torkell: (Default)


I've finally upgraded to FTTC broadband! I've no idea where the cabinet is but it can't be far given the speed I've got out of it (and I suspect from the modem's stats there's another 5Mb/s of download I could achieve).

My previous ADSL connection was fairly decent as I'm not that far from the exchange - I was on about 11.5Mb/s down and 1Mb/s up (it used to be faster but something glitched out on my line and the exchange decided to target a larger SNR) - but the comparatively slow upload has been annoying for a while. So while doing price comparison for car insurance (protip: if you want to keep your customers, don't add a 20% loyalty penalty!) I also plugged the numbers in for broadband, and for less than what I was paying BT and Sky I could move to plusnet's top package. So I did just that!

Of course as with anything involving phones in this flat nothing was simple and it took 6 weeks to actually switch over! From the updates I got from plusnet either the broadband or the phone order would be rejected by Openreach which meant they had to cancel the whole lot and resubmit it again. It appears that Openreach's system is really confused by the concept of transferring an existing number - one sequence of messages went "transfer this number to plusnet", "you can't have that number, it's in use", "I know, that's why I said transfer". I've ended up with a little collection of "sorry you're leaving" letters from BT all with a different end of service date!

But the order finally went through last week, phone service switched across this morning (taking my ADSL out with it, as expected), and while at work I got a text to say the new broadband had been activated. And I now have shiny fast intertubes!

Fair play to plusnet, they did all the pestering of Openreach without me having to chase them and they've not started billing me until the move actually went through.
torkell: (Default)
And now for something less rant-inducing...

Today's discovery is Firebug Lite, which is a cut-down version of the Firebug web developer tools that can be run within a browser... without a browser extension. I'm not entirely sure how they manage it - presumably some horrible abuse of Javascript - but the result is that I can get at the Javascript console inside Safari on iOS without having to hook it up to a Mac. Neat!

Actually running it in iOS Safari is a little tricky - they give a bookmarklet that loads the script, but you can't directly add that. Instead you have to take an existing bookmark and edit the URL to contain that of the bookmarklet (the easy way to do this is to use a desktop browser to copy the URL into Pastebin or whatever, then go to that page on iOS and copy'n'paste the URL from there).
torkell: (Default)
Isn't it wonderful the way modern websites make it impossible to search for what you want? Either the search engine is overly precise and fails to find documents containing the words you've entered, or it's overly fuzzy with no way to say "search for this phrase only and nothing else"...



I was originally searching for "quicktime" - see, I've got this crazy idea that the Quicktime plugin might actually be a reasonable way to play MPEG4 videos in Firefox on an old laptop, and this even crazier idea that a Firefox extension could replace <video> tags with a suitable <embed> tag to achieve this. This was how video playback actually worked in the HTML 4 days, after all.
torkell: (Default)
So if you've just updated Firefox to version 34, have the Flashblock addon installed, and are wondering why videos no longer work at all... it's because the Firefox devs changed part of the plugin API such that Flashblock is now broken (they freely admit that "this has the potential to break addons"!). The only fix is to disable Flashblock, which may or may not happen automatically (on my system it was enabled, but marked to be disabled when Firefox was restarted).

Unfortunately all the workarounds range from "buggy" to "horrible". The obvious patch to Flashblock (apparently just setting a "yes, this plugin really is allowed to rewrite HTML" flag) doesn't work on some sites, Flashstopper only deals with flash videos and not flash in general, and Firefox's native click-to-play is clunky and too coarse-grained (no I don't want to enable all flash applets on a site just to watch a video).

Now in theory there would have been some warning when updating Firefox that the addon was incompatible, as each addon is supposed to declare which versions it works with. Except Firefox has followed Google Chrome's lead in continually releasing new major versions (presumably to be able to say "my version number is bigger than yours!"), with the result that the version number has no semantic meaning and version N+1 is equally likely to be fully compatible as it is to break everything. This means there's no point in specifying a maximum version that your addon is compatible with unless you want to have to release a new one every 6 weeks.


For added annoyance, the update appears to have reset the configured search engine back to Google (which is odd as Mozilla has just ended their partnership with Google). It also appears to be rather unstable - it's crashed a good half-dozen times already. I wonder if I'm better off just reverting back to version 33, or possibly going back further to version 31 (which at least will continue to get security updates, unlike version 33 which is now considered end-of-life despite having only been released for less than two months).
torkell: (Default)
We've completely forgotten how to make user interfaces, haven't we?

Today's example is the permanent link for a tumblr post, which is the cute little maybe 16x16 pixels folded-corner-of-paper icon that only appears when you mouseover a post, and has absolutely nothing to suggest that it might even be a link. Let alone the link to the current post.

Compare with, I don't know, just about any other blogging platform where the post title is the permanent link. And even if there is no title (such as on this post), then it's usually something like the post date or even something saying "link", and in any case it looks like a hyperlink.

Router++

Dec. 2nd, 2013 09:45 pm
torkell: (Default)

I've been having issues for some time now with my router, a Netgear DG834Gv3, where the ADSL connection becomes very unstable at times. The Internets suggest that this is likely due to the AR7 chipset in it arguing with whatever Be O2 Sky use at the exchange, and losing. Netgear have gone through several hardware revisions and the v4 uses a completely different Broadcom chipset. As it happens, I've got one in the spares box.

As with all good experiments, some before and after comparison is required. So, the obligatory before statistics (with the DG834Gv3):

ADSL LinkDownstreamUpstream
Connection Speed14502 kbps1203 kbps
Line Attenuation18 db4 db
Noise Margin3 db5 db

Important tip: don't try running a speed test at the same time as running inSIDDer - the searching-for-wifi-networks stuff causes frequent interruptions to the network connection and so really eats bandwidth. Presumably this is from the channel-hopping it has to do to discover the other networks.

Some reconfiguration and rewiring later, it's time to repeat with the DG834Gv4:

ADSL LinkDownstreamUpstream
Connection Speed16307 kbps1217 kbps
Line Attenuation17.0 db8.8 db
Noise Margin6.4 db6.8 db

That is... quite a significant increase in both downstream sync speed, noise margin (an increase from 3dB to 6dB is actually an improvement, especially when the 3dB would go negative at times!), and real-world speed. Interestingly the upstream sync speed has increased slightly despite a large worsening in the attenuation, though in absolute terms 9dB is still very good and 1217kb/s is about as fast an upstream as ADSL2+ can do. The reported values are a lot more stable as well - with the v3, the noise margin was all over the place. Here it's varying by maybe ±0.3dB which is nothing.

And yes, I know this has nothing on Craig's ridiculous 70Mb/s connection (and note that his upstream is actually 20Mb/s!). Though BT have just wired up the local cabinet for fibre...

torkell: (Default)
This morning I'd just finished reading a book on the Kindle and since I hadn't yet turned my computer on, I decided to quickly check my email from the Kindle. Now, the webhost has a webmail service and the Kindle has a web browser so this should all work. Unfortunately, since I last used it the webhost has upgraded to the latest and shiniest version of cPanel, wherein the only visible difference is a shiny new login page. So, enter username, enter password, select login button... and absolutely nothing happens. The Kindle will happily highlight the button as something that can be clicked on, but whatever Javascript monstrosity is actually powering the login page prevents me from, well, logging in.

Later, for a laugh, I decided to try the Wii's browser (which is based on Opera). After persuading it to connect to the webmail address (hint: if you want to connect to a non-standard port, you have to enter the full URL including the "http://" bit) I then entered my username and password and selected the login button. At which point everything on the page disappeared except for the background image. I suppose this is a slight improvement on the Kindle. Actually if you then reload the page it seems to work, but I only tripped across that by accident when hitting the Wii browser's "change layout" button.

So I pulled up the source of the login page to try and work out just how they've managed to break such a simple thing. What seems to happen when you click the login button, some Javascript takes over and sends an AJAX request off. When the response comes back it'll display a message and if the login was successful redirect to the webmail interface a few moments later.

It's all shiny and AJAXy and Web 2.0 and completely pointless, not to mention broken in at least two web browsers. What was wrong with just using a server-side redirect (like the previous version did)? For that matter, what was wrong with using HTTP-based authentication (like the version before did)?

Oh, and in Internet Explorer 8 the background image on the login page doesn't show up. That's not due to a browser bug, but rather it's a deliberate decision in the CSS.

Multifail!

Jun. 1st, 2012 09:20 pm
torkell: (Default)
Somewhere within the depths of Amazon is a RSS client made of several kinds of fail, judging by the logs on a server I admin.

Multifail! )
torkell: (Default)
Someone on Facebook posted a link to Gizmodo. Specifically, they posted http://gizmodo.com/#!5782366/someday-youll-keep-a-rainbow-in-a-cage. So, being curious, I clicked on it.

In Firefox 3.6, this redirects to http://uk.gizmodo.com/#!5782366/someday-youll-keep-a-rainbow-in-a-cage which displays the Gozmodo home page. Opera 11.01 behaves identically.

Internet Explorer 7 instead redirects to http://uk.gizmodo.com/. The initial redirect is actually carried out server-side and the location header only contains "http://uk.gizmodo.com". Interestingly the HTTP RFC doesn't define the handling of fragments with redirects (and one could argue the handling either way), so IE7 is just as compliant as the others here.

Presumably there is an article, but I'm not sure I can be bothered to go and find it given how utterly broken their website is. Not only does it appear to rely on the fragment being preserved, but it doesn't even work after the forced redirect!

Things like this are why Web 2.0 should be fired into the sun.
torkell: (Default)
With Google's recent announcement that they're dropping H.264, I thought I'd have a quick look at what codecs will be supported by the various browsers...

BrowserCan play Ogg TheoraCan play H.264Can play WebM
FirefoxYesNoYes
OperaYesNoYes
ChromeYesNoYes
SafariNoYesNo
Internet ExplorerYesYesYes


That last row may come as a surprise if you've not paid attention to how Microsoft have implemented the video tag and just read all the FOSS propaganda. Microsoft have used the existing DirectShow framework, and as a result IE should be able to handle anything that there's a DirectShow codec for. And there's DirectShow codecs for pretty much any format in existence, including Theora and WebM.

It amuses me no end that the browser usually ridiculed for poor compatibility has the best HTML5 video support.

Edit: corrected table headings.
torkell: (Default)
Today was supposed to be phone line day, but unfortuantly that failed to happen. And I can't even rant at the person who's fault it is (and I was very nice and didn't rant at the engineer - it's not his fault), so I shall instead rant at the Internets!

It turns out that I'm not only the first person in this flat to want a landline, but the first person in the whole block - the other tennants all use mobile phones. But that shouldn't be a problem, the block itself should be wired up.

Ha. Ha. Ha.

Turns out that the builders fitted a DP and wired all the flats up to that, but then left the cable out of the DP in a heap on the floor. And they ran a couple of ducts in the cupboard, but a) they aren't the same colour as the ones underneath the BT manhole cover right outside which leads to b) the engineer has no idea where they go! The landlady doesn't know how the phones should be run, and the person who would know is on holiday. So the phone sockets in the flat are all wired up, but the cable between the flat and the exchange isn't connected and no-one knows where it should run!

With any luck BT'll send another engineer soon to sort out the ductwork (the one today didn't have the necessary equipment), and hopefully that one will manage to get it wired up.

At least at the end of this I should get an excellent broadband connection as the new flat is much closer to the exchange!
torkell: (Default)
...since when did my connection get upgraded? Well, that explains the glitch last night.

I just downloaded a 4.21MB file in 6 seconds. I thought this link was only supposed to be 2Mb/s. Let's see...

Speedtest.net claims this:
Speedtest.net: Download: 6.54Mb/s, Upload: 0.79Mb/s, Ping: 34ms

Pure internet speed tests are however generally a bad way to test your connection speed, as there's far too many variables in the list. A much better way is to download something chunky from as close to you as possible, and time that. My ISP runs a public ftp server, so let's download something from it. According to my FTP client I managed a transfer rate of 7.64Mbps (I'm not sure how it worked that out), which as about what my connection is actually synchronised at.

Yay faster internets!
torkell: (Default)
Well, that's rather annoying. I had a wonderful plan for today's (ok, technically yesterday's but I've not gone to bed yet) post, involving an inline version of last year's Storyteller. Unfortuantly this was scuppered by several things: 1) LiveJournal does not allow IFRAMES, 2) using the OBJECT tag with HTML doesn't reliably work in IE6 (it loads the page but doesn't display it), and 3) when you use the OBJECT or EMBED tags in a LJ entry, what actually happens is their backend wraps it in an IFRAME using the lj-toys.com domain. This eats the referer, and so I can't find out which user is viewing that entry.

I could probably make it work by dynamically generating an image and including that - images still get directly included, and so the referer header should still be present. Doing it that way feels like a horrible hack, but that's the Internet for you.
torkell: (Default)

Virgin media have recently started redirecting missing DNS queries to their own search page. So, if you go to an address that doesn't exist, like http://www.bad.pleaseremoveyourfeet.com/ (a non-existant subdomain of a friend's site), you get redirected to a search page like http://advancedsearch.virginmedia.com/subscribers/assist?url=www.bad.pleaseremoveyourfeet.com. On the one hand this breaks the expected behaviour of DNS, while on the other hand this is much better than Verisign's Site Finder as a) only Virgin Media's customers are affected, b) they seem to be applying actual intelligence to this system and c) you can opt out of it in a way that actually works

So, on the whole I don't mind this. It would have been good if they'd given some notice of this (though I recall a Register article from a while back talking about this for their cable customers), but this is far superior to Verisign's attempt to take control of the internet, and for non-tech-savvy users it's probably ideal.

Technical ramblings )
torkell: (Default)
At the moment, I can only get to LiveJournal via a mobile Vodafone 3G connection - when I try my landline Virgin Media ADSL link, the packets get stuck in a routing loop.

That's not the weird part. The weird part is both connections at one point go through the same routers (marked in bold)!

Vodafone 3G traceroute )

Virgin Media ADSL traceroute )

So, who do I pester about this?

Edit: at some point over the past day this mysteriously fixed itself. Of course, something else had to disappear from this connection to maintain the cosmic balance of the routing tables :(
torkell: (Default)

Virgin Media recently did a free connection upgrade...

ADSL Link Downstream Upstream
Connection Speed 8026 kbps 992 kbps
Line Attenuation 41 db 11.5 db
Noise Margin 9 db 5 db
torkell: (Default)
Today's discovery: IPCop's traffic shaping is really odd.

About a year ago, I set up traffic shaping at 10240 kbits/sec down, 448 kbits/sec up and forget about it.

Today, I was puzzling over why the slower ADSL2+ connection (3.5Mb/s down, 1.2Mb/s up) actually performed better than the ntl cable (rated for 10Mb/s down, 512kb/s up, in practice about 4Mb/s down) and realised that all testing with ADSL2+ had excluded the IPCop firewall. So I disable the traffic shaping on a whim, and download speeds doubled.

Testing with Virgin Media's FTP server, I just managed to download a 252MB file in about 6 minutes 30 seconds. That's 6.55Mb/s. That's about 2Mb/s more than we generally get from this connection.

Right, anyone know how to set up IPCop traffic shaping such that it doesn't suck?
torkell: (Default)
So there's an interesting little debate going on on a certain forum I frequent.

The gist of it is that member A was banned from the community-run listservers (the official ones having gone 404 a long time ago) for crashing servers. He was originally banned for one month, but that got extended for various reasons to 2 months. The ban officially expired on 30th November last year (2005).

Members B and C run a listserver each. Member C unbanned member A a while back, as per the original agreement. Member B has not, and has expressed no desire to ever do so. There is possible circumstancial evidence of member A continuing to crash servers, however nothing is currently known for certain (and there is also less circumstancial evidence of it being another member).

Given that, who out of members B and C is taking the correct action?
torkell: (Default)

Seen on a 404 page:

ERROR

Please be advised that you almost broke the Internet. Trying to get somewhere? Use the navigation at the top of the page. If you're still having problems, update your browser or your version of Flash. Just stop whatever it is that got you here because it's seriously freaking us out.

torkell: (Default)
2Mb/s ADSL and large downloads still stall.

December 2025

S M T W T F S
  1 2 3 4 5 6
78910111213
14151617181920
21222324252627
28293031   

Syndicate

RSS Atom

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags
Page generated Dec. 24th, 2025 05:16 am
Powered by Dreamwidth Studios