OmegaDelta » internet

Request: browser tabs should be default:mute

Will — Thu, 09 Jun 2011 09:32:38 +0000

The other day I found myself with about 200 tabs in 5 windows opened. Travel searches (flights, visas, etc), development, questions, and some entertainment.

Then suddenly some music starts playing on my computer. Where is it coming from? Who knows‽ I had to go on a big hunt.

Since I rarely want the non-focused tab to be emitting any sounds whatsoever, I really think a default:mute policy would make a lot of sense here.

Or at least, some visual indication of which tab is emitting sounds, maybe by replacing the favicon.

Preventing Google from indexing AJAX HTML fragments/partials.

Will — Tue, 07 Jun 2011 14:22:46 +0000

So you’ve created a neat new site using AJAX. Your server code renders up partial HTML pages which you slap directly into the DOM. Problem is, somehow Google found these partials (even though they are only referenced in Javascript), and has indexed them. Furthermore, because they are not whole HTML pages (and not designed to be), there is no tag where you can place a robots tag to prevent them being indexed.

Enter the new X-Robots-Tag HTTP header. You can set this header to ‘noindex’ when you return the HTML fragment, so that if google somehow finds it, it still won’t cache it.

Details are here.

I applied this today to geospike.com, Google indexed a few of these -less pages which is pretty terrible if the user clicks on them (they see unstyled crap), hopefully they will drop from the index shortly.

Link Shorteners and the Eternal URL

Will — Mon, 17 Jan 2011 02:16:30 +0000

URLs are (or should be) eternal. Once you create some content and slap it up on the web, people will start referencing it. If you change the URL or remove the content, then these references break – which is sad.

Enter link shorteners. All of a sudden, you are referencing a whole bunch of content (your own, and others) through an additional 3rd party. So now you not only have to worry about whether the referenced content will still be there, but also that the link shortener hasn’t run out of money and closed up shop too.

A big problem I see is that link shorteners earn no money (read: operating costs) when you simply click the links. They *do* get brand recognition, which helps drive people to their site to shorten links of their own (which in turn can earn money through ad impressions, etc). But what happens when we’ve all moved on to something new. The link shortener won’t be earning as much money (nobody shortning urls any more), but still has to pay to keep their infrastructure online.

This troubles me. And, with log.ly – a new link shortener for Geospike.com I am also now a link shortener operator!

To combat this potential issue, I propose the following idea: Somebody reputable (The Internet Archive would be a good candidate) creates a global link shortener database with 3 basic keys [domain, link key, full link]. Link shorteners can then periodically – or just when they go bust – import their database into this master database.

Then, if they do go bust, then can transfer their domain to The Internet Archive, to operate the links on a read-only basis. In the event such a transfer is not possible (such as Libya suddenly commandeering all .ly domains), then at least people could manually look up the short link through a form on The Internet Archive’s website. If link shorteners operators are concerned about people farming the DB, then perhaps the archive could be private, while said link shortener is in active operation.

Without such a service, the situation link shortener operators would face if they became unprofitable would be to keep paying the bills for the good of humanity, or hear a great many URLs cry out in terror before being suddenly silenced.

The short URL for this post is: http://b.log.ly/fmD1EG ;-)

AWS Customer Support not like Amazon.com’s

Will — Sat, 06 Nov 2010 02:20:21 +0000

AWS, or Amazon Web Services is Amazon.com’s cloud computing arm.

They do some pretty cool stuff like S3 storage and EC2 cloud computing.

Recently EC2 announced the availability of a free cloud “instance” (basically a linux box running in the cloud). (google ‘aws free tier’ for info). Publicaly they say it is to “To help new AWS customers get started in the cloud”.

Seems like a good deal to me. Lets my startup save $200 in the first year and experiment with the cloud.

However, unfortunately I have already been using an AWS service – S3. While many EC2 instances will use S3, the inverse is not true. I use S3 on a standard shared hosting setup and it works well. I would argue that this doesn’t mean I am using “cloud computing” yet (since I’m not running any programs in the cloud, just hosting some images there).

I applied to AWS Customer support to use the free tier, and I was denied, purely because I had already used S3 in the past. Given my S3-using site is not even live yet, I even asked if I could “delete my entire account and sign up again” (what better way to prove I am new to cloud computing, if not a new signup?).

Regretfully, the AWS Customer Support refused (albiet politely). Now I can’t understand why they would deny a free EC2 instance to a startup like me – given that I am probably the exact type of customer they want to attract with the free plan – new business, thinking of getting their feet wet with EC2, but unsure how to proceed.

So in conclusion, I have to say that the AWS Customer Support is not nearly as understanding or accommodating as their Amazon.com counterparts (who are absolutely amazing at acknowledging and re-shipping lost orders, for example).

All this has left a bit of a sour taste in my mouth, because I can’t really work out why my startup should pay $200 for something almost everyone else who is in my position gets for free. It seems like a poorly thought out, inflexible arbitrary policy.

And now I’m considering other providers.

WHR-HP-G54 running OpenVPN

Will — Tue, 12 Oct 2010 07:00:24 +0000

Today I set up a second wired+wireless network in my home – this one running a secure VPN (provided by StrongVPN).

What follows is my setup experience, and some links that were helpful to me. YMMV.

Firstly I had to install the DD-WRT firmware. This is an open source firmware that runs on a bunch of routers (I chose the router specifically because I learnt it was capable of running this firmware well). Make sure you check before you buy.

Following the steps here worked fine (on OS X). Apparently you have to get the timing just right, I managed to get it to flash on the first attempt. Now that DD-WRT is installed, I can easily re-flash the device with my desired version of DD-WRT (in my case, the vpn build) from the web interface.

My first crack at setting up the VPN failed (using the latest vpn version from here), I think because I had too little available flash memory (they say you need 5.5kb and I didn’t).. The helpful folks at StrongVPN linked me to this version to try, which worked a treat (after following the StrongVPN setup instructions to the letter). StrongVPN is not only a great service, but they actually officially support DD-WRT, provide a single shell script which you run on the device to install it, and can even provide live troubleshooting help. wow.

$ scp ovpn272_ddwrt.sh root@192.168.11.1:/tmp/
$ ssh -l root 192.168.11.1
# once you're connected...
$ sh /tmp/ovpn272_ddwrt.sh
# which should output something like "size: 27011 bytes (5757 left)"
$ reboot

First troubleshooting tip: make sure you can actually connect to the VPN using your computer (via the router’s network) before you try to debug it on the router. If you can’t connect from your PC then what hope does your router have?

If you’re uncomfortable working in a bash shell with ssh/scp tools, then these steps may be pretty difficult for you. Otherwise it’s not that hard.

So now I have two networks to pick from my devices which is handy. Previously the VPN would only work on my laptop, whereas now I can use it on all devices, and also switch rapidly when needed.

WHR-HP-G54 is a damn fine piece of hardware (even if you don’t need DD-WRT). I recognise it from the awesome Ace Inn Shinjuku hostel which had one on every floor, and I’ve seen the distinctive default SSID network names around (they assign a random one, not a stupid one like ‘dlink’).

Starting and stopping:

killall openvpn
sleep 10
openvpn --daemon --config /tmp/ovpn/ovpn.conf

A tale of 2M

Will — Mon, 11 Oct 2010 04:26:16 +0000

If you want something done right, do it yourself :-/

I moved into this house back in February. It has a 1M unlimited internet connection. I was happy with this at the time, it was better than anything I used throughout 2009, but nevertheless I asked to be upgraded to the best available. The account is in the landlord’s name (in China, the internet is like gas and electricity – they don’t switch it off between tenants which is really freaking smart). So I asked my landlord to make it happen.

It took so long I basically gave up. Until this week, it really started to piss me off (the iPhone SDK is now a whopping 3.5GB which takes a while).

Landlord said he had tried many times to upgrade me at the shop, but they didn’t let it happen. We tried the chinese language hotline and they said I had to go to the office with my ID card and the landlords.

So today I rang the company (10000), got the English hotline, spoke to #9434. Asked to be upgraded. They said sure. Cost is ¥1958 for the year (pro-rated for the remainder of this year as I’d already paid for 1M). Contract is 3 years with a minimum of 1. I asked what happens if I cancelled it within a year – she said I’d have to go to the store for that (???). She then asked me the name of the landlord (I gave her the surname but only knew the english nickname – “close enough” she said). Then she asked me for the ID card number, which fortunately I had from the lease agreement. And now I will have 2M (and still unlimited) internet in 24 hours :)

I also feel the ripe fool. Why didn’t I do this back in Feb? Old assumptions I guess – no Australian teleco would have let me do what I just did. It seems even the chinese ones don’t – unless you ring the English hotline! We get special treatment (probably because they know if we go to the shop an hour of miscommunication would ensue). how nice…

Hope the contract I agreed to on behalf of someone else wasn’t too onerous… :-/ (shouldn’t be… I think).

petition @GoogleChromeDev to reinstate full URLs (with the http://) in the address bar http://act.ly/2dw RT to sign

Will — Wed, 15 Sep 2010 04:15:28 +0000

Lets get put the http back in the address bar of Google Chrome. Please read & sign my petition here: http://act.ly/2dw

Steven Conroy Announces Additional Protocol Filtering

Will — Mon, 14 Jun 2010 12:06:01 +0000

PARLIAMENT HOUSE, CANBERRA: Today Steven Conroy accepted the mounting criticism that his proposed “Great Firewall of China – Down Under” would be inept at achieving it’s stated goal of preventing distribution of refused-classification material due to only filtering web traffic.

Henceforth, he directed the engineering team to add the GOPHER, TELNET and ~~LIBERAL.ORG.AU~~ protocols to the filter.

When asked to quantify just how much porn was shared over GOPHER, he waved his hand and said “I am not the Communications Minister you are looking for” and promptly left the podium to stunned silence.

._.

MiFi

Will — Wed, 09 Jun 2010 09:34:53 +0000

Gigamom is clueless on the actual problem, confusing cellular and WiFi. The issue isn’t Jobs asking people to turn off their abusive MiFis,or AT&T’s problematic network. It’s that devices like the MiFi are an incredibly stupid replacement for a one-foot USB cable, which is all these audience members were using it for. 802.11 does not support AP densities of more than three per 100′ radius under the best conditions. The WWDC had more like 50 per 100′.

If the clueless MiFiers (and other cellular personal hotspot users) simply USB-tethered their hotspots the problem would have never occurred. It’s scary that so-called “developers” couldn’t predict this problem and sensibly head it off.

“MiFi Jamming” is a usability killer in many Starbucks now, where a dozen or more thoughtless users smear the room with unnecessary interfering 802.11 radiation, jamming access for everyone trying to use Starbucks’ wireless. All for want of a one-foot cable that ships with the cellular modem! Idiots all.

The thing is, the 802.11 standards committee predicted this problem FIVE YEARS ago. MiFi is just the first wave a self-defeating wireless decablers, wielded by clueless users that ignore spectrum realities, potentially destroying WiFi as a useful technology.

Packetguy on http://gigaom.com/2010/06/07/steve-jobs-survives-gizmodo-but-not-mifi/

Using Wifi to network a single device does seem like overkill.

Conroy is an idiot

Will — Tue, 01 Jun 2010 04:29:53 +0000

It’s confirmed

My sentiments exactly. There is no need for an internet filter to slow down the internet and remove our liberties, while not even satisfying the task it was put there for.