Slashdot Mirror
Most of the Web Really Sucks If You Have a Slow Connection (danluu.com)
Dan Luu, hardware/software engineer at Microsoft, writes in a blog post: While it's easy to blame page authors because there's a lot of low-hanging fruit on the page side, there's just as much low-hanging fruit on the browser side. Why does my browser open up 6 TCP connections to try to download six images at once when I'm on a slow satellite connection? That just guarantees that all six images will time out! I can sometimes get some images to load by refreshing the page a few times (and waiting ten minutes each time), but why shouldn't the browser handle retries for me? If you think about it for a few minutes, there are a lot of optimizations that browsers could do for people on slow connections, but because they don't, the best current solution for users appears to be: use w3m when you can, and then switch to a browser with ad-blocking when that doesn't work. But why should users have to use two entirely different programs, one of which has a text-based interface only computer nerds will find palatable?
No idea how your connection speed adds anything to this.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
Why does my browser open up 6 TCP connections to try to download six images at once when I'm on a slow satellite connection?
i used to have this problem until i switched to Lynx. Now pesky things like popups and adware are a thing of the past! six streams indeed.
Good people go to bed earlier.
This is one of the most obvious statements I have ever read on /., or anywhere on the Internet. As long as I have a fast enough connection, of course.
Everything Internet-related really sucks if you have a slow connection.
#DeleteFacebook
Why does my browser open up 6 TCP connections to try to download six images at once when I'm on a slow satellite connection? That just guarantees that all six images will time out!
The problem is not opening 6 connections, or failure to retry, but a timeout that's too short.
Tell that to all the people with slow connections, who's operating system is constantly phoning home to a dozen different servers and hijacking their bandwidth to spread itself on a p2p network.
I have the same problem with smartphone apps. If you don't have the highest LTE connection possible, the app is a pain to use. Go to a rural area, and you may not even get it to open. Web sites are the same way. They give developers super fast connections, and they develop applications that require that speed. They don't put them on slow networks and test to see if they are even useful on a basic level.
Most web browser engines are open source. Go and modify one or many of them to handle slow connections better.
I'm not sure why your provider isn't offering a HTTP proxy server (as is standard for Satellite providers) that has all the sufficient timeout values you need as standard to workaround these latency issues?
Change is certain; progress is not obligatory.
You can configure this setting in Firefox. It doesn't look like Chrome has a similar configuration.
http://kb.mozillazine.org/Abou...
network.http.max-persistent-connections-per-server - default = 6
Try setting this to 1.
Source:
https://support.mozilla.org/t5...
A lot of what drives modern internet design is e-commerce. If you're on a slow connection, you probably don't have much money to spend, so why should anyone care? Or so the thinking goes....
Competition Good, Monopoly Bad.
Even with fast Internet connections, websites are so bloated with ancillary scripts and tracking code and cross linking to 20 different various advertising and content servers that you get stuck waiting no matter what. CDNs helped but you're still hostage to X advertising companies one slow server because it's not on that CDN.
With so much crap advertisement on each page, all trying to get a few cents from adchoices, and google crap, no wonder your connection is slow dude.
all show which side their bread is buttered on.
When the advertising content loads first and the page rebuilds/rearranges itself 6 or 8 times and finally the content you want to see becomes visible or stabilizes enough to click a link.
I think some of the pages are designed to do this on purpose.
You get a glimpse of the content you are looking for and click on a link just as the page rebuilds itself and the link has changed to an ad and the cash register rings on a click through.
Rick B.
Are designed for faster connections than yours where doing things asynchronously is much preferred (Though not it appears by the developers I tend to deal with) to synchronous.
Build a Man a Fire, and He'll Be Warm for a Day. Set a Man on Fire, and He'll Be Warm for the Rest of His Life.
It had a number of options where you could set up the number of global connections and per page. And yes, it was useful during the early days of dialup.
But anyway, most of the web sucks in this regard. Period. Sites are horribly designed these days - a gazillion JS dependencies, unnecessarily large images which get scaled, zero concern for mobile devices, etc.
This is part of a much larger pattern in software development: Targeting a subset of the user base and assuming that everyone is on a fast connection, every Android user is on a phone with a small screen, every Linux user and potential user is comfortable with the command line, etc. We live in a time of tipping points; if you and your system and your use of it fall into the favored group then your life is somehwat-to-vastly better than if you're walled off in a shunned minority.
The digital divide in the US became most evident (to me) in this last election cycle.
If you look at the page weights of 'conservative' vs 'liberal' news sites the former are much smaller and tailored to people on even a dial up, in large part because they know their demographic. Rural internet in the US flat out sucks. We have counties in my state, not more than 3 hours outside of Chicago that still have dialup as a viable option.
Drudge Report loads amazingly fast. Huffington Post does not. Drudge was 1.13 MB in size with 44% of that images. (The site I used to analyze them was done with Drudge's 14 assets long before Huffington Post stalled at 220/222 assets.)
The art of optimization seems to have disappeared, it made a small resurgence when web developers tried to optimize for the mobile web, but it doesn't look like most developers ever tried that hard.
It's a closed feedback loop. Developers live in places with fast Internet, test in places with fast Internet and then don't understand what it's like anywhere else. Students on college campuses live with gigabit internet and Internet2 connections to peer universities. They move to cities that Comcast pays attention to.
The best suggestion I have: Turn off images, configure the browser not to thread connections, and get involved in local government to get faster internet to your area.
Be sure to run adblockers, stripping out adverts makes a big difference.
But even slashdot is a big fat bloated pig. no reason at all to load everything and a giant pile of JS.
Do not look at laser with remaining good eye.
Most of the matrix is clogged with heavy footprint advertising and is light on content.
Websites that have to load 50+ JavaScript ads before displaying the page can easily overwhelm a slow processor even with a fast cable connection.
Opera started it. Other browser makers had to follow suit to keep up with Opera's speed.
Get an extension for Chrome or Firefox that blocks images unless asked not to. Turn off Javascript unless the site needs it.
When one of the tactics for those on slow connections is to employ ad blocking technology, one doesn't have to look far as to the impact on bandwidth. Couple that with the fucking telemetry bullshit infecting software these days, and the problem is obvious in both directions.
I find it odd that our society will label physically tracking everything a human does a crime, but virtually tracking and selling everything a human does is called capitalism.
I think Opera came up with an ideal compromise. A middle-man proxy that did filtering and compression. Fast pipe on their end, slow on yours.
Shai Schticks:"You don't make peace with friends, you make peace with enemies"
Ssh, and telnet work just fine over a slow connection and so does email so long as it doesn't have a load of attachments plus other protocols such as gopher. People did manage to use the internet over dial up before HTTP/HTML came along and sucked up as much bandwidth as it could!
There is a hilarious (and sad) commentary on website bloat at http://idlewords.com/talks/web... that shows truely outrageous examples of this sin.
Users like myself who are on a metered connection also suffer, but in a slightly different way. All those pesky videos that (partially) download on page load when you're just trying to read a news article. The bandwidth is plenty fast, which actually exacerbates the problem. How is it that Chrome provides a switch to turn off images, but not videos?
Another commenter here talked about pipelining. Again, this comes back to a lack of proper browsing features in our de-facto web browser, Chrome. We want power options, Google!
-IOVAR Web Dev Platform
1. Sites that play videos when the load.
2. Sites that display the entire page for three seconds and then cover it with a full screen ad.
3. Sites that constantly reorganize the page as it loads new ads.
4. Sites that load ads FIRST instead of the actual content.
Bottom line is the web sucks because Madison Avenue got a hold of it. They aren't content with placing an ad like they do in papers or magazines. They all in your face and FORCE your participation in message delivery. And before you even mention Ad Block, more and more sites simply refuse to load when you have that installed/enabled.
My solution (though my connection is fast enough) is to turn off images and javascript (and cookies) all the time, specifying some sites as exceptions, and leaving the browser config tab open for the rare one-off exception when I want to see images for one page refresh. I also separate browsing activities by my multiple user accounts, depending on the type (eg, banking in a separate account from general browsing), and either use separate consoles (Ctrl-Alt-Fn) for the accounts or put them on the same X desktop with "ssh -X ..." for most things, "ssh -Y ..." for very few, and some convenience scripts to automate starting some apps that way. This has worked well enough for me on multiple free OSes.
The benefits include speed, simplicity, I don't have to see garbage ads or images of things I just don't want in my mind, and hopefully a lower attack surface thus more secure browsing. The stuff I want is almost always in the form of text anyway, and I don't mind text ads.
Suggestions/critiques welcome.
A Free, fast personal organizer for touch typists: onemodel
I think that has to do with a really big company up in Redmond, which made a lot of money with a rip-off of another graphics-only operating system and produced code of dubious modularity and unnecessary bloat and complexity. I think their name starts with "M" or something. Maybe you should go and work for them and fix their software.
Meanwhile, somewhat more innovative companies than that Redmond company have come up with solutions like putting much of the browser and rendering into the cloud or using HTML-rewriting proxies. Well, it was innovative 20 years ago at least. Maybe people at that "M" company haven't heard of it, but the rest of us use this when we need it. (Of course, the rest of us usually can't afford satellite phones anyway; you need some old Microsoft stock options for that.)
I hate pages that look like they stopped loading and decide to do one last refresh AFTER I've started scrolling. That and images that don't load until you get near them. All that does is rearrange text while you're trying to read.
Only the State obtains its revenue by coercion. - Murray Rothbard
Most browsers open 6 connections because most connections these day can handle it. If you live in a time warp (like most of the US) with tech that's 20 years old, then you have to use 20 year old tricks of the trade.
"Back in the day" we had 56k at home or for the rich folk, 128k (ISDN), double ISDN if you were really lucky. We had the same problems, the 'web' was getting fancy with things like Flash, video and high-def images because all the 'work' was being done at places that had either access through an institution with at least Fractional T1's and things like 100Mbps home-internet copper/fiber connections for $10 were being promised as less than a decade away by the ISP's which were then repeatedly subsidized by the governments to do just that.
How did we do it:
a) Set up your own DNS caching servers
b) Set up your own HTTP/HTTPS proxy caching servers with giant caches
c) Proper QoS to make sure certain traffic had priority over others, small packets and small buffers
d) Set up your own servers (such as IMAP/SMTP, gaming etc) and have them sync during times of lesser activity. These days, with a bit of API tinkering, even social media can be done that way.
We had LAN parties with 100's of computers behind a 1Mbps cable modem, no problem.
Custom electronics and digital signage for your business: www.evcircuits.com
Be sure to run adblockers, stripping out adverts makes a big difference.
But even slashdot is a big fat bloated pig. no reason at all to load everything and a giant pile of JS.
I highly recommend https://alterslash.org it removes all the bloat from slashdot.org Thanks to Jonathan Hedley!
That was true up to 2012 or so, now it should block them from loading.
It's a shame Opera Software abandoned their old Presto engine for Chromium instead - Opera (through version 12) had a setting for number of connections per site and total number of connections. And ad blocking, and on and on.
As much as I'm not a fan of APK and his hosts file spam, this is somewhere he's right.
The only way to really use the internet these days is with either a hosts file blocking thousands upon thousands of entries, or better yet, a DNS server that does the same for all your devices at once (I use the latter).
Without that, many ads still waste your time and bandwidth, even if you don't have to see them at the end.
Of course the downside is that it makes it even easier for sites to detect when you're blocking ads, but there are ways around that too on a case by case basis.
Of course the normal way of writing web pages these days includes at least 10 different servers (mostly ads and tracking), so it's easy to only open 2 connections to any one server while still having your browser opening 20 or more connections at a time.
The problem with a satellite connection is not precisely related to latency but rather to jitter, the large differences in latency from one packet to the next. This happens as a result of rain fade, or a poorly engineered link to your transponder on the bird, or a variety of other more infrequent issues.
You can (and should) up your TCP timeout values from the default 3 seconds on a satellite connection, and adjust the http keep-alive timeout, etc, but a lot of times this just means you wait longer to be told when the connection fails.
The solution is a combination of caching, compression, and a performance enhancing proxy or PEP. The PEP does TCP spoofing, basically faking the acknowledgements to speed up the transmission of packets. Compression is similar to the MNP5/v.42bis stuff from modem days applied to a satellite connection. Caching is basically Squid. A lot of PEPs combine all three functions into one - Riverbed is a really good example, though i've worked with pretty much every vendor and they all do the same stuff, with differences in ease of use and efficency.
Implement the timeout fixes, implement a good PEP with all three of the ingredients noted, and make sure the connection is dialed in well with a good shot (line of sight) without physical impediments like trees, buildings, and most importantly microwave interference, and you should have a fairly reliable internet connection. You will still take hits, but you can look at the front of the satellite modem and see that is happening if it's an iDirect or something similar.
Bottom line though is that unless you are taking hits, you should be able to set up downloads of a lot of images and never see a timeout.
- someone who has spent a lot of time doing this (and living off sat connections) in awful places in the world
HBI's Law: Frequency of calling others Nazis is directly correlated with the likelihood of the accuser being Communist.
1. People are egocentric. Everyone thinks (by default) that everyone else is in the exact some relationship with their environment as they are. I've been in corporate IT for many years (in the past) and have seen a large number of major initiatives designed from the point of view of the HQ folks who are on top of the data center, network-wise. You know, millisecond latency to the servers. Then the app is deployed out to the remote masses, and... it sucks. So the app developers demand to know when the network staff is going to fix the problem. Bloated web pages happen because the people who design them do so when they are milliseconds away from the servers, and it all works fine for them.
2. Networks are layered. The idea of the layering concept is that each layer has its own set of responsibilities. The application layer through which the web browser communicates doesn't have any responsibility for managing data traffic flows. That is the responsibility of a lower layer. So the application just requests the data, and lower layers are supposed to take care of making it happen. As others have pointed out, parallelization of the work is useful, much of the time, so URL requests aren't serialized. When you are on a slow link, or a high latency link, this strategy fails. Should the network layer take care of it, or should the browser somehow become aware of the speed of the physical/data-link and adapt its behavior? Having the browser adapt might be seen as a violation of the layering principles; having the network/session layers interfere with which TCP connections are prioritized would definitely be a violation of layering. Come up with a good way to handle this type of problem, and you can help design the next version of the TCP/IP protocol suite!
3. It's mostly crap anyway. As others have pointed out, its mostly crap anyway. And in my experience its the crappier sources that load more pictures and other cruft onto the pages that you are accessing. An ad blocker helps, and just avoiding those pages works even better. My ad blocker is tuned to let in everything from certain sites, and helps me determine which sites to avoid by showing me the rapidly incrementing "I blocked this" count in my browser's top bar. When that number climbs too high, too fast, I just bail out of the page.
For each page of text and every image, HTTP requests a new TCP connection and tears it down after transfer.
Which version of HTTP are you using? Even HTTP 1.1 has keep-alive.
A lot of times, bad MTU settings are the problem with a sat link. The problem is simply stated: GRE tunnels are common on such links, and a GRE tunnel will encapsulate each packet and add a 16 byte header. Since the modems usually only permit a 1500 byte MTU, this means the maximum packet size you can get through the GRE tunnel will be 1484 bytes long, inclusive of header. If someone sends out a packet that is maximum size for a 1500 byte MTU, and sets the DF (Don't Fragment) bit, when the packet hits this GRE tunnel it will be dropped. This happens frequently with bad SSL implementations.
This is only one version of MTU problems with sat links. There are others.
HBI's Law: Frequency of calling others Nazis is directly correlated with the likelihood of the accuser being Communist.
Use a browser like Opera Mini on your phone. Use the browser with an Amazon tablet. Both user cloud technology to cache pages, greatly enhancing the loading of pages. Similar solutions are available on the desktop. Adblock helps reduce page load time also.So...with just a little work you can find your own solution.
Some settling may occur during posting.
unless you're a developer to start with, creating a website happens through a low-code or no-code suite, usually. (or you spend a few weeks learning html and javascript, which is actually less efficient, since you can't learn good practices and sufficient vocabulary to be good in that short time). also, with the vast majority of people using broadband, there's little incentive to work extra hard so that last 0.002% of your user base still connecting through 33.6Kbps dial-up connections has a good experience. you could have just as easily asked "why are OSes so bloated? what about those of us still working on pentium series processors?" the answer would be the same: "they don't think of you, because you're not an influential enough part of the market". imagine you're a developer tasked with creating a news site. do you spend months optimizing your work for the tiniest fraction of the market, or do you spend that time putting polish on the parts the vast majority of your users are going to actually see?
oh boy....
Your DNS can easily be local, on your network, and it means that everyone connecting to your network, regardless of device, has the advantage of ad blocking.
I don't need to jailbreak my mother in law's iPad for her to get the advantage of the DNS blocking, I don't have to root my wife's android, every device that comes in to the house and connects to the network automatically gets the advantage of ad blocking just by virtue of connecting. And best of all, I can make updates to one device instead of dozens as new ad domains appear.
Hosts is great for people who only use a PC, never use a mobile device, have few devices, and don't know how to configure a DNS server, but for true geeks (the main audience of this site) it's lacking on many levels.
There's a reason we don't all just have hosts files for the whole internet, there's a reason DNS was invented, those same reasons are what makes it superior to a hosts file for this application as well.
Providers don't offer a proxy because an HTTPS proxy requires each user of each browser on device to add the proxy's root certificate to the certificate store used by the profile associated with that user, browser, and device. Non-technical users are unlikely to successfully complete this process for both the OS-wide certificate set and the separate NSS set used by Firefox.
Why does my browser open up 6 TCP connections to try to download six images at once
The website could easily be serving those 6 images as 1 image and use CSS to display them individually.
So it's not only a browser issue.
Actually, it's a protocol issue.
I believe this sort of thing is being changed for HTTP/2, with multiplexing.
- Don't do what I do, it's probably not healthy nor safe. -
Some how win10 isn't allowing me to edit the host file even as Administrator
Since Windows 8, the administrator has to exclude %SystemRoot%\system32\drivers\etc\hosts from Windows Defender. It's under Windows Defender protection in the first place because in the past, malware has used this file to redirecdt popular social media sites to phishing sites. Go to Start (Windows logo) > Settings (gear icon) > Update & security > Windows Defender > Exclusions.
This is the disconnect (no pun intended) between internet application developers, most of whom live in a big city with huge data pipes, and the rest of the country that doesn't (and doesn't always have cellular internet either).
I'm sorry that your screw driver isn't a ratchet, and that your ratchet isn't a hammer.
But have you tried to actually lace up your shoes? They'll work better that way.
Read, learn, and re-configure your browser, your operating system, and your network to actually do whatever it is that you want it to do. Most browsers can easily be configured to limit the number of simultaneous connections. It's a number somewhere -- think registry, about::config, etc. Did you try? No.
You're on something that's both atypical and inferior. So configure it to your preferences.
Also, you could configure your router to limit connection counts. You could configure your browser to access only one domain at a time. You could load your hsts file to block ad servers.
You could choose to block most typical third-party sites when you aren't interested in them -- like blocking facebook would block like buttons everywhere. Unblock it when you want them. Block them when you don't.
You seem to be surprised that the default out-of-the-box isn't for you. Big surprise -- it isn't for anybody. It takes me three days to configure a new machine. There are about ten thousand settings to look configure to my liking. I remap keys on the keyboard, configure macros on the mouse, firewalls and hosts files. Does my browser show image placeholders for still-loading images? Security settings, privacy settings, themes, colours, mouse cursor sizes, the list goes on. Toolbars, a dozen little UI tools, notes and reminders, icons a'plenty, work drive, play drive, system drive, memory limits, processor limits, user permissions.
Your car has another 100 -- radio stations, temperatures, seat positions, steering wheel positions, tire pressures, cargo nets, folding seats, et cetera.
Geez, do you buy photographs that come pre-hung on your walls?
Make a decision on your own for a change.
I've changed my mind. Call me. I'll run through the whole thing with you. You can pay me to configure everything with your usage-scenarios in-mind. I'll happily take your money. And nearly all of it can be done remotely -- I'll need your help for a few boot-time (e.g. BIOS) settings, like boot disks, power management, and power failure recovery settings.
Don't be on a satellite connection. duh.
Then what instead? You buying?
I think that has to do with a really big company up in Redmond, which made a lot of money with a rip-off of another graphics-only operating system and produced code of dubious modularity and unnecessary bloat and complexity. I think their name starts with "M" or something.
Or was it an N?
Nintendo of America is also in Redmond, Washington, and the feature set of the Picture Processing Unit in the Nintendo Entertainment System was heavily inspired by that of the Texas Instruments TMS9928 in the ColecoVision. NES games likewise ballooned up to half a megabyte over the system's life,* compared with the more compact ROM sizes of the Atari 2600 and ColecoVision. I mean seriously, 48K for Tetris ?
* Only three licensed NES games were larger than 512 KiB: Kirby's Adventure, AD&D: Pool of Radiance, and Uncharted Waters.
If you have a shitty connection, just turn off images - you'll be amazed how much time you stop wasting when you don't have eye-candy to lead you to rubbish sites. Looking at you, Buzzfeed.
I've noticed over the last few years that most websites suck as far as loading time, even on DSL connections. There is so much garbage loaded that even a relatively new computer on a good connection loads slowly. Web developers today should all be lined up and shot.
I'm working on a system now, and for the first time in my career, I'm taking the time to really do it right (mostly because, for the first time in my career, I'm being allowed to). Even though it's mostly used in-house, we do have outside customers with access so I do make sure it runs acceptably on slow connections or devices.
Images are small and scarce. That's not exactly a big deal these days, but honestly, it makes styling easier.
There is one CSS file. It caches just fine. I just ran a test simulating a 22kbps connection - the first page load is about ten seconds, but after that, nothing takes more than a second or so to load, save for one page I've already marked for revision.
Javascript is minimal, mostly used to make nested menus load with AJAX or to do form submission with nice validation. This saves bandwidth overall, so I'd call it a win. The site does not degrade gracefully without Javascript but it doesn't require that much CPU horsepower to do what we do with it.
There are no ads or tracking. Not even Google Analytics. The designer wants to add it but I'm not gonna let it happen. We have server logs we can analyze for all the info we need, why would we send that on to Google, not to mention the perf hit?
Tested in current versions of Firefox, Chrome, Opera, Vivaldi, and Edge, and also IE10, under Windows and Linux (for browsers that run on Linux). IE misses some styling features, Firefox has some alignment glitches I haven't gotten around to yelling at the designer to fix it yet, but everything is fully usable. Safari and OS X testing will happen if we ever get a Mac to test with. Mobile browsers technically work but the site is very information-dense and nobody in their right minds would use it on a phone. (And I've tested it with Lynx, you can read it but it's not easy, and saving stuff is broken... if I ever run out of stuff to do, I might fix that, but it's not exactly a priority)
On a normal business-class internet connection (25Mbps shared across a dozen people), and a normal business-class browser (dual-core, 8GB RAM, Windows), most pages load in about 30ms. Only one takes longer than 100ms, and that's already queued for revision.
The fundamental problem is that too few programmers have a direct profit incentive to make a good website user experience. The big sites make money from ads, which are inherently anti-user. Contractors (which I used to be) care more about getting it out under the hours estimate than about quality. I work directly for the people who use this site eight hours a day, you can bet your ass I'll make this a good site for them to use.
The problem with your link isn't that your browser is opening up multiple connections, it's that your satellite link has shitty (or non-existant) QoS. I run a satellite network to a remote site, that is serving about 60 users using 3.3Mbps. Between WAAS and some carefully tuned CBWFQ, it's slow but reliable, you click on something you can be pretty sure that it will load. Heck, this link is also running VOIP for everyone, and calls are nearly toll quality even when the network is saturated (which is most of the day and evening).
...si hoc legere nimium eruditionis habes...
Didn't see you upthread where people were discussing the limits of ad blockers, and wondered if you were doing OK. But here you are.
Socialism: a lie told by totalitarians and believed by fools.
Back in the day we were used to time-shifting our collection of information, and the viewing of information. This was accomplished on BBSs - such as FIDONET - by up and down loading content for later viewing with offline viewers. You would just set up some automation to run during off times (while you were asleep for example). Back in those days -- even as slow as things were, your time didn't seem to be wasted as much as today.
I don't have a bad connection - I stream videos no problem - so I can only assume the problem is the advertising cruft layered on top. As a result, I'm in the early stages of putting together a web crawler of my own...basically I go to the same sites day after day -- so most of what I read comes from the same sources - so why not crawl those sites and draw down what I want to read at my leisure? I can also automagically separate the multimedia from the text, and deal with that as I want to - rather than how a standard browser decides to do for you.
Website owners and ad people have gotten lazy - and disrespectful of users; time to claim back our time.
Lodragan Draoidh
The more you explain it, the more I don't understand it. - Mark Twain
were planning on making the same comment if it wasn't already here?
Bloat, yes, bufferbloat, no.
Recently I made a trip to nicaragua and deployed a few fq_codel and cake equipped routers there. The effects were *marvellous*. The kind of total failure to load problems dan describes, in particular, went away.
Still, I'd like to banish those that think IW of greater than 2, TSO and GSO, and 6 simultaneous connections in web browsers - as a good idea - to remote locations for a while, until they learn the errors of their ways.
https://tools.ietf.org/html/dr...
is also a good read.
People on slow connections like satellite make up such a ridiculously small percentage of the market that it makes no sense to optimize or create functionality for them. Most people's mobile connections are faster than their home broadband these days. There are plenty of extensions out there to help you on a slow connection, click-to-load images, etc. but ultimately it's just the nature of the beast. We aren't downloading web pages, we are downloading entire applications that we are running through our browser virtual machines. Slow-connection functionality shouldn't be added to mainline browsers--they should be focused entirely on speed and efficiency. Browser extensions are the best solution.
and use the DMCA [...] or foreign counterparts
DCMA, yeah. The world does not care about DCMA, only the US.
Technically, you are correct. But in practice, two things are true. For one thing, Slashdot and many of its users are in the United States. For another, I mentioned not only DMCA but also "foreign counterparts" to the DMCA, by which I meant other countries' implementations of the WIPO Copyright Treaty of 1996, such as Article 6 of the Directive 2001/29/EC in the European Union and the "digital locks" provision of the Copyright Modernization Act of 2012 in Canada.
I have a proxy (squid) that supplies HTTPS just fine. It just can't cache anything served over HTTPS.
Polipo is the same way: it caches cleartext HTTP and uses CONNECT for HTTPS. But with HTTPS traffic having already exceeded cleartext HTTP traffic, the majority of traffic passing through the Squid or Polipo proxy will be under the CONNECT method. And as more and more public sites continue to adopt HTTPS, your cache hit rates will fall to single digit percentages, at which point there becomes less of a business case for continuing to maintain the proxy in the first place.
"No shit, Sherlock?"
Birds are not dinosaur descendants;birds are dinosaurs, for all useful meanings of "birds", "are" and "dinosaurs"