/g/ - Technology

/g/ - Technology
Text Board: /tech/

Posting mode: Reply

[Return] [Bottom]

Attention 4chan extension/user script/archive developers: Some time in the next few days, we'll be rolling out a complete HTML rewrite of the imageboards.
The design will remain the same, but the underlying HTML/CSS is completely new, and validates HTML5/CSS3 (with some tweaks to account for cross-browser compatibility).

Please visit this thread to read more about the changes, and here to preview the code.

As a regular user, these changes should not affect you. You will need to update your 4chan browser extensions/user scripts when their maintainer updates them to be compatible with the changes.
The official 4chan Chrome extension will be ready to go when the updates happen, and 4chan X should be ready soon. We'll post more details on the day of the migration!

File: 1335613239.png-(71 KB, 1024x1024, hypertrollmarkuplanguage.png)

Anonymous 04/28/12(Sat)07:40 No.24511647

all of my why.

>>	Anonymous 04/28/12(Sat)07:42 No.24511664 One word The forced indentation of the HTML Thread over.

>>	Anonymous 04/28/12(Sat)07:42 No.24511665 You forgot to strip the whitespace.

>>	Anonymous 04/28/12(Sat)07:43 No.24511669 >495 bytes clog my link Get a better internet connection

>>	Anonymous 04/28/12(Sat)07:44 No.24511677 >href="javascript:" >HTML5

>>	Anonymous 04/28/12(Sat)07:44 No.24511682 HTML5 is bloat.

Anonymous 04/28/12(Sat)07:46 No.24511694

>>24511665
>>24511664
They're all live samples. This is exactly the stuff what we're going to get unless they change it (and I hope they do.) The new sample has tabs for the whitespace and removing them is not going to make much difference. ~100 bytes at most.

>>	Anonymous 04/28/12(Sat)07:46 No.24511695 This isn't about reducing bytes. At any rate, this will be much faster as having that many tables slows everything down. See: Stickies, /vg/ threads, etc.

>>	Anonymous 04/28/12(Sat)07:46 No.24511701 Yeah, it sure is tedious waiting for all those bytes to come in at 2400 baud.

>>	Anonymous 04/28/12(Sat)07:47 No.24511710 >>24511669 goddamn fatmster

>>	Anonymous 04/28/12(Sat)07:47 No.24511712 >minify class names >use b for block and q for inline elements exclusively for shorter code

>>	LAwLz !LAwLzaWU1A 04/28/12(Sat)07:48 No.24511718 >Think more code == bad >removing the white space on two of them but not the third one. >Think you can't use both HTML5 and JS on the same page.

>>	Anonymous 04/28/12(Sat)07:48 No.24511719 >>24511669 Tell moot that. It's not much for each user but remember how many users 4chan has.

>>	Anonymous 04/28/12(Sat)07:48 No.24511720 Parses quicker. Nobody gives a fucking about raw file size. This is like faggots thinking more LOC = bloat

>>	Anonymous 04/28/12(Sat)07:48 No.24511726 File: 1335613713.jpg-(19 KB, 331x240, negromongler.jpg) >>24511712 > don't care about semantics ISOCKYSDRAWERGIGGIDY

>>	Anonymous 04/28/12(Sat)07:49 No.24511734 >>24511720 >>24511695 These. This will be a huge benefit to everyone. Stop whining.

>>	lachs0r 04/28/12(Sat)07:49 No.24511743 You forgot the part where 4chan’s servers use gzip compression.

Anonymous 04/28/12(Sat)07:52 No.24511779

>>24511695
Are you living right next to 4chan's servers?

The network bandwidth is the bottleneck here, and if it isn't then you really need to upgrade. That monster thread in /b/ right now barely registered any CPU usage when I tried opening it. Most of the time was spent waiting for the network, loading all the images. And this was on an 800MHz Pentium III.

>>	Anonymous 04/28/12(Sat)07:53 No.24511789 >>24511779 >network bandwidth >bottleneck Move away from Nigeria.

>>	Anonymous 04/28/12(Sat)07:54 No.24511800 >>24511720 How much quicker? A few microseconds? Compare that to the milliseconds it takes to get the page data over the network in the first place.

>>	Anonymous 04/28/12(Sat)07:54 No.24511807 >>24511779 Try being a Chrome user.

>>	Anonymous 04/28/12(Sat)07:56 No.24511831 >>24511647 YEah, he should get rid of all that whitespace. Don't parsers get rid of them anyway? I'm pretty sure removing it would help programs parse easier

>>	Anonymous 04/28/12(Sat)07:57 No.24511841 HTML 5 is the Ubuntu of GNU/Linux, slow and bloated.

>>	Anonymous 04/28/12(Sat)07:58 No.24511845 >>24511779 It took half a minute for Firefox to open that thread. wget: 2,733,868 1.34M/s in 2.0s

>>	Anonymous 04/28/12(Sat)07:58 No.24511846 content is compressed while sending try zipping them for fairer comparison

>>	Anonymous 04/28/12(Sat)07:58 No.24511847 >>24511822 It's 495 bytes more per post, not per whole page. The stylesheet is probably getting cached so it won't make much difference there.

>>	lachs0r 04/28/12(Sat)07:58 No.24511850 >>24511831 See >>24511743

>>	Anonymous 04/28/12(Sat)07:58 No.24511851 >>24511841 >HTML 5 is the Ubuntu of GNU/Linux >the Ubuntu of GNU/Linux

Anonymous 04/28/12(Sat)07:59 No.24511861

>>24511800
You're not considering the reduction in the stylesheets, neither browser caching. Also, it's not that much more. If you'd bothered to profile, you would see that you use most of your bandwidth to get the image thumbnails anyway.

To conclude: You're full of shit.

>>	Anonymous 04/28/12(Sat)07:59 No.24511865 >>24511841 >HTML 5 >GNU/Linux HTML5 is a distro now?

>>	Anonymous 04/28/12(Sat)07:59 No.24511867 >>24511851 >>24511682 what are you faggots talking about. I see no html5

>>	Anonymous 04/28/12(Sat)08:00 No.24511870 current doctype >none newhtml doctype ><!DOCTYPE html> BLOATED PIECE OF SHIT

>>	Anonymous 04/28/12(Sat)08:00 No.24511871 >>24511865 >>24511851 Autistic kids in charge of analogies

>>	Anonymous 04/28/12(Sat)08:01 No.24511882 >>24511871 This whole thread is autism. It's pointless bitching about something we can't even judge yet. Fuck moot should have gotten rid of tables years ago.

>>	Anonymous 04/28/12(Sat)08:02 No.24511887 >>24511847 >>24511847 Anyway, you're using most of the bw to get image thumbnails anyway, that is the true bottleneck. Also, see >>24511743

>>	Anonymous 04/28/12(Sat)08:02 No.24511891 >>24511871 You don't know how to put together an analogy.

>>	LAwLz !LAwLzaWU1A 04/28/12(Sat)08:02 No.24511893 >>24511789 Ehhh.... Seriously? You're not bottlenecked by connection speed when you load a webpage?

>>	Anonymous 04/28/12(Sat)08:02 No.24511895 File: 1335614560.jpg-(22 KB, 490x367, what_cracker.jpg) >>24511865 Yes. u mad?

>>	Anonymous 04/28/12(Sat)08:02 No.24511896 >>24511891 not the guy who made the analogy, but it was pretty fucking clear even for a retard.

>>	lachs0r 04/28/12(Sat)08:04 No.24511907 >>24511847 >It's 495 bytes more per post gzip compression removes most of that redundancy. Numbers: >>>/newhtml/ uncompressed: 55051 bytes gzip (default compression level): 9867 bytes Even at the lowest compression level, gzip still achieves 11854 bytes.

Anonymous 04/28/12(Sat)08:05 No.24511914

>>24511845
Try Chrome.

>>24511882
>implying tables are bad just because the w3c says so
Protip: it all renders to the same thing in the end, and a table with 1 row and 2 columns isn't going to make much difference compared to the same set of divs, only it takes more effort to get the divs in the right place.

It's a moot point.

>>	Anonymous 04/28/12(Sat)08:05 No.24511915 >>24511893 You're Swedish. Get a better uplink.

Anonymous 04/28/12(Sat)08:06 No.24511929

>>24511914
When you're loading 500+, and using 4chan X, and styling everything on top of it it absolutely makes a difference. At that point it becomes a necessity to use fixed tables and that only slightly cuts down on the poor performance.

Tables have no reason for being in threads outside of the post form.

>>	Anonymous 04/28/12(Sat)08:07 No.24511937 >>24511914 Former Chrome user here, I have to use Nightly just to browse /vg/ threads because Chrome performs so poorly.

>>	LAwLz !LAwLzaWU1A 04/28/12(Sat)08:07 No.24511939 >>24511915 Do you have any idea of how much faster a computer is compared to the network connection? You'd need to have a connection directly to the server with a couple of hundreds of GB/s to not be bottlenecked by the connection.

Anonymous 04/28/12(Sat)08:09 No.24511949

jesus FUCK i hate you stupid autistic faggots

>hur durr get beter connection

you are a fucking moron and should kill yourself. the fact that the filesize is NOT an improvement doesn't mean YOU should need to upgrade

you stupid beta faggots holy fuck

i'm going to fill half your house with dogshit and when you see you have less room in the house I'll tell you to get a bigger fucking house

faggots

>>	Anonymous 04/28/12(Sat)08:09 No.24511951 >>24511939 No, no 4chan thread is even close to be anywhere near 1 GB.

>>	Anonymous 04/28/12(Sat)08:10 No.24511954 >>24511949 You seem visibly upset over something that is rather trivial. May I suggest taking a break from the internet and relaxing a bit?

>>	LAwLz !LAwLzaWU1A 04/28/12(Sat)08:11 No.24511964 >>24511951 I don't see what that has to do with anything. I was just talking about the ridiculous network connection speed you would need in order to remove the network bottleneck when loading a webpage.

Anonymous 04/28/12(Sat)08:11 No.24511966

ITT: People who just found 4chan this month.

4chan has had stability and bandwidth issues for years.

Posting is still down all the time due to ddos, though cloudflare makes reading boards a bit more reliable.

Doubling the bandwidth per page is just idiotic when your servers are so shitty and unreliable as well as being the target of frequent attacks by skiddies

>>	Anonymous 04/28/12(Sat)08:12 No.24511970 >>24511966 You're so retarded it hurts.

>>	Anonymous 04/28/12(Sat)08:12 No.24511971 File: 1335615149.png-(523 KB, 689x689, brilliant.png) >>24511914 >It's a moot point.

>>	Anonymous 04/28/12(Sat)08:12 No.24511973 >>24511058 Why have HTML5 when they aren't even going to make the effort?

>>	Anonymous 04/28/12(Sat)08:13 No.24511978 >>24511949 >i'm going to fill half your house with dogshit and when you see you have less room in the house I'll tell you to get a bigger fucking house >faggots 14/6

>>	Anonymous 04/28/12(Sat)08:13 No.24511985 >>24511966 >Posting is still down all the time due to ddos, though cloudflare makes reading boards a bit more reliable. This isn't 2008 anymore...

>>	Anonymous 04/28/12(Sat)08:14 No.24511989 >>24511966 What part of g-zip compression don't you understand?

>>	Anonymous 04/28/12(Sat)08:14 No.24511990 >>24511669 >sacrificing load time for whitespaces I never see

>>	Anonymous 04/28/12(Sat)08:15 No.24511996 Why the fuck do you think the whitespace will be in the `production' version. It's just for development, you retards.

Anonymous 04/28/12(Sat)08:15 No.24511997

>>24511939
>I don't see what that has to do with anything. I was just talking about the ridiculous network connection speed you would need in order to remove the network bottleneck when loading a webpage.
You obviously don't know what you're talking about. No 4chan thread, counting all thumbnails and stylesheets and everything is close to be big enough for my VDSL-line to become the bottleneck. The 0.2 seconds it takes to get all (the rest is spent on rendering) is due to network latency, not throughput.

>>	Anonymous 04/28/12(Sat)08:16 No.24512006 >>24511996 >This It's only up and in a readable format to help developers you autistic fucks.

>>	Anonymous 04/28/12(Sat)08:16 No.24512015 >>24512006 >>24511996 If you retards (that is all of you that claim whitespace is a problem) actually bothered to view the raw HTTP responses, you would see that it is gzip compressed.

LAwLz !LAwLzaWU1A 04/28/12(Sat)08:17 No.24512021

>>24511997
Are you seriously saying that your connection is faster than your CPU and GPU? Sounds like it's time to update your computer if that's the case. Also, I count network latency as "network bottleneck" as well, since you know, it is related to your network connection.

>>	Anonymous 04/28/12(Sat)08:18 No.24512022 moot does one cool thing FINALLY for 4chan which will be amazing for userstylers, extension/script developers, and people who struggle with performance in stickies, and /g/ has to go and shit all over it with negativity. Fuck you guys.

>>	Anonymous 04/28/12(Sat)08:18 No.24512027 >>24511989 Don't forget compression increases server load too, and having it need to compress more data is an overall increase of resource usage.

Anonymous 04/28/12(Sat)08:18 No.24512028

>>24511997
No, you're a fucking idiot.
Network throughput is on the order of a few megabytes per second, your computer is capable of gigabytes per second of data throughput (ram->ram) and hundreds of mega bytes per second (hdd->ram).
Network is definitely the bottleneck when displaying webpages.

>>	Anonymous 04/28/12(Sat)08:19 No.24512031 >>24512022 Why do you think he closed the complaint thread?

>>	Anonymous 04/28/12(Sat)08:19 No.24512032 >>24511695 >See: Stickies, /vg/ threads, etc. you know, when your userscript is inefficient, you could change that. when the code is bloated, it is out of our control.

>>	Anonymous 04/28/12(Sat)08:19 No.24512035 >>24511726 Is that Uncle Phil?

>>	Anonymous 04/28/12(Sat)08:19 No.24512038 >>24512021 >has no idea what bottleneck is >still argues about it People like you are the reason why threads get shat up so often. Stop replying to everything you see and just think for a fucking second.

>>	Anonymous 04/28/12(Sat)08:20 No.24512048 File: 1335615618.png-(45 KB, 190x250, 1331288578435.png) >inline javascript

>>	Anonymous 04/28/12(Sat)08:21 No.24512069 ></span> > > <span class="dateTime">04/27/12(Fri)23:29:30</span> > > <span class="postNum desktop"> > I think they are trying too hard.

Anonymous 04/28/12(Sat)08:21 No.24512070

>>24512031
>Why do you think he closed the complaint thread?
Because he was mostly looking for feedback from relevant people like Mayhem, who gave it to him already, and the thread had completely derailed into "Hurr why are you a tripfag omg moot put forced anon everywhere"

Anonymous 04/28/12(Sat)08:22 No.24512075

>>24512021
>Are you seriously saying that your connection is faster than your CPU and GPU?
No I'm not saying that, why are you being so dumb. Can't you read? I'm saying that 4chan threads are so compact that the DNS lookup + SSL handshake + HTTP requests + HTTP responses are no match for my uplink at all.

>Also, I count network latency as "network bottleneck" as well, since you know, it is related to your network connection.
Well, it is mainly due to propagation time and processing time in routers along the path + bridging in the DSLAM.

For fuck sake, wget uses 0.2 seconds to retrieve the HTML alone, just profile it for yourself.

LAwLz !LAwLzaWU1A 04/28/12(Sat)08:22 No.24512077

>>24512038
Okay, please define bottleneck for me. When your connection can only handle a few megabytes per second (with a good connection) and your computer can handle tens of gigabytes per second then it is pretty clear to me that the network connection is a bottleneck.

>>	Anonymous 04/28/12(Sat)08:22 No.24512079 >>24511997 YOU ARE SO FUCKING RETARDED HOLY SHIT

>>	Anonymous 04/28/12(Sat)08:22 No.24512081 I'm personally offeneded that 4chan hasn't got a RESTful API that it uses to construct pages on the client with node.js and mongodb comprising the backend. GET WITH THE TIMES MOOT

Anonymous 04/28/12(Sat)08:23 No.24512086

>>24512022
>people who struggle with performance in stickies
That's because they bloat up their browser with all sorts of extra shit. I bet even IE6 could beat all the other browsers at loading huge threads because all it's doing is (poorly) rendering HTML.

Keep drinking that HTML5 Kool-Aid.

>>	Anonymous 04/28/12(Sat)08:23 No.24512093 File: 1335615824.png-(132 KB, 512x384, 1260918629280.png) >>24512081 >node.js

Anonymous 04/28/12(Sat)08:24 No.24512102

>>24512075
>For fuck sake, wget uses 0.2 seconds to retrieve the HTML alone, just profile it for yourself.
And?
Do YOU not know what a bottleneck is?
Your computer would be capable of doing that in microseconds if it had local access, that is ORDER OF MAGNITUDE FASTER YOU IDIOT, ie the network is seriously bottlenecking your computers potential data throughput.

>>	Anonymous 04/28/12(Sat)08:25 No.24512109 >>24512077 >>24512028 >>24512079 A bottleneck is what requires the most resources (usually time) to complete it's task. MORE TIME IS SPENT ON RENDERING THAN ON NETWORK TRAFFIC, THERFORE RENDERING IS THE BOTTLENECK.

Anonymous 04/28/12(Sat)08:25 No.24512112

>>24512077
> it is pretty clear to me that the network connection is a bottleneck.
Jesus fuck, are you dense?

It's not a bottleneck if having a better connection wouldn't change a thing in the first place. The connection itself isn't fucking maximized by a single 4chan page, it's not bottlenecking anything.

>>	Anonymous 04/28/12(Sat)08:25 No.24512113 File: 1335615945.png-(95 KB, 247x302, 64.png) >>24512081

>>	Anonymous 04/28/12(Sat)08:26 No.24512116 >>24512102 MORE TIME IS SPENT ON RENDERING!!!! RENDERING IS THE BOTTLENECK

>>	Anonymous 04/28/12(Sat)08:26 No.24512121 >>24512102 See >>24512109

>>	Anonymous 04/28/12(Sat)08:27 No.24512124 >>24512081 I don't know what people's problem with node.js is, but I really wish there were an API for 4chan so external tools could access it better without having to fetch all the html

>>	Anonymous 04/28/12(Sat)08:27 No.24512129 >>24512028 Nope. The browser uses more time on rendering than the system uses on network traffic.

LAwLz !LAwLzaWU1A 04/28/12(Sat)08:27 No.24512131

>>24512075
I see what you're getting at, but a better connection (please note that better connection does not necessarily just mean higher bandwidth) would still reduce the load time, therefore it is the major bottleneck. Also, you can skip some of the steps you listed such as DNS lookup and SSL handshake if you want.

Anonymous 04/28/12(Sat)08:28 No.24512136

>>24512109
>A bottleneck is what requires the most resources (usually time) to complete it's task.
No, that is completely wrong.
A bottleneck is something that prevents other more powerful parts from doing their job, for example if you have a SLI GTX 680 with a P3 CPU, you wouldn't be able to use all the potential computing power of the GTX 690 because the CPU is bottlenecking it.
Same with network, your CPU, RAM and HDD are capable of MUCH MUCH faster data throughput than your internet, but they have to sit idle and wait for the network, ie a bottleneck.

>>	Anonymous 04/28/12(Sat)08:29 No.24512140 >>24512086 Probably doesn't help when scripts have to act upon 500+ posts for backlinks etc.

>>	Anonymous 04/28/12(Sat)08:29 No.24512142 >>24512131 >therefore it is the major bottleneck No it is not. When more time is spent on rendering, you faggot, rendering is the major bottleneck. See >>24512109

>>	Anonymous 04/28/12(Sat)08:30 No.24512148 >>24512129 Either your browser is retardedly slow or you need a hardware upgrade if rendering pages takes more time than getting them from the network.

>>	Anonymous 04/28/12(Sat)08:30 No.24512149 >>24511867 >I see no html5 ><!DOCTYPE html> You didn't look very far. That's on the first line.

Anonymous 04/28/12(Sat)08:30 No.24512151

>>24512136
>but they have to sit idle and wait for the network, ie a bottleneck.

But it's not the network that is the bottleneck, even if we had better networks the performance would be the fucking same. The transfer rate isn't maximized by a single 4chan page, it's not filling the whole network with it.

If anything, I'd say the rendering of the page (or the actual 4chan server) are way slower than retrieving a single page (in bytes) from the internet.

>>	Anonymous 04/28/12(Sat)08:30 No.24512156 >>24512121 >>24512142 see >>24512136 Also: http://en.wikipedia.org/wiki/Bottleneck

>>	Anonymous 04/28/12(Sat)08:31 No.24512161 >>24512148 Either your connection is retardedly slow or you need a better ISP if connecting to pages takes more time than rendering them.

LAwLz !LAwLzaWU1A 04/28/12(Sat)08:31 No.24512164

>>24512109
>A bottleneck is what requires the most resources (usually time) to complete it's task.
That's not really what a bottleneck is though. A bottleneck (when talking about computers) is when one component is limiting other components. In this case, your CPU and other computer components are bottleneck by the speed at which your computer receives data from the server.
Your CPU is much faster than the speed at which the network sends data over, therefore the connection between you and the server is the bottleneck.

Anonymous 04/28/12(Sat)08:32 No.24512167

>>24512136
Well, I my system is just as idle when waiting for the CPU and GPU to complete, isn't it?

Also, you're full of shit. The process is pipelined, the CPU begins rendering the document while retrieving the thumbnails. It still isn't done rendering when all is fetched.

>>	Anonymous 04/28/12(Sat)08:33 No.24512174 The bottleneck is rendering the thumbnails. /thread

>>	Anonymous 04/28/12(Sat)08:33 No.24512180 >>24512167 >The process is pipelined, the CPU begins rendering the document while retrieving the thumbnails. It still isn't done rendering when all is fetched. this

Anonymous 04/28/12(Sat)08:33 No.24512182

>>24512151
>But it's not the network that is the bottleneck
That's exactly what it means.

>Even if we had better networks the performance would be the fucking same
No, what the fuck are you talking about?
If you can get the 0.2s down to 0.000001s, you would have a significant speedup.

>The transfer rate isn't maximized by a single 4chan page, it's not filling the whole network with it.
You would need internet speeds in the range of 5GB/s to saturate RAM throughput, and a few hundred MB/s to saturate SSD speeds.

Anonymous 04/28/12(Sat)08:34 No.24512192

You fucking retards, you don't start rendering AFTER you retrieve everything.

You start rendering WHILE everything is getting streamed to the browser. And guess what? After everything is already delivered by the network, the browser is still rendering and adjusting the page.

Network bottleneck is non-existent.

>>	Anonymous 04/28/12(Sat)08:35 No.24512194 Know what really rustles my semantic pantaloons? When people talk about graphics cards bottlenecking performance.

>>	Anonymous 04/28/12(Sat)08:35 No.24512195 >>24512167 >Well, I my system is just as idle when waiting for the CPU and GPU to complete, isn't it? fuck off idiot

Anonymous 04/28/12(Sat)08:35 No.24512198

Are you retards still bitching about bottlenecks? Look, if transfer speeds are reduced and the code is left the same thats an improvement.
If the code is increased, but rendering time is reduced to a point where it outweighs the increase in transfer THAT is an improvement too.

Even if there are a minute amount of bytes extra to transfer the improvement in rendering will be a benefit to all.

>>	Anonymous 04/28/12(Sat)08:35 No.24512202 i seriously hope moot doesn't listen to these retards

>>	Anonymous 04/28/12(Sat)08:35 No.24512204 that's several HUNDRED bytes, I mean, can we really handle that kind of data throughput with present technology? I am really sensitive to picoseconds so this concerns me greatly

Anonymous 04/28/12(Sat)08:35 No.24512205

>>24512182
>If you can get the 0.2s down to 0.000001s, you would have a significant speedup.

A speedup in what? Download a few 4chan pages locally and stopwatch how much it takes for your browser to render them locally.

It would still be more or less the fucking same because it actually takes time rendering and not downloading few useless kilobytes of data. (don't forget caching too)

>>	Anonymous 04/28/12(Sat)08:36 No.24512208 >>24512167 >>24512180 >>24512192 Most obvious samefag shit I have seen in my life

Anonymous 04/28/12(Sat)08:36 No.24512209

>>24512167
>Well, I my system is just as idle when waiting for the CPU and GPU to complete, isn't it?
No, when CPU and GPU is working that is the definition of being non-idle.

>The process is pipelined, the CPU begins rendering the document while retrieving the thumbnails.
Doesn't matter how many pipes you use, network is STILL ORDER OF MAGNITUDE SLOWER THAN RAM AND HDD SPEEDS.
By doing things over the network you are seriously bottlenecking the potential data throughput of a computer.

LAwLz !LAwLzaWU1A 04/28/12(Sat)08:36 No.24512212

>>24512164
Look at it this way. Which is faster.
Loading a webpage from the hard drive (as in locally) or loading a webpage on a server in a another country.

If the local webpage loads faster, then it is the network connection which is bottlenecking you.
If both loads just as fast then the network is not limiting factor.

Very simple test which yields very accurate results.

>>	Anonymous 04/28/12(Sat)08:36 No.24512214 >>24512124 So do I, anon. So do I.

>>	Anonymous 04/28/12(Sat)08:36 No.24512216 It's fast enough for me to read the posts as I scroll down the page, so I'd call that good enough. Even with the new HTML that is unlikely to change, but the issue here is more with server resources and bandwidth.

>>	Anonymous 04/28/12(Sat)08:37 No.24512231 >>24512209 >Doesn't matter how many pipes you use, network is STILL ORDER OF MAGNITUDE SLOWER THAN RAM AND HDD SPEEDS. This. You fucktards.

>>	Anonymous 04/28/12(Sat)08:38 No.24512233 >>24512182 >If you can get the 0.2s down to 0.000001s, you would have a significant speedup. The improvement in rendering time is probably more significant than 0.2 seconds on large threads.

Anonymous 04/28/12(Sat)08:38 No.24512234

>>24512164
I'm not arguing your definition of bottleneck, but your statements are flawed.
Yes, the connection from the server -> you will be the data transfer bottleneck. However you seem to think this automatically equates to it being the slowest step in the process. It's not- the amount of data you receive is not automatically directly proportional to the amount of CPU time spent processing that data. A relatively small amount of data- like telling your computer to calculate pi to the 100,000th place- can take quite a while to process.

Anonymous 04/28/12(Sat)08:38 No.24512239

>>24512205
>A speedup in what?
Data throughput.

>It would still be more or less the fucking same because it actually takes time rendering and not downloading few useless kilobytes of data. (don't forget caching too)
It would be significantly faster, especially image have threads.
Ever wondered why the first time you view a thread it always takes longer? Because you have to wait for bunch of images to get downloaded from 4chan's shitty servers, after that most of them get cached and you have significantly faster page loads.

>>	Anonymous 04/28/12(Sat)08:39 No.24512252 >>24512239 >It would be significantly faster, especially image have threads. show me some proof. I don't really care that much to test it right now, but again, show me some proof and we might agree on something.

>>	Anonymous 04/28/12(Sat)08:40 No.24512255 >moot gets flamed at for being against image preloading >the boards get rewritten, making it easier for parsers to get the content >there is additional whitespace >people are okay with image preloading but freak out over some whitespace fuck you /g/

>>	Anonymous 04/28/12(Sat)08:40 No.24512257 >>24512233 Fuck your shit RIGHT OFF. Stop talking about other shit in an attempt to detract from the statement FUCK!

>>	Anonymous 04/28/12(Sat)08:42 No.24512279 >>24512257 Oh, I'm not talking about that bottlenecking garbage. I'm just saying that the probable improvement in rndering time will outweigh the extra data that needs to be downloaded.

>>	Anonymous 04/28/12(Sat)08:43 No.24512296 >div classes for fuck-everything Even a moron could style this.

LAwLz !LAwLzaWU1A 04/28/12(Sat)08:44 No.24512304

>>24512252
>show me some proof.
Do we seriously have to post proof that cached websites load faster now? Seriously? That's common knowledge.

Anyway, I think it's nice of moot to do improvements like these to 4chan. It proves that he still cares about us.
I am just arguing against those people who say your network interface is not a bottleneck when loading webpages.

>>	Anonymous 04/28/12(Sat)08:44 No.24512306 >>24512296 m00t.

Anonymous 04/28/12(Sat)08:44 No.24512310

Save the huge thread in /b/ to your hard drive.

Open it and time how long it takes to refresh compared to refreshing the one from the URL.

It took ~500ms for it to refresh locally, compared to over 5 seconds from 4chan's servers.

Rendering is a tiny amount of time.

Anonymous 04/28/12(Sat)08:45 No.24512319

>>24512304
>Do we seriously have to post proof that cached websites load faster now? Seriously? That's common knowledge.

1) If it's common knowledge, show some proof for it
2) It's not what I asked or what I'm arguing about. What does "load" mean to you? Load into memory? Rendering? If you have a very heavy page full of images and dynamic content, it matters jackshit where you're taking your data (be it on the network, locally or cached), you still need some time to render it in your browser.

>>	Anonymous 04/28/12(Sat)08:45 No.24512321 >>24512310 But I have Opera. So it takes nearly ~9000ms to fresh. Also, fuck moot.

>>	Anonymous 04/28/12(Sat)08:45 No.24512322 smaller is not always better.

>>	Anonymous 04/28/12(Sat)08:45 No.24512323 >>24512310 Yet its not how long it takes to load that freezes up browsers. Just stop your bitching and wait til the html conversion.

>>	Anonymous 04/28/12(Sat)08:46 No.24512333 i dont know what any of this means but how come other websites are fast EVERYDAY ALL THE TIME and 4chan STILL gets slow once in a while

>>	Anonymous 04/28/12(Sat)08:47 No.24512337 >>24512319 Its common knowledge that the earth has oxygen, yet you don't put someone on the spot to prove it has oxygen when proof requires tools and time. He doesn't need to prove anything to you as it IS common knowledge.

Anonymous 04/28/12(Sat)08:47 No.24512339

>>24512252
Time a wget on 100 images.
Time a copy of those images from one folder to another.

If you have a 100Mbit connection, and you manage to saturate it from 4chan (very unlikely), that is around 10MB/s, decent SSD have around 300MB/s read speeds, average RAM transfer speeds are in the 5 GB/s area (although in practice, this is around 500MB/s in typical applications).
That is a factor of 30-50x speedup.

>>	Anonymous 04/28/12(Sat)08:47 No.24512341 >>24512296 >>24512306 As a moron, I'm glad. Styling this will be easy-as-fuck. I'm sure the script devs aren't having much trouble with it either. It's considerate to include so many classes for modifications.

Anonymous 04/28/12(Sat)08:48 No.24512350

> If you have a very heavy page full of images and dynamic content, it matters jackshit where you're taking your data (be it on the network, locally or cached), you still need some time to render it in your browser.
4chan is plain static HTML. Rendering images shouldn't take much time either.

It's the pages full of bloated JavaShit that slow browsers down the most. And I'm glad 4chan isn't like that.

Anonymous 04/28/12(Sat)08:49 No.24512357

>>24512337
>when proof requires tools and time.

You don't have to repeat an experiment every single time you need to prove it's true, you can take data from reliable sources and show that.

Stop being stupid, the burden of proof falls on the one that makes the claim. Then show me some proof if you want to go on with this conversation.
"Common knowledge" means nothing, it's "common knowledge" for the uneducated that the Moon has no gravity and yet it is flat out wrong.

>>	Anonymous 04/28/12(Sat)08:49 No.24512361 >>24512319 If it wasn't faster, why would it be implemented in pretty much every browser?

>>	Anonymous 04/28/12(Sat)08:49 No.24512363 What's the point of even arguing that images takes longer to load remotely. That has nothing to do with the original post. Fuck you, Lawlz, you fucking moron.

Anonymous 04/28/12(Sat)08:49 No.24512364

>>24512319
>If you have a very heavy page full of images and dynamic content, it matters jackshit where you're taking your data (be it on the network, locally or cached), you still need some time to render it in your browser.
Render time is tiny compared to data loading times.

!seaweedKJc 04/28/12(Sat)08:49 No.24512367

>>24512296
I'm extremely excited about it. Some of the selectors I have are just awful because of lack of classes.

Hell, just a simple class added to the announcements means I can stop using "font[color=red]" and then have to put in workarounds for everything else that's red that's not an announcement, like error pages.

>>	Anonymous 04/28/12(Sat)08:49 No.24512368 >>24512350 You've never used 4chan X I take it?

>>	Anonymous 04/28/12(Sat)08:50 No.24512378 Downloads on 4chan will already be slow thanks to TCP slow start, so differences between files less than 1 MB in size are going to make almost no difference at all.

>>	Anonymous 04/28/12(Sat)08:50 No.24512379 >>24512192 This

>>	Anonymous 04/28/12(Sat)08:51 No.24512385 >>24512035 Yes, from the episode in Season 3 where they go on Oprah.

Anonymous 04/28/12(Sat)08:51 No.24512387

>>24512339

1) you access thumbnails, not full pictures
2) I'm not talking about downloading those thumbnails vs loading them from your pc, I'm talking about rendering the page while streaming the content off the network vs your hard drive.

I fully well know that cache is faster and that local content is faster, this doesn't mean anything though, since a better layout for the page means it's easier to be parsed and rendered by the browsers (and scrapers) which means that it will be actually easier for the 4chan servers AND the browser rendering itself which means that it will eventually lead to better performance.

>>	Anonymous 04/28/12(Sat)08:51 No.24512391 >>24512364 tell that to my net book

>>	Anonymous 04/28/12(Sat)08:52 No.24512395 OP here, the main point of this is not about faster or slower for the end-user (probably insignificant difference), it's about the network bandwidth and server-side CPU usage.

>>	Anonymous 04/28/12(Sat)08:53 No.24512409 >>24512395 >it's about the network bandwidth and server-side CPU usage. Fair enough, however with a better page layout, it'll be easier to parse by bots and scrapers which could actually ease a lot of load on the server itself

Anonymous 04/28/12(Sat)08:53 No.24512412

>>24512387
>1) you access thumbnails, not full pictures
Doesn't matter, time it anyway, you'll be surprised.

>I'm talking about rendering the page while streaming the content off the network vs your hard drive.
And what do you think your renderer is doing most of the time?
Protip: Waiting on data.

>>	!seaweedKJc 04/28/12(Sat)08:53 No.24512415 >>24512395 >OP here No one cares. Your thread is stupid and your opinions are shit.

>>	Anonymous 04/28/12(Sat)08:54 No.24512421 >>24512415 Hi weedy-poo.

>>	Anonymous 04/28/12(Sat)08:54 No.24512426 >>24512412 >And what do you think your renderer is doing most of the time? >Protip: Waiting on data. Prove it, again.

>>	Anonymous 04/28/12(Sat)08:54 No.24512427 >>24512395 If you are so concerned about that, maybe you should be asking moot to lower the image limit per thread to 10 images. That's what eats most of the bandwidth.

>>	LAwLz !LAwLzaWU1A 04/28/12(Sat)08:54 No.24512428 >>24512415 So seaweed what do you think. What is the real bottleneck when accessing a webpage. Your computer, or the network connection?

>>	!seaweedKJc 04/28/12(Sat)08:55 No.24512434 >>24512421 Hi. I'm pretty damn happy for the HTML change. Tables to divs is something that will be EXTREMELY welcome.

>>	Anonymous 04/28/12(Sat)08:55 No.24512440 >>24512412 this

>>	Anonymous 04/28/12(Sat)08:55 No.24512441 >>24512434 >Tables to divs is something that will be EXTREMELY welcome. That is the important thing here.

Anonymous 04/28/12(Sat)08:55 No.24512443

>however with a better page layout, it'll be easier to parse by bots and scrapers which could actually ease a lot of load on the server itself
Parsing is done in the CLIENT, not the server. Maybe they'll parse faster but they still have to pull data from the server to do it, and there is a slight increase in that amount now.

>>	Anonymous 04/28/12(Sat)08:55 No.24512445 I'm actually disappointed they're changing the HTML. 4chan still renders properly in Dillo in Damn Small Linux, but that's not going to be true after the switch to HTML5.

>>	!seaweedKJc 04/28/12(Sat)08:56 No.24512450 >>24512428 Network connection

Anonymous 04/28/12(Sat)08:56 No.24512451

>>24512209
HDD speeds are irrelevant. So is RAM speed, bus speeds, CPU speeds.

I fire of a HTTP request and get a response. Then my browser parses the response and requests the thumbnails and begins rendering. When the thumbnail responses are complete, the CPU IS STILL OCCUPIED RENDERING THE GOD DAMNED HTML. The reason for the pipelining is both that you MAY have a slow internet connection and or the images might be very big, in this case YOU DO wait for the network. But this isn't the case here, the case here is that the thumbnails are already retrieved while the browser is still struggeling to render the document properly.

AFTER THIS IS COMPLETE the thumbnails are rendered. If the network actually WERE the bottleneck, you would see it as the thumbnails being gradually loaded as they would be rendered bit for bit (aka porn using ISDN in the old days), but I don't because I don't fucking have a cheap ass connection.

>>	Anonymous 04/28/12(Sat)08:56 No.24512454 >>24512378 ???

>>	Anonymous 04/28/12(Sat)08:56 No.24512456 >using http >2012 gee, shit is already bloated no matter html5 or not

>>	Anonymous 04/28/12(Sat)08:56 No.24512461 >>24512426 Load up Firefox in a profiler and see for yourself. Or, go to a image heavy thread (100+ posts), time the first load, press F5 and time that load. I guarantee you the second time will be significantly faster.

>>	Anonymous 04/28/12(Sat)08:57 No.24512465 >over double the size to do the same shit WHY IS HTML5 SUCH A FUCKING PIECE OF TRASH

>>	Anonymous 04/28/12(Sat)08:57 No.24512472 >>24512451 >>24512451 >>24512451 >>24512451 >>24512451 >>24512451

>>	LAwLz !LAwLzaWU1A 04/28/12(Sat)08:57 No.24512479 >>24512450 Nice to see that there are some sane people here. Also, you will be updating Onee-chan to support the new HTML, right?

>>	Anonymous 04/28/12(Sat)08:57 No.24512480 >>24512451 Get a faster browser or CPU if parsing HTML is that slow on your system.

>>	Anonymous 04/28/12(Sat)08:58 No.24512485 <board> <thread> <post> blablabla </post> </thread> </board> moot why u no xml?

Anonymous 04/28/12(Sat)08:58 No.24512486

>all that bloat

Classes and Ids, classes and Ids everywhere.

That shit isn't needed at all.
You only need those things every 5 levels deep, or at main sections.

Most of the stuff in each major section is UNIQUE since it is a parent of something with an ID.
Remove all that other shit and style them by travelling down the DOM.

Also, seriously having mobile code in browser code?
The fuck?
Detect mobiles on server side.
Yeah, it is pretty neat watching the site change to small sizes when it gets really crushed, but why would any person use that on a desktop?
Resolutions are locked on mobiles last I checked. And CSS has an orientation detection component of it so you can style for vertical or horizontal screens.

Anonymous 04/28/12(Sat)08:58 No.24512487

>>24512461
>I guarantee you the second time will be significantly faster.

You're the one making those claims, go show me some proof. Once again, I don't give enough fucks about this HTML5 change to be bothered by it. You people are the one whining like retards not realizing that new HTML5 pages will just make it easier for everybody else, you show us some proof about it being a bad thing.

>>	Anonymous 04/28/12(Sat)08:58 No.24512490 File: 1335617935.jpg-(73 KB, 480x349, songayku.jpg) ITT: moot defense force in full swing

>>	Anonymous 04/28/12(Sat)08:59 No.24512498 >>24512486 Now, kids, this is a legit complaint and I totally agree with it.

>>	Anonymous 04/28/12(Sat)08:59 No.24512500 >>24512485 That wouldn't change much.

!seaweedKJc 04/28/12(Sat)08:59 No.24512503

>>24512479
'Course. I'll try to get it done in a branch before it goes live.

CPU bottlenecking is still an issue with really heavy CSS, however. I can't use 4chan Rewired with all its gradients and transparency on my computer because my Core 2 Quad performs poorly with it, but even on a new i7 stickies are going to perform badly due to network bottlenecking.

>>	Anonymous 04/28/12(Sat)08:59 No.24512504 >>24512490 I want to suck moot's cock

>>	Anonymous 04/28/12(Sat)08:59 No.24512505 >>24512480 It's not the PARSING that is slow, but the RENDERING. Learn to read, faggot.

LAwLz !LAwLzaWU1A 04/28/12(Sat)09:00 No.24512509

>>24512451
Please do the test I posted here >>24512212
Save this thread on your computer, clear your cache/disable all addons and load the page locally. Now, delete the page from your computer, clear the cahe and keep your addons diabled. Load the page from 4chan's servers.

Which will be the fastest? The local copy being loaded, or the one on 4chan's servers?

>>	Anonymous 04/28/12(Sat)09:00 No.24512511 >>24512487 shut the fuck up you autistic piece of shit nobody cares about your shitty need for proof which you can find yourself in 2 seconds if you're fucktarded enough to not already know

>>	!seaweedKJc 04/28/12(Sat)09:01 No.24512522 >>24512486 Classes are absolutely needed for easier styling. They might not be necessary to you but whether they're there or not makes no difference to you. As for IDs, its just for the test version, they won't actually be there for the main thing.

Anonymous 04/28/12(Sat)09:02 No.24512528

>>24512451
>HDD speeds are irrelevant. So is RAM speed, bus speeds, CPU speeds.
Then you must go clueless about the relative difference between them.

>When the thumbnail responses are complete, the CPU IS STILL OCCUPIED RENDERING THE GOD DAMNED HTML.
No.

>But this isn't the case here, the case here is that the thumbnails are already retrieved while the browser is still struggeling to render the document properly.
It's struggling because it's waiting on image data to display.
You might have had a case if 4chan was "WEB TWO DOT OH" with kilobytes of JS, that isn't the case here however. 4chan HTML rendering time is insignificant compared to network loading times.

>If the network actually WERE the bottleneck, you would see it as the thumbnails being gradually loaded as they would be rendered bit for bit
That is exactly what is happening.
It's generally not perceptible in the days of millisecond retrieval times - but that is completely irrelevant to the argument, all you have to do is scale up the images in size or numbers and eventually you WILL hit your connections bottleneck (or more likely, 4chan server's cap).

Anonymous 04/28/12(Sat)09:02 No.24512529

The case of network bandwidth is completely irrelevant after gzip compression. The new HTML is much more consistent and will gzip down to about the same amount of bytes. Every browser capable of rendering the mess on the left properly supports gzip compression.

Of course, the new HTML isn't that good because it sprinkles duplicated and useless IDs everywhere, assuming that you didn't edit it anyway.

Anonymous 04/28/12(Sat)09:02 No.24512536

>>24512509
>Which will be the fastest? The local copy being loaded, or the one on 4chan's servers?

This isn't what we're talking about though. We're talking about the fact that having a few more bytes sent over the network (not considering removal of whitespace and compression!!) could actually make it faster for the browser itself to render (and for the bots to read the data), which ends up being better performance.

Network bandwidth or "bottleneck", which seems to be your only concern and you seem to be so focused (pointlessly so) about it, has nothing to do with it.

Until we get a definite HTML5 code for the new 4chan interface, we can't profile jackshit.

>>	Anonymous 04/28/12(Sat)09:02 No.24512537 >>24512509 HDD speeds ARE IRRELEVANT you dumbwit. Of course, the INITIAL REQUEST WILL TAKE SOME TIME TO COMPLETE, but after this is done, the speed is the same. Your "test" is bullshit.

>>	Anonymous 04/28/12(Sat)09:02 No.24512538 File: 1335618172.jpg-(187 KB, 500x500, 1327281826812.jpg) >moot in charge of improving 4chan

Anonymous 04/28/12(Sat)09:03 No.24512545

>>24512511
>shut the fuck up you autistic piece of shit nobody cares about your shitty need for proof which you can find yourself in 2 seconds if you're fucktarded enough to not already kno

>dat ad hominem
well, looks like we know who's grasping at straws now...

>>	Anonymous 04/28/12(Sat)09:03 No.24512548 >>24512487 >You're the one making those claims, go show me some proof. How the fuck can I show you this? You would have to be in my fucking room at my computer you idiot. That's why I'm telling you to do this yourself.

>>	!seaweedKJc 04/28/12(Sat)09:03 No.24512551 >>24512545 An insult inside an argument is not ad hominem.

Anonymous 04/28/12(Sat)09:04 No.24512561

>>24512537
>INITIAL REQUEST WILL TAKE SOME TIME TO COMPLETE, but after this is done, the speed is the same.
And WHY do you think that is you fucking moron?
Because of CACHING, the data is being loading from HDD or RAM afterwards, which is significantly faster than network.
That has been my fucking point all along.

>>	Anonymous 04/28/12(Sat)09:05 No.24512565 >>24512551 https://en.wikipedia.org/wiki/Ad_hominem#Abusive

Anonymous 04/28/12(Sat)09:05 No.24512566

>>24512528
They are irrelevant, because you don't have internet saved on your god damned computer. Of course, the initial request will take some time to complete, but see >>24512536

>That is exactly what is happening.
No it isn't, and 4chan server's cap are irrelevant. I'm talking about the thumbnails, not the images. The images (because they are fucking huge compared to the HTML) would actually be bottlenecked by the network. No one argues about that.

>>	Anonymous 04/28/12(Sat)09:05 No.24512569 >>24512548 >You would have to be in my fucking room at my computer you idiot. Yeah, how can scientists demonstrate physical laws to the world, the whole world can't fit in a single laboratory

Anonymous 04/28/12(Sat)09:05 No.24512571

>>24512479
Maybe he's just fucking around and feels like making the extensions/scrapers/etc. guys do more work.

You have a functioning imageboard. People write stuff to work with it. Now you change everything to make it "easier" for those people.

But they have already expended the effort to do it for what it was before, and they are never going to regain that time they spent. Instead they now have additional work to do because of these changes.

It's a burden of work that no one should really want to do unless they're getting paid for it.

>>	Anonymous 04/28/12(Sat)09:06 No.24512574 >>24512561 You will always load the HTML, dimwit. The stylesheets are cached, and that is the assumption I'm using.

>>	Anonymous 04/28/12(Sat)09:06 No.24512578 >moot defense force derailing the thread with trolling after being told

>>	Anonymous 04/28/12(Sat)09:06 No.24512582 ITT: faggots arguing over a couple of bytes.

Anonymous 04/28/12(Sat)09:07 No.24512588

>>24512566
>No it isn't, and 4chan server's cap are irrelevant. I'm talking about the thumbnails, not the images. The images (because they are fucking huge compared to the HTML) would actually be bottlenecked by the network. No one argues about that.

HOOOOOOOOOOOLLLLLLLEEEEEEEEEEEEYYYYYYY FUUUUUUUUUUUUUUUUCKING SHIIIIIIIIIIIIIIIIT OMFG

you faggots are picking CERTAIN EXCEPTIONS?! HOLY FUCK

i'm out

>>	Anonymous 04/28/12(Sat)09:07 No.24512590 File: 1335618460.png-(80 KB, 649x490, enwhtm.png) Is this considered good practice in HTML5? I hate code and images floating around in space

>>	Anonymous 04/28/12(Sat)09:08 No.24512594 >>24512588 We were talking about a few extra bytes in the HTML, when the fuck did images become relevant retard?

>>	Anonymous 04/28/12(Sat)09:08 No.24512598 >>24512588 Thanks for (not) contributing to this thread whatsoever. Go in peace.

!seaweedKJc 04/28/12(Sat)09:08 No.24512601

>>24512561
>fucking moron
>>24512548
>you idiot
>>24512537
>dumbwit
>>24512536
>jackshit
>>24512511
>shut the fuck up you autistic piece of shit
>>24512505
>faggot
>>24512487
>retards
>>24512465
>FUCKING PIECE OF TRASH
>>24512363
>you fucking moron
>>24512296
>moron
>>24512279
>garbage
>>24512257
>Fuck your shit
>>24512255
>fuck you
>>24512231
>You fucktards
>>24512588
>you faggots
>>24512582
>faggots
>>24512578
>moot defense force
>>24512574
>dimwit
>>24512594
>retard

/g/ - Quality Posting

Anonymous 04/28/12(Sat)09:09 No.24512604

>>24512566
>No it isn't, and 4chan server's cap are irrelevant.
It's extremely relevant, I can't even cap my connection to 4chan.

>I'm talking about the thumbnails, not the images. The images (because they are fucking huge compared to the HTML) would actually be bottlenecked by the network. No one argues about that.

10 thumbnails are bigger than the entire HTML code.
They are the bottleneck when rendering a page.

>>	Anonymous 04/28/12(Sat)09:09 No.24512606 I welcome the new change.

>>	Anonymous 04/28/12(Sat)09:10 No.24512617 So how much does 4chanX slow down parsing a long thread? Is the official 4chan extension any faster?

>>	Anonymous 04/28/12(Sat)09:10 No.24512618 >>24512601 Welcome to 4chan, a place for personal insults.

>>	AMD☃Tea !Reeeeeeeeg 04/28/12(Sat)09:10 No.24512620 >>24512601 Gotta love Anonymous. Always so insightful with their shit flinging antics.

>>	Anonymous 04/28/12(Sat)09:10 No.24512621 >>24512601 >10 years old realizes how the world works.jpg

Anonymous 04/28/12(Sat)09:10 No.24512627

>>24512569
>Yeah, how can scientists demonstrate physical laws to the world, the whole world can't fit in a single laboratory
They don't need to demonstrate it to the whole world.
But to the ones they are demonstrating it to, you can bet your ass they are present.

>>	Anonymous 04/28/12(Sat)09:11 No.24512635 >>24512604 >They are the bottleneck when rendering a page. And so we admit that changing a few bytes in the HTML CODE won't matter at all because you're still going to have thumbnails, regardless.

>>	Anonymous 04/28/12(Sat)09:11 No.24512637 >>24512620 Gotta love namefags. Always so impartial in their judgements.

>>	Anonymous 04/28/12(Sat)09:12 No.24512649 >>24512574 >You will always load the HTML, dimwit No, you wont, html gets cached just like everything else.

!seaweedKJc 04/28/12(Sat)09:12 No.24512651

>>24512617
A good amount. Negligible on Nightly but on Chrome its really bad (on Windows).

The official 4chan extension is faster at quote previews and image expanding and the like in those large threads, but when you scroll down you'll notice huge performance hits as it tries to load the quick reply / reply hiding / report buttons, on a higher level that will continue slowing down the thread.

Anonymous 04/28/12(Sat)09:12 No.24512652

>>24512627
>They don't need to demonstrate it to the whole world.
>But to the ones they are demonstrating it to, you can bet your ass they are present.

No, you see, you're supposed to document what you do, explain what you did (so you let other people have a chance to repeat the same steps) and show some goddamn proof of your claims.

In this case, screenshots or terminal outputs in a pastebin are acceptable. Also external links to websites where other people already tried to profile this shit.

Anonymous 04/28/12(Sat)09:13 No.24512657

>>24512582
On average, about half a kilobyte more PER POST.

That /b/ thread with 3K posts will increase by 1.5MB.

For comparison, a thumbnail is about 2KB. There's 780 thumbnails in that thread, and about 1.6MB total.

So in this case, it could be said that with the new HTML for that thread the extra data transferred is equivalent to having twice the number of thumbnails. But that is for this case only.

>>	Anonymous 04/28/12(Sat)09:13 No.24512659 >>24512649 >html gets cached just like everything else. >yfw all threads have the same content and same posts because "it gets cached"

Anonymous 04/28/12(Sat)09:13 No.24512668

>>24512635
>And so we admit that changing a few bytes in the HTML CODE won't matter at all because you're still going to have thumbnails
I have never argued that, I simply jumped it and pointed out that network is in fact a huge bottleneck for webpage rendering.
Obviously the HTML code is almost completely insignificant.

Anonymous 04/28/12(Sat)09:14 No.24512669

m00t, change we can believe in
>m00t, change we can believe in
m00t, change we can believe in
>m00t, change we can believe in
m00t, change we can believe in
>m00t, change we can believe in
m00t, change we can believe in
>m00t, change we can believe in
m00t, change we can believe in
>m00t, change we can believe in
m00t, change we can believe in
>m00t, change we can believe in
m00t, change we can believe in
>m00t, change we can believe in
m00t, change we can believe in
>m00t, change we can believe in
m00t, change we can believe in
>m00t, change we can believe in
m00t, change we can believe in
>m00t, change we can believe in
m00t, change we can believe in
>m00t, change we can believe in
m00t, change we can believe in

>>	Anonymous 04/28/12(Sat)09:14 No.24512674 >>24512604 No, because they're fetched while your browser is struggeling to render moot's shitty HTML/CSS.

>>	Anonymous 04/28/12(Sat)09:14 No.24512681 >>24512657 You're completely ignoring gzip compression and the fact that the final HTML5 version will be stripped of useless shit. It's just there to show the developers how the new layout would work, to give them an easier job at parsing it.

>>	Anonymous 04/28/12(Sat)09:16 No.24512703 yeah guys it will be faster because of GZIP compression ! yeah !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! wahoo !!!!!!!!!!!

>>	Anonymous 04/28/12(Sat)09:16 No.24512706 >>24512649 >>24512649 In that case every thread, every thing would look exactly similar. The CSS is cached, HTML is NOT. Just look at the god damned HTTP response, it contains the HTML for fuck sake.

>>	Anonymous 04/28/12(Sat)09:16 No.24512708 >>24512681 gzip is not lossy. It reduces the size but eventually all that comes in one end has to go out the other. So something has to process those extra 1.5 million bytes in that case. There's no way around that.

Anonymous 04/28/12(Sat)09:16 No.24512709

>>24512649
It doesn't get cached. The server has the ability to communicate when the site will be outdated.

In case of dynamic websites such as boards, they set this to 0 or something, so it will "always be out of date" -> you never retrieve from cache unless working in "offline mode".

Anonymous 04/28/12(Sat)09:17 No.24512711

>>24512652
>No, you see, you're supposed to document what you do, explain what you did (so you let other people have a chance to repeat the same steps) and show some goddamn proof of your claims.
I have told you the steps, I have given you the numbers on network, hdd and ram speeds.
You said you didn't believe me, so I said test it yourself.

>In this case, screenshots or terminal outputs in a pastebin are acceptable. Also external links to websites where other people already tried to profile this shit.
Just do it yourself, the test is so simple.
Time the first page load, hit F5, time subsequent loads.
I guarantee the second load will be significantly faster.

>>	Anonymous 04/28/12(Sat)09:17 No.24512719 gee zip compression!!!!!!!!! yippee !!!!!!!!!!!!

>>	Anonymous 04/28/12(Sat)09:17 No.24512722 >>24512703 >>24512681 We should move on to 7z compression. It's multithreaded and compress more.

>>	Anonymous 04/28/12(Sat)09:17 No.24512723 File: 1335619077.png-(118 KB, 207x287, boogaloo_faggot.png) You'd think /g/ would appreciate the changes, if anyone. The old code was an absolute mess. Oh, well. None of you are in charge of anything here, so I guess it's all good.

>>	Anonymous 04/28/12(Sat)09:17 No.24512724 >>24512668 >I simply jumped it and pointed out that network is in fact a huge bottleneck for webpage rendering. So you're just a fag wanting to argue then.

>>	Anonymous 04/28/12(Sat)09:17 No.24512725 >>24512708 >So something has to process those extra 1.5 million bytes in that case. There's no way around that. OH NO MY CPU WILL HAVE TO PROCESS ONE WHOLE MEGABYTE OF DATA!

>>	Anonymous 04/28/12(Sat)09:18 No.24512737 >>24512708 The stuff doing the processing has no issues processing an additional 1.5 million bytes of data.

>>	Anonymous 04/28/12(Sat)09:18 No.24512740 Why the fuck will I have to buy GZIP compression when I already bought WinRAR? What the actual fuck?

Anonymous 04/28/12(Sat)09:18 No.24512741

>>24512522
No. There are way too many of them. They simply are not needed.

#this sort of .thing .exists in #css { you:know}
It can be combined in any way imaginable.

You only need a major node to have a class or ID, or something that is absolutely 100% unique and requires one.
Having them on everything is retarded.
Classes on image container? Why?
It can be styled by travelling through the DOM.
div.post div a img {outline:5px solid #ff0!important}
That hits only the post images.
There are a fuckton of other things that could be done in exactly the same way to style anything on there.

>>	Anonymous 04/28/12(Sat)09:19 No.24512744 >>24512711 Read the fucking thread instead of living your own conversation that you created in your own mind. We're not talking about cached vs non-cached. Read >>24512536 Can we drop this retarded burden-of-proof shit now? thanks.

>>	Anonymous 04/28/12(Sat)09:19 No.24512749 >>24512722 gzip is significantly faster and better suited to webpage compression.

>>	Anonymous 04/28/12(Sat)09:19 No.24512753 moot is mai waifu

>>	Anonymous 04/28/12(Sat)09:19 No.24512754 >>24512725 >>24512737 See: >>24511949

>>	Anonymous 04/28/12(Sat)09:20 No.24512757 >>24512737 True. overall, this is an improvement. it breaks "backwards compatability" and has unnecessary whitespaces, the first which will happen in any case of change, the second meaningless.

>>	!seaweedKJc 04/28/12(Sat)09:20 No.24512765 >>24512723 Mayhem's in charge of stuff, and he likes the changes

>>	Anonymous 04/28/12(Sat)09:20 No.24512770 >>24512709 I might need to fire up curl for that, but I'm sure 4chan sends 304 responses. At least, that code was shown on some auto-updater.

>>	Anonymous 04/28/12(Sat)09:21 No.24512772 >>24512749 If we're talking about speed then we would have gone with the likes of LZO/LZF/Snappy The point is that GZ is not multi-threaded.

>>	!seaweedKJc 04/28/12(Sat)09:21 No.24512774 >>24512741 >by traveling through the DOM Which userstyles can't do. Just calm your shit down. You're not someone styling the site.

Anonymous 04/28/12(Sat)09:22 No.24512777

>>24512754

See:

>the whole thread

We've already talked about this. The improvement of a few more bytes (that will be compressed either way and won't make a big difference with the network bandwidth) will make it easier for browsers to process and render AND will make it easier for bots and scrapers.

Result: a few more bytes can actually make it faster and perform better.

File size isn't everything, you have to view the whole picture.

Anonymous 04/28/12(Sat)09:22 No.24512781

>>24512744
>Read the fucking thread instead of living your own conversation that you created in your own mind. We're not talking about cached vs non-cached.
Why the fuck have you people consisted on denying that network is a bottleneck then?
You could have simply said "yes, that is true" and that would have been that.
I've never claimed or hinted that the new html code will be faster or slower.

>>	Anonymous 04/28/12(Sat)09:22 No.24512791 I'm not fucking buying gzip to view this shitty website. What a fucking stupid idea to limit your userbase to who bought that shit. FUCK

>>	Anonymous 04/28/12(Sat)09:23 No.24512794 >>24512770 would be interested in the result. it gets parsed by php right? would be cool if it can track changes/non-changes to the thread

Anonymous 04/28/12(Sat)09:23 No.24512795

>>24512723
The old code works in just about every browser and doesn't have more than it needs to. People have been able to work with it, otherwise there wouldn't be all these extensions and whatnot.

2chan's been around longer than 4chan and they haven't changed the basic page in over 10 years.

>>	Anonymous 04/28/12(Sat)09:23 No.24512796 >>24512772 LZO was invented in 1996, I believe gzip was used prior to that. It'd be a nice thing for the future though.

>>	Anonymous 04/28/12(Sat)09:23 No.24512797 >>24512781 >Why the fuck have you people consisted on denying that network is a bottleneck then? Because it's not the bottleneck in this case.

>>	Anonymous 04/28/12(Sat)09:24 No.24512807 >>24512709 Seems very inefficient if 4chan doesn't serve cached threads. Would be pretty retarded not to.

>>	!seaweedKJc 04/28/12(Sat)09:25 No.24512812 >>24512795 With that kind of mentality we wouldn't even have the internet.

Anonymous 04/28/12(Sat)09:25 No.24512813

>>24512795
>The old code works in just about every browser and doesn't have more than it needs to.

The new code performs better and is more compatible with newer standards. Also it's not a clusterfuck to parse and work with external tools.

Don't live in the past, you don't need to have backwards compatibility for everything, this is just a webpage you can access with every web browser out there, if you're using a browser that is not compatible with it, then you may want to get a better one.

Anonymous 04/28/12(Sat)09:26 No.24512831

>>24512797
>Because it's not the bottleneck in this case.
Yes, it still is.
Whether it's perceptible or not is irrelevant.

Fact: Computers are capable of much much more data throughput than any consumer (or experimental) network is capable of today.
Conclusion: Loading data over network will ALWAYS be a bottleneck.

>>	Anonymous 04/28/12(Sat)09:28 No.24512850 >>24512831 Fact: <something irrelevant> Conclusion: <something completely unrelated to this topic> So yeah, can we go back talking about real stuff or do your autistic senses tingle too much for you to pay attention and not get sidetracked?

Anonymous 04/28/12(Sat)09:29 No.24512866

>overall, this is an improvement. it breaks "backwards compatability" and has unnecessary whitespaces, the first which will happen in any case of change, the second meaningless.
Changing the backend won't break backwards compatibility.

It's hard to comprehend the reasons for this change (beyond the usual "it's better because it's valid HTML5!") because 4chan has remained essentially unchanged for this period of time and built up a considerable amount of "legacy", and the work involved in this rewrite was probably not trivial. It looks more like a marketing/economic than purely technical reason. A lot of things are like this. Perhaps it marks some significant event taking place with the people in charge of this site.

>>	!seaweedKJc 04/28/12(Sat)09:29 No.24512867 >>24512850 Don't ever go into debating or politics, I don't think you'd do well...

>>	Anonymous 04/28/12(Sat)09:30 No.24512877 >>24512850 They are both relevant, read the thread. People did claim that internet speeds were irrelevant to loading 4chan.

>>	Anonymous 04/28/12(Sat)09:30 No.24512878 >>24512867 >tripfag in charge of anything

>>	Anonymous 04/28/12(Sat)09:31 No.24512892 File: 1335619885.png-(380 KB, 592x450, imlistening.png) >>24512866 >It looks more like a marketing/economic than purely technical reason

>>	Anonymous 04/28/12(Sat)09:32 No.24512908 >>24512794 Nah, I think I'm wrong, so nevermind.

>>	Anonymous 04/28/12(Sat)09:34 No.24512929 File: 1335620052.png-(269 KB, 1193x984, thehellyouonabout.png) >>24512774 Yes they can. They have been able to do this since CSS existed.

Anonymous 04/28/12(Sat)09:34 No.24512932

>>24512812
The internet would be much more pleasant if that was the case. 4chan is one of the few large sites remaining that has (for now) not changed much, and that stability is a good thing. I'm not saying that newer standards, with their advertised advantages, are always a bad thing, and new sites adopting them are good, but when applied to existing sites these look like they're more for generating work than anything else.

Anonymous 04/28/12(Sat)09:34 No.24512935

>>24512813
>Also it's not a clusterfuck to parse and work with external tools.
In all honesty, only hardcore ricers are bothered by this.
Regex based crawlers don't give a fuck about markup, and quote previews/backlinks don't need any advanced selector-fu.

I'm not saying the changes are bad or anything. The best parts though, are the addition of the doctype and the "compatibility" with mobile devices.

Anonymous 04/28/12(Sat)09:35 No.24512945

This is what Moot should do:
>Combine external CSS
>Leverage browser caching
>Leverage proxy caching
>Minimize cookie size
>Serve static content from a cookieless domain
>Specify image dimensions
>Optimize the order of styles and scripts
>Put CSS in the document head
>Remove unused CSS rules

Anonymous 04/28/12(Sat)09:38 No.24512994

>>24512935
Regex based crawlers are bad fucking code practice.

In 2012 we have a much more evolved and standardized internet. We shouldn't be using technologies from the past if unnecessary, with more modern crawlers it will be easier to work with these pages and better tools will be able to get developed in the future.

Technology is here to make things easier for us humans. This is the fucking technology board, one would think people in here would approve of technological advancement...

>>	Anonymous 04/28/12(Sat)09:40 No.24513019 >>24512935 Doctypes and other semantic fluff were never really needed.

KING_JAFFE_JOFFER !7yCwooLsmg 04/28/12(Sat)09:40 No.24513022

the new 4chan extension renders a ~4000 post thread around 3x faster than the current one

~1000ms for the thread vs ~3800ms in the current huge /b/ thread

also it has filters and text butanz, spoiler revealing and backlinking everywhere (including on board indexes) and some other things

>>	Anonymous 04/28/12(Sat)09:40 No.24513028 http://validator.w3.org/check?uri=http%3A%2F%2Fboards.4chan.org%2Fhtmlnew%2F&charset=%28detect+a utomatically%29&doctype=Inline&group=0 Oh, yes. This is soo much better.

>>	Anonymous 04/28/12(Sat)09:41 No.24513035 >>24513022 Really? Where's the changelog?

>>	Anonymous 04/28/12(Sat)09:41 No.24513037 >>24512994 >Regex based crawlers are bad fucking code practice. It's easier to work with and more fault-tolerant. Regexps are the way to go when you don't have control or are unsure of the validity of the markup.

>>	Anonymous 04/28/12(Sat)09:42 No.24513044 >>24512994 RegEx is best ex. Nothing beats it. It is the most powerful filter there is.

>>	Anonymous 04/28/12(Sat)09:42 No.24513045 >>24513028 better than http://validator.w3.org/check?uri=http%3A%2F%2Fboards.4chan.org%2Fg%2F&charset=%28detect+automat ically%29&doctype=Inline&group=0&user-agent=W3C_Validator%2F1.3

>>	KING_JAFFE_JOFFER !7yCwooLsmg 04/28/12(Sat)09:42 No.24513046 >>24513035 up my butt son, i'm still programming

>>	Anonymous 04/28/12(Sat)09:43 No.24513060 >>24513037 Exactly.

>>	Anonymous 04/28/12(Sat)09:43 No.24513062 >>24512994 Spoken like a true web2.0 evangelist. You people never think about all the work involved in making these changes. Or maybe you do, and view that as an economic benefit.

Anonymous 04/28/12(Sat)09:44 No.24513071

>>24513037
>easier to work with
no, regex aren't easier to work with than simple XML/HTML DOM parsing.

>more fault-tolerant
This may be, but with a standardized and fixed layout there will be no space for faults. If something breaks then it means there's a problem, one should not ignore it. (also it's not like you can't keep using regexes with the new layout, you'd just be stupid for doing so).

>Regexps are the way to go when you don't have control or are unsure of the validity of the markup.

The whole point of re-structuring the layout is because there is going to be more control and validation in the new layout, making this complaint moot.

>>	Anonymous 04/28/12(Sat)09:44 No.24513078 File: 1335620677.png-(277 KB, 586x334, 1325514389684.png) >>24513022 >the official extension I thought we talked about this.

Anonymous 04/28/12(Sat)09:45 No.24513086

>>24513062
I'm not a web developer, I'm actually a kernel developer. I fucking hate web technologies and everything related to the internet, however I can recognize advancements when I see them.

There's nothing wrong with web2.0, if you still think it's bad, then you need to wake up and realize this isn't 2006 anymore.

>>	Anonymous 04/28/12(Sat)09:45 No.24513092 >>24513071 Regex is still faster.

>>	!seaweedKJc 04/28/12(Sat)09:46 No.24513103 >>24512929 See, now you're doing something that could be applied to EVERY element. md5 allows me to differentiate between thumbnails and expanded images in a faster way.

>>	Anonymous 04/28/12(Sat)09:46 No.24513105 >>24513092 [citation needed]

>>	Anonymous 04/28/12(Sat)09:47 No.24513115 >>24513103 >now you're doing something that could be applied to EVERY element. But it's not, because navigating through the DOM make it so you're pointing at that very specific element without needing an ID. It's the whole point he was trying to make.

Anonymous 04/28/12(Sat)09:47 No.24513118

>>24513028
>>24513045
And yet, the original code works in all browsers since the early 90s while the new stuff breaks in them. Newer is better? You think about it.

Let's face it, the W3C only makes "recommendations", the real standards are what the current browsers implement. What really matters is whether it looks OK to the user.

>>	Anonymous 04/28/12(Sat)09:47 No.24513120 >>24513092 You would be surprised at how slow a regex is in the case where half of the damn page is a partial match.

Anonymous 04/28/12(Sat)09:49 No.24513146

>>24513118
>And yet, the original code works in all browsers since the early 90s while the new stuff breaks in them. Newer is better? You think about it.

Read
>>24512994

This is not 2006 anymore, all browser support this kind of mild HTML5. It's not doing anything weird, if your browser doesn't support it, then it's time to upgrade.

(And no, upgrading is not a bad thing and you should've done so years ago already)

!seaweedKJc 04/28/12(Sat)09:49 No.24513153

>>24513115
Sure but that doesn't always work and sometimes scripts add in their own images you don't want styled or change it.

Adding classes to your own personal website absolutely isn't necessary, but in a huge website like this where so many people have their own stylesheets it just makes everything easier for everyone. People have been bugging mrvacbob to add a class to this or that for years and now its no longer a problem.

Anonymous 04/28/12(Sat)09:52 No.24513180

>>24513153
Well, fair enough.

I'd reply that people shouldn't add their shit in their own personal scripts and expect all unrelated extensions to work out of the box. If they want to tweak their shit they have to be responsible for it.

However, as a developer, I can agree that work with simple IDs can be easier and better to manage.

>>	Anonymous 04/28/12(Sat)09:52 No.24513184 >>24513046 It's not my fault when you 'advertised' your statement in present tense. It gives off the meaning that you're talking about a new release, you know.

>>	!seaweedKJc 04/28/12(Sat)09:52 No.24513191 Headed home, later /g/~ >>24513180 Of course, as we've been doing for years. moot's just choosing to listen now and he's a pretty cool guy for it.

>>	Anonymous 04/28/12(Sat)09:52 No.24513196 Well, Don't Optimize Unless You Have Performance Problems. When moot is going to see his provider bill or 100 computer ddos take entire website off, he'll make some optimizations, i guess.

>>	Anonymous 04/28/12(Sat)09:54 No.24513210 >>24513196 There's no need of optimisation. CERN has agreed to host 4chan.

Anonymous 04/28/12(Sat)09:54 No.24513211

>>24513071
>regex aren't easier to work with than simple XML/HTML DOM parsing.
They are when you don't need any advanced selection/xpathing, etc.
7 lines of code (the size of the regex to parse threads on /htmlnew/), contained in one place, is much easier to maintain than any libxml2 based parser.

>be more control and validation in the new layout
No. When you have a twitter xml feed, you can safely assume that it's valid and won't rape your parsing lib.
On 4chan, anything can happen. Literally, anything.

Anonymous 04/28/12(Sat)09:55 No.24513223

>>24513153
They can still be styled.

CSS has nth selectors. (pretty god damn powerful at that, you can put rules in that badboy like 3n+3)
So even if someone added another 100 images in that same depth, you could select the first one easily.
div.post div a img:nth-child(1) {outline:5px solid #ff0!important}

It isn't a case of us styling the site would be a pain to them.
This would be the source and they'd be styling around it.

Anonymous 04/28/12(Sat)09:55 No.24513229

>>24513211
>On 4chan, anything can happen. Literally, anything.

Moot is explicitly working with extensions developers just for this fucking reason.... If anything, I'd give him more credit now that he acknowledged such shit and might be supporting it in the future.

>>	Tripfaglookatmeohnoaddinganextrakilobyteonurbrowser !RNQUHOpCIo 04/28/12(Sat)09:56 No.24513235 File: 1335621388.png-(124 KB, 1935x1429, LE REDDIT.JPG.GIF.AVI.LOLYNOKV(...).png) Fixed that for you.

>>	Anonymous 04/28/12(Sat)09:57 No.24513247 >>24513211 On Twitter, you parse xml. In Soviet Russia, xml parses you. On 4chan, Steve Jobs is a Japanese Nigger hybrid cyborb who rapes tree-hugging script-kiddies, giving birth to battlestations which run cron jobs that post on Deskterp Thrads.

>>	Anonymous 04/28/12(Sat)09:58 No.24513256 >>24513028 >tripcode + email still generating invalid html Really, moot? Really?

Anonymous 04/28/12(Sat)09:59 No.24513259

>>24513086
The "advancements" are just so developers like you can have work to do?

>>24513146
Upgrading to view a site that has always worked? If they're adding new features of HTML5 to do some new stuff then that's a good thing, but this is just showig the same old posts, and they're changing things around just because they want to be "newer".

Sometimes I wish it was still 1966. When things were made to last and not deliberately half-made just to give more people work to do.

>>	Anonymous 04/28/12(Sat)09:59 No.24513266 >>24513247 Why don't you go back at "developing" your kernels if you don't understand how similar xml and html are?

>>	Anonymous 04/28/12(Sat)10:00 No.24513270 >>24513259 There is no need to upgrade. Any browser that supports CSS will support this new layout. Any browser that doesn't never supported the old one properly either.

Anonymous 04/28/12(Sat)10:02 No.24513293

>>24513259
>so developers like you can have work to do?

what

>but this is just showig the same old posts, and they're changing things around just because they want to be "newer"

You obviously have no idea how computers work. As a developer, 90% of the time your "improvements" don't even reflect to the end product graphically, however the engine behind it is much more stable, better coded and easier to work with.

With a simpler code you will have an easier time setting up modifications and more improvements, it's easier to interact with for non-official extensions (4chanX for example) and easier to parse by bots and scrapers.

All this shit is innovation compared to a website from 2005, if you don't like it, you should probably deal with it like the rest of the world does.

>>24513266
>implying that one is me

[Return] [Top]

Delete Post [File Only]
Password

Style [Yotsuba | Yotsuba B | Futaba | Burichan]