An Even Faster Browser?

octavian755 asks: "Seems that a 16-year-old Irish student has created an Internet browser called XWEB, which is the fastest browser known to date. This browser is said to be capable of boosting surfing speeds on a dial-up connection by 100 to 500 percent. What I would like to know is something like this even possible?" Update: 01/20 07:30 GMT by C : As folks have pointed out, this story is a duplicate. Also, a minor title gaffe corrected. Sorry about that.

Comments

Already Covered on /.

[grantdh]

What, weren't the responses given the last time this was posted enough???

Damn, even I remember this one and I'm notorious for my short term memory loss. Who was smoking what when this one got posted? :)

Re:Already Covered on /.

[jsse]

Calm down, I see the differences:

1) This is posted by Cliff, last time it was posted by Hemos. Isn't it a pity when those who blocked Cliff's or Hemos missed this great news? The news will be reposted by Taco very soon, so he majority who blocked both Cliff and Hemos will not miss it
2) Last time we don't have typo in the headline. This is very untolerable for slashdot readers
3) This time we've same article from other source. Slashdot editors must make sure they don't miss any reference to great news like that
4) Last time Hemos said 'Quadruples Surfing Speed', today Cliff found out that it can actually boost the speed up to 500% percent! This is definitely an improvement and should be posted as a follow-up

It's so fast...

[Hubert_Shrump]

It warps time, and reports itself again!

Hey Cliff

[presearch]

Hey Cliff.
You oughta, like, read Slashdot more often.

How is it possible to be so fast?

[Pyromage]

It's not.

No, I haven't used it. But there are simple facts, like the speed of light. The thing is this: data can only be transferred so fast.

If I have a 56k connection, than the fastest I can transfer is 56k (I know there are other considerations, but that's not important). It's that simple.

There are some things that can be done to speed it up: cache things. Render things faster, but they are all stopgap measures. I don't care what it says, the fact is that you only get so much speed. Any more just is not there.

It's like saying my car gets infinite miles per gallon. I can improve things, but I still need some fuel no matter what I do.

Re:How is it possible to be so fast?

[babbage]

Come on, be nice :)

Let's profile this rather than just flame. The article claims a 100% or greater speedup, which is of course twice as fast. If the download time is half of that, and as you say it cannot be shurnk further, then you can still realize a 100% gain by getting the other half of the work to approach zero time.

Very fast rendering (but very broken? I don't see anything saying this browser is actually usable or at all standards compliant...) is pretty much the main way to bring down that chunk of time. Good caching can minimize the amount of data transferred, and as another commenter noted, if the browser can take advantage of mod_gzip they'll get a significant download reduction on many sites.

Stopgaps? Sure, but no one is saying that things will be infinitely fast. Have you actually spent any time profiling what portion of the time is spent on which tasks in getting & displaying a web page? If the average downloading time is 50% or more then okay, your flame scorched the right target. But if other factors can accumulatively account for 50 or 80 percent of the work time, then your objections become just one of several relevant bottlenecks.

The thing is, whether or not you have ever used software profiling tools, I'm sure that the developers of the major browsers all have. That's what makes me skeptical of this. If there are any major gains to be made in download compression, caching, rendering or other areas, I would think that optimizations from each area would have shown up in the mainstream browsers by now (and in fact, all of this does exist in some form). While there is still room for improvement -- as the release of Safari clearly shows to the MSIE team -- extravagant claims are unlikely to be true.

I skipped out this article the last time it was posted, so may be asking what is a FAQ by now, but can anyone provide a better source of material on this browser than Rupert Murdoch's little puff piece? It would be interesting to hear how this browser supposedly works, against which browsers it supposedly does so much better, and whether those gains hold true as you adjust variables like bandwidth (does the gain wash out at DSL or T1 speeds?) and processing power (does the gain get even better on a very fast computer?). It would be interesting to see which browsers it was benchmarked against, and if there was any obvious problems with them when the tests, if any, were conducted ("what, I shouldn't be running against mod_inflate_data on that side?" :).

/. & vacations

[Anonymous Coward]

Nice to know that I'd be able to go on vacation, come back to /. and see everything I missed reposted.

And I bet he did it...

[Cinematique]

In a small codebase, too. Take that, Mozilla! First KHTML, now a single 16-year old boy! And people think Microsoft produces bloated software. /me runs and hides ;)

I'm really impressed here..

[zcat_NZ]

No technical details, not even an 'open' demo so we can see it's not rigged.

The usual excuse; this is such advanced, groundbreaking stuff and he doesn't want anyone to steal his ideas until after he's been given some development capital.

Scam. Scam. Scam...

Re:deja vu

[WIAKywbfatw]

Damn this is a recent repeat.
Not even any more info either. I'm sure all of us would like to examine this browser just to see if it is true.

More info: http://radio.weblogs.com/0103966/2003/01/14.html
Even has every media player: http://www.taipeitimes.com/News/worldbiz/archives/ 2003/01/13/190872
http://www.esatbtyoungscientist .com/

Uh-oh. It's deja vu all over again. Someone call Yogi Berra...

It's easy making it faster

[skinfitz]

All you do is supply a cached copy of the entire web with the browser, then when on a dialup, all it does it look at the site URL and serve the page up from the cache.

Easy!

The easy explaination...

[OneFix]

Turn off HTTP 1.1 in IE's settings and it won't be able to use compression (Mod_GZIP, etc)...

The other explaination is to modify Mod_GZIP or similar on the server side to report as some odd name (Mod_XWEBS might be a nice one) :) So that only your client recognises the compression method...this is real easy with Mozilla...and that or something similar is likely to be the real client. Mod_GZIP alone can give up to 12x the speed on a purely text page.

The other features (Built-In TTS, Access to multiple search engines, etc) are all fairly standard in the browser market now.

Now, I don't think it's been said enough...this kid supposedly did 1.5 million lines of code in 2 years...

Which would be ~2054 lines of code per day... or 85 lines of code an hour...

If we assume that he needs a minimum of 6 hours of sleep per night, that brings it to ~115 lines of code per hour...

This doesn't allow for eating, testing, rewrites, attending classes, reading documentation, etc...

Now, even the best coders only do ~100 lines of code per day...

I refuse to belive that this kid could do the work of 20 coders over a 2 year period...

You'ld think that this kid would release a binary only distribution for testing so that everyone would stop doubting his sincerity.

Re:The easy explaination...

[stevey]

Now, even the best coders only do ~100 lines of code per day...

That's not true; some coders may write 500, some may write 50, and others may write only five.

I'm not even that sure it is worth measuring the lines of code written as a performance indicator either; I've had days at work where I've written only one line of code - but it was the line to solve a random threading deadlock; and so it was the correct line to write.

Re:The easy explaination...

[stevey]

Creating IDL definitions, and wrappers for libraries is large in terms of LOC even if it's short in complexity - just to give one example. (I once worked on an CORBA based distributed reporting project)

Other projects written in Java or C++ with lots of comments, and interfaces would be doable - it really depends upon the implementation language the level of the coding, and the individual developer.

karmaho'ing

[bolind]

Hey, just find all the highmodded posts from the last time this story was on /., and post them again for instant karma... hey, here's my take:

"It's just Mozilla; it seems to cache the entire web during the build process."