Archive for February, 2006|Monthly archive page
We start with a quick Announcement from Feedburner which is an open API to Feedflair !
it’s a big panel tonight, first question is about how the web now has lots of apps that do one thing very well and what do the panel think of this, the response was Yes!
now identity, mention of Sxip and Mylid then one guy pitches in with the statement that everyone uses the same login and password for all the services anyway so it’s not needed.
– Joshua chips in with why not skip all that and have authentication URL sent to you to get it, i quite like that.
– desktop apps will become more llike web apps
– a short discussion of how do you price apps, given the level of detail we go into on this I reckon it’s covered
and we close and the day is over.
Future of Webapps > Steffen Meschkat > Greater Expectations – Reality-Checking the AJAX Web Application Architecture
an overview of how AJAX is a bad name etc etc again.
some differences between REST and SSSS apps and AJAX like whole page refresh versus updates
AJAX enables complex interactions and manipulations and fixes the architectural problems of client and server session states
AJAX – what’s really in it
- layout, fonts colours
- just two words:insert semicolon
- single threaded, but nobody tells you so
- Data Marshalling
it’s very hard to understand this guy as he is speaking very fast and has a relatively strong German accent, plus the slides have very little text, in fact I’ve never heard anyone talk so much about one slide with so few words!
cross browser compatibility
seperation of interaction logic and application logic
Due to his speed and accent and possibly to a degree his subject matter the last speaker of the day is losing the crowd..
I got to my laptop last night and a new screen had appeared that asked me if I wanted to save all my conversations in my Gmail account, this is cool I thought so I accepted it (mind you I always do of course) and I even screenshotted it but forgot to paste it and save it so missed that.
The actual functionality is very cool and I like it, you get a message/chat per user/per some period of time, if you close a chat and go back after 5 or so minutes it goes into the same file but if you leave it for a hour or so you get a new one which is not as cool as if you always had one per user per day!
They are all stored in a new folder in Gmail called ‘Chats’
Once you open a file it’s a very clean (what else!) version of the exact chat you had, all in all very cool and a reason to try get more of my buddies to start using GTalk…
figure to spend < $30,000 for a web2.0 startup app (in relation to most projects we do that is small!)
DropSend is there to solve the problem of sending large files – simple
make sure your idea is financially viable – be cautious – estimate and then cut by 45%
forget about acquisition – aim for profit
forget about the “is this another bubble>” stuff
Branding and UI Design: £5,000 (Ryan Shelton)
Development: £8,500 (Plum Digital Media)
Desktop Apps £2,750
Hosting/Maintenance: £800 per month (BitPusher)
Linux Specialist: £500
Merchant Account: £200
Payment processor: £500
Total £25,680 all in for an operational company
Now on to build your team on a budget:
avoid the ‘RockStars’ as expensive and busy
offer a % of equity in the end product
next up scalabillity on a budget – i.e. buy just enough
Don’t waste money – like £1000 on stationary – dumb
Holy Crap – Lawyers are expensive !
Use cheap software like Basecamp, Trac (bugs), Skype & AIM, Subversion (version control) and LAMP
Use cheap hardware like a £200 linux box and if needs be plug it in to your office
Dont spend anything on marketing, use blogs, word of mouth, viral etc
Overall an interesting expose of how much it cost to get a business up and running that now has 950,000 users for less tham £30,000 – makes me wince when I think about the project budgets we have.
The main sponsor ‘Adobe’ gets a slot to push Flex 2 (in beta)
AJAX and Flex – shared web2.0 themes – engaging UE – seperation of UI and data (RSS, public API, open data formats)
an AJAX programmer can switch to Flex easily
key point – allows easy offline and online working ! better multimedia functions
seamless experience across browsers, OS and devices (including increasingly mobile devices)
We have started looking at Flex in the Cycom project and after this session need to investigate more
98% of connected PC’s have Flash installed !
and now the science bit
he now does a demo of how he built a version of Flickr in a morning, and shows real time message collaboration and picture sharing built within the app, quite stunning..
you can now also ‘view source’ and see all the code that built the app that you are looking at, cool
we got given a cd with all the tools on it so will give it a try at home
check out their Labs page
What is an API – can be likened to a bank ATM
connectivity problems prevented too much on this session – however it was very dry, there was nothing really new here in terms of the wider scope and this conference.
APIs are good, you need them, you should have them, they add value, they bring value, they help expand your business
the talk ran very short so for the first time today we switched to a questions session which focused mainly on Mint which is the product Shaun built
Future of Webapps > David Heinemeier Hansson > Happy Programming and Sustainable Productivity with Ruby on Rails
Ruby on Rails framework
introducing the Silver Bullet – which is – TaDa – motivation – which comes from happiness – not sure where this is going but I’m sure the reveal will happen soon..
your app is not a snowflake – you are not special (crowd laughs) i.e programmers copy each others code to the same mundane tasks
we move into some very detailed Ruby code on screen and he talks us through it – too detailed for me!
Every developer starts off with a Devil and an Angel, the Devil says “don’t worry you can tidy up later – you’ll have more time” the developers in the crowd are lapping this up
we get the first full on clap for the reveal that PHP is the Devil as it encourages coders to be sloppy
some more things that Ruby enables you to do by making you code cleaner
<getting conference back and wrists again>
A good speech on the benefits of clean and structured programming – the inside slant for us is that it is getting very easy to code apps with things like ruby – one to watch out for
missed the first 15 mins as not allowed to bring my hot coffee into the hall….
money to be made from APIs – again not getting a new message here, less central development, syndicate platform.
the last point I agree with is that “we will see more paid for API services” – lots of knowing laughter again as we move onto how you have to be part of the ecosystem – he is trying to sell us a vision it turns out.
How do you add value to the aggregate web ..
can you find ways to help people explore data from multiple data sources, network effect etc etc
Now we move to architectural principles – I hope it’s not another 101 on data use, ref to Matt Biddulph application of web ? to data
we move to 9 or so principles, they are good ones and the basis for good design, i.e. get the data right not the pages, I assume he means get the schemas right and then apply a UI as this can change, the underlying strucure must be capable though
Identify first order bit, very true, need to think about what that is in Cycom (and the URL schema)
good URL must be a permanent forever reference, Tom has a fetish for URLs apparently – guffaw from behind me, do not expose the underlying structure but reflect the data structure so you can ‘hack’ navigation
Tom is a very good presenter and has the crowd in his grasp, he has good material too and throws in the odd quip, he has a lot of ‘fanboys’ in the crowd too which helps of course and every quip gets a laugh
have list views and batch interfaces, good for interacting with lots of data, Flickr Organizr on screen, there is a lot of nepotism here, it’s UKETech on a small scale
we have Flickr on screen as an excellent example of Ajax in use, elegant etc etc, we are on to Odeo, basically the darlings of Web2.0 are up for show and I have to agree, these a probably the two best on the web right now….
nice summary of APIs over microformats over Parallel XML and RSS etc
and then a one page summary of the principles and a quick ad for Yahoo developer page, the obligatory “all my pictures are from Flickr” ” my font is …..”
time for lunch
Cal Henderson from Flickr
Architect at Flickr – he’s very web2.0 cool as he wears shorts and it’s not that warm in here or outside
10 reasons to love web2.0
- Open APIs
- Clean URLs
- Desktop Integration
- Open Data
- Open Content
It’s Flickr’s 2nd birthday in SF with free cake for anyone that can come – first geek laugh of the day
he’s not the most confident speaker, lot’s of ‘er’ ‘er’ repeating some of the things Joshua stated about user behaviour etc, generating passionate users
bit of Flickr history and how they built a MMOG and then used the same tech to build Flickr
collaborative metadata and aggregation of views like latest photos from everyone etc etc
he means web services API, REST, SOAP etc but one ‘naive’ programmer once told him that all programs have an API – duh!
start off with a read only API but open it all up but then once you understand what is happening you can go for write API as well, it allows other people to build their own UI and Flickr is just storage, latest app built is Fastr
some idiot in front of me is busily uploading to Flickr so he can be cool and get piccies in the pool first, don’t worry about the slow bandwidth or requests not to do this
interesting note, he types in a URL from memory and have in the past removed the actual link to the page from the site, be careful
AJAX is possibly the worst product name ever – 2nd GL the other worst is XMLHTTPRequest
<I’m surrounded by people with hacking coughs, in fact there is barely a minute goes by without a cough or sniffle, says something about the sort of people here I guess>
when using AJAX think carefully about how to map URLs into the app
This is basically another 101 lesson on web2.0, he starts going into some actual details of unicode and states he is getting too ‘geeky’ now which gets a knowing laugh from the crowd, however if they think this is new then they aren’t knowing geeks, and certainly not alpha geeks 😉
He asks us to remember how WAP affected our lives and we get a big laugh – haha, he moves on to browser implementations and how dificult it is, at last he has a point XHTML mobile profile is the app to build against, he makes an amazing statement that not all data on a web page is suitable to be used on a mobile….
another comment on letting people feel they own all their data and are free to leave at any time, this encourages people to come and stay..
a quick stint on how Flickr lets the user own the data unlike some other sites which actually own the data when it is uploaded to their service
we are asked to check out blog.flickr.com for a person that has created a song using creative commons licensed pictures
http://iamcal.com/talks for the presentation
and that’s it
- browsers are a massive pain in the butt
- first off, don’t do it because what you predict as a problem is most likely not going to be the proble after all
- read the Cal Henderson paper
- monitoring system
- make sure you know SQL, partition, seperate tables, indexes etc, tags dont map to sequel
- use caching everywhere
- figure out where you can get away with being sloppy, RSS feeds can be slow
- idiots are smarter than you, they will find a way to break it in ways you dont know
- learning the innards of Apache is a dark art
- put a proxy in front of it to sttop everything coming off the same server etc
- build APIs early !
- make them as simple as possible, del is not even REST its HTML
- no API key required to get access
- not infintely scalable
- dont expose the internal identifier as some bonehead will scrape them
- the ones you take out are important as the ones you put in
- freeform tags are
- dont add in a feature from another site that someone wants you to add
- thinks important, allows another set of tools
- put them on everything you can
- it used to be the thing that made people come to try out a site but he doesn’t think so now, maybe tags are
- make them plain and simple to allow people to paste the around
- look for a new behaviour, decide to amplify, modify or reduce
- do something that you really love or need
- he started out with a text file full of 25,00 URL in 2000 which became a database for his own use which then became a public system in end of 2003
- lots of holding pages, closed betas ,limited invite, he thinks it’s a bad idea, too closed and quiet, you dont get the interaction
- any sort of aggregation of attention becomes something that people are interested in
- attention hogging
- they don’t do a top 10 because the long timers would always be there but also it would be something for spammers to have a go at
- dont give the spammers any attention – doh !
- they are about UI and fairly useful for recall and ok for discovery
- no point in auto tagging as defeats the object
- there is a feeling at the moment that if you build something that a few people use then lots more will use it
- be careful about where you spend your effort – doh!
- if a user is using something in week 1 are they still using in week 5
- also the system itself of course
- look for behaviour rather than claims for what they are doing
I realise I am blogging verbose almost everything again and there really is no need and it will kill my wrists…
It’s a good summary of how to build an app but if you have built a system before it’s a 101 you dont really need, interesting points only from now on
For testing go to Starbucks and buy people a coffee and chat about what you are doing.
Allow a user to completely remove himself from a system and to completely remove data too
You have to infect every communication means you can, try and turn RSS feeds into a client app almost
Any questions – ooops there is no time left and we’re hitting the break