulling it over to C-side. That's simply not the situation we're in, however.


Pages: 1 2 3 4 5 [6] 7

This thread is older than 90 days and has been locked due to inactivity.


 


The new forums are live

Please adjust your bookmarks to https://forums.eveonline.com

These forums are archived and read-only