Dave Verwer

With the hype growing every day, I figured I was being left behind by the cool kids in not having my very own Facebook application!

Overall I like what I see there, they do not allow any custom JavaScript in FBML apps but do give you some reasonable tags to do AJAX callbacks and stuff (Note to self: I need to test how badly these degrade) and basic hide/show markup stuff which is useful.

However I just had to blog about the pain I have had with the rFacebook Gem. From what I can gather, the latest version of the gem simply doesn’t work as every time I redirected via Facebook to login (a necessary step) I got the following error:

NoMethodError (You have a nil object when you didn't expect it! You might have expected an instance of Array. The error occurred while evaluating nil.length):

/usr/local/lib/ruby/gems/1.8/gems/rfacebook-0.6.4/lib/facebook_rails_controller_extensions.rb:37:in `fbparams'
/usr/local/lib/ruby/gems/1.8/gems/rfacebook-0.6.4/lib/facebook_rails_controller_extensions.rb:60:in `fbsession'
/usr/local/lib/ruby/gems/1.8/gems/rfacebook-0.6.4/lib/facebook_rails_controller_extensions.rb:130:in `require_facebook_login'
[snip...]

Much Googling, wailing and gnashing of teeth later I found someone talking about this issue being specific to 0.6.4 of the gem (the latest) and sure enough downgrading to the 0.6.2 version of the gem and everything is sweetness and light.

Anyway, a morning of frustration out of the way and I am proud to present…

Apart from the gem issues the development process is fairly smooth. The FBML Test Console is nice enough and is very useful to prevent hundreds of checkin/deploy cycles to test out your markup as your site (obviously) has to be deployed to be accessed through Facebook.

I am hoping to build a little app for fun (more than “Hello World”) so expect a few more posts here as that develops…

Since the presentations section of the Shiny site is gone for now as we have a reshuffle over there and I haven’t got around to putting up a presentations section here yet, I thought I would add a quick post tonight to make the slides available from the various presentations I have given recently.

• Ruby on Rails for .NET Programmers
• Unobtrusive JavaScript
• Introduction to the afeeda API
• Hosting and Deploying Rails

If you have any questions about any of these, please contact me.

Note: I would love to put these on SlideShare but it really seems to bork on Keynote PDF’s :(

I have been thinking of doing some work with Flash and/or Flex on afeeda recently and one thing that has always held me back from considering Flex is its dependence on v9 of the flash player.

Now we have a decent amount of traffic for afeeda, I thought it would be worth taking a look at the version distribution across flash players for our visitors.

That is 89% of visitors ready to run Flex content, that is pretty reasonable. I am not planning on being dependent on any Flex so I think that is the answer I need to go ahead with investigation/development with some Flex now.

The only worrying thing is why people have such a huge distribution of v9.xxx players, I thought the flash player had an auto-update in it after a certain version so what is up with that?

Silverlight will really have to hustle to get that kind of penetration.

Disclaimer: These are the stats for afeeda only, your visitors may differ to ours.

As a Camino user, I am a huge fan of bookmarklets for browser add-on functionality as Camino doesn’t support Firefox extensions so when we created afeeda we naturally created a bookmarklet for browser integration rather than an extension or IE toolbar, get to all browsers in one… especially since the JavaScript that we need to execute is so simple:

javascript:location.href='http://afeeda.com/dashboard/discover?u='+encodeURIComponent(location.href)

This has been working fine in all browsers until someone reported a bug with it tonight on IE7. It seems that if you execute JavaScript from the feed view of IE7 you get this error:

Ignoring the oh so useful error message for a second, it seems like this is the culprit:

Feed View in Restricted zone

The IE7 Feed View displays feeds in the Restricted security zone, no matter where the feed originated, even if for example the feed came from a site in the Trusted Sites zone. By default script is disabled in the Restricted zone. In addition, the Feed View disallows URL Actions including script and active content.

(via Microsoft RSS Blog)

What a stunningly effective way to break all bookmarklets which were just about the only way to get cross platform browser add-on functionality, good job! According to that article, all active content is already being stripped out of any HTML content in the RSS before it is displayed so this extra step seems like overkill somewhat.

Thanks to Matt Brindley for helping out finding the reason for this.

I have been doing some work for a company recently who wanted to deploy their rails application using capistrano but had a problem in that their remote servers were firewalled from their development network, which included their Subversion server.

The solution still seemed absolutely ripe for automation, it just meant we had to dive into writing our own capistrano task to basically do the following:

• Check out latest version of code from subversion (svn export)
• Zip exported source up into a file
• Transfer the zipped source to the deployment server(s)
• Unzip into the relevant releases folder on the deployment server(s)

So, I spent a few hours on Sunday morning tidying up what we did and making it a little more generic and I thought I would share it on here in case anyone else would find this useful.

The capistrano task is as follows:

desc "Export source from SVN, Upload it to an SFTP location, Download it to the release path" 
task :export_svn_deploy, :roles => :app do
  tmp_path = "/tmp" # A path to a *local* temporary directory
  sftp_server = "your.sftp.server" # The SFTP/SSH server for storage of the zipped source
  sftp_user = "sftp_user" # The username for the storage server
  sftp_path = "deploy_tmp" # The remote path on the storage server
  tmp_tgz = "exported_src.tar.gz" # Temporary file name for the zipped source

  # Export from SVN, zip and upload to the remote storage server
  system "svn export -q --force #{repository} #{tmp_path}/src/" 
  system "cd #{tmp_path}/src/ && tar czf #{tmp_path}/#{tmp_tgz} ./" 
  Net::SFTP.start(sftp_server, sftp_user) do |sftp|
    sftp.put_file("#{tmp_path}/#{tmp_tgz}", "#{sftp_path}/#{tmp_tgz}")
  end
  system "rm -rf #{tmp_path}/src && rm -f #{tmp_path}/#{tmp_tgz}" 

  # On each server, scp the file from the remote location and unzip it to the release_path
  run "mkdir #{release_path}" 
  run "scp -q #{sftp_user}@#{sftp_server}:#{sftp_path}/#{tmp_tgz} #{release_path}/#{tmp_tgz}" 
  run "cd #{release_path} && tar xzf #{tmp_tgz} && rm #{tmp_tgz}" 
end

Notes

• The remote_path on the SFTP storage server must exist or the SFTP commands will fail.
• It is assumed that you are using ssh keys and no pass phrase to connect to the SFTP server from both the client machine and also all of the remote servers, this way both the Net::SFTP and the scp commands will not need passwords to be sent.
• The mkdir #{release_path} before the scp should not be needed if this is being used in a real deployment, let the regular capistrano tasks take care of that
• I chose to use Net::SFTP on the client to upload the file because this script was originally developed for a client using Windows and Net::SFTP works whereas shelling out to do an scp, wouldn’t. The rest of this cleaned up task however would need some modification to work on Windows as it stands as the tar commands would fail.
• This blog post needs less notes ;)

Next Steps

I am pretty sure that the best way to implement this would be as a custom source control plugin for capistrano so by doing:

set :scm, :svn_via_sftp

and have some extra options to specify SFTP server name and user account then this process could be completely seamless with a regular cap deploy.

I would be really interested to see if this is a common scenario, I suspect it will be with larger clients who do not want their SCC systems exposed to the web. Is anyone in this situation?

So like everyone else I have been reading the back and forth between Alex from Twitter and DHH about scaling Rails up to the level of traffic Twitter is seeing at the moment.

However, I was thinking about this tonight and the numbers just don’t sit right with me:

At that time they were fielding spikes of up to 11,000 requests per second across some 16 cores with very little caching thrown into the mix to mitigate. No wonder their site had been feeling slow.

11,000 requests per second!! That seemed very high so I did a little digging for stats from some other sites:

Even if you take the Twitter number as a peak against the others which are averaged out across a month, it just seems way too high, YouTube could spike to 7x its average traffic before it came close to the twitter peak. I know it has become very popular (and I am a fan, in case you think I am just biased against the service) but I find it hard to believe these numbers.

If these numbers are true, then it is great news for Rails on MySQL. If Twitter can just about scale to this level of traffic across only 16 cores with one back end MySQL server and “very little caching” then that pretty much blows any performance complaints about Rails out of the water.

However, even if those numbers are out by an order of magnitude or even two, I still think that this is good news for Rails scaling because it shows that it follows exactly the same scaling pattern as other web frameworks and you will eventually become database bound and that is the hard problem to solve. Even better I think it shows the power of Ruby/Rails that as a response to this argument Nic Williams has released a gem which implements multiple database server pooling in ~75 lines of Ruby and about a day or so of work. Using a master write and multiple read database cluster can scale a massive amount more than a single DB or an Active/Passive set-up.

Note: The above quote is taken from DHH’s article, I am sure that the original interview article mentioned 11,000req/sec and you can see plenty of references to 11,000req/sec in the comments it seems to have been edited now, maybe I am right to doubt the numbers.

So my cousin-in-law (if there is such a word!), Simon Wheatley is starting to learn Rails and after writing his very first line of Rails code and struggling for a while eventually IM’ed me for some help. I must admit I assumed it would be a bug in his code, I mean who gets it right the very first time? However a little investigation revealed it was indeed a bug.

What was the line of code?

scaffold :nursery

Turns out singularisation was at fault, ActiveRecord and scaffolding use pluralize and singularize all over the place and “nursery” was not being singularised correctly.

>> "nurseries".singularize
=> "nurseries"

This was being caused by the regex in the rule for correctly serialising “series” to “series” being too loose and catching any word that ended in “series”.

So I spent a little time last night creating a patch for this. I hope it gets accepted but I fear it may not because additions/changes to singularisations have been rejected in the past. I am keeping my fingers crossed though because I have kept this patch very focused and it does fix a rule which is currently too loose.

That brings up an interesting point though, at what point do the inflector functions graduate from being utility functions for use in things like DB table name pluralisation into a full featured linguistic framework? Obviously a full linguistic framework needs to cope with multiple languages and maybe even context and I would agree this is way out of the intended scope of the current inflector. However I do feel that it could be expanded a little without straying into that territory.

For example while investigating this last night I came across a few examples that I feel should go in:

>> "foot".pluralize     => "foots" 
>> "feet".singularize   => "feet" 

>> "tooth".pluralize    => "tooths" 
>> "teeth".singularize  => "teeth" 

>> "goose".pluralize    => "gooses" 
>> "geese".singularize  => "geese"

I feel these, along with “nurseries”, “miseries” and the others that my patch fixes have as much right to be in there as “oxen” -> “ox” and “octopi” -> “octopus” which already have special cases in there. If this patch gets accepted, I might try my luck with this (patch here).

Anyway stick with it Simon, I am sure you won’t find a bug on every subsequent line :)

So I finally got around to finishing off the video from the NWRUG in February, this video is of Rob Lee (http://kodefoo.com, http://monkeyhelper.com) talking about REST in Rails.

I have to say I am still not convinced about REST as a general design principle (although I must say this is not because of Rob’s presentation!) and I just cant see many benefits over how I already write Rails applications, maybe this makes me a Rails luddite and if so, i don’t really care ;)

Now, if I was implementing a specific API I would be all over doing it with REST and I am more than happy to consume other peoples API’s with REST and ActiveResource.

We have Graham Pengelly (http://goingspare.wordpress.com/, http://www.mainlineconsultants.com/) to thank for the fabulous intro/ident that starts the video… it really makes us look way more professional than we are ;)

Finally, just a quick plug for next weeks NWRUG meeting (on the 22nd March), anyone unlucky enough to be going will have to watch me talk about Unobtrusive JavaScript but all is not lost because after that, Paul Robinson will be doing a “Mystery Ruby” session.

Anyway, on with the video from February! Enjoy.

Ever wanted to be able to delegate ActiveRecord fields between objects when using belongs_to relationships?

What do I mean? Well, lets say you have a really simple normalised database for storing people and the companies they work for:

class Person < ActiveRecord::Base
  belongs_to :company
end

class Company < ActiveRecord::Base
  has_many :people
end

Lets say you want to write the persons address out in a view, you would obviously have to traverse the relationship and might end up with code like this:

<%= @person.company.address %>

Not a massive hassle, but you also need to check if the person actually belongs to a company before you do that or you will get a NoMethodError complaining that company is nil.

<%= @person.company.address unless @person.company.nil? %>

This can pretty quickly become tiresome and clutter up those clean and tidy views you have been striving towards. Wouldnt it be nice to cut out some verbosity and be able to do this?

<%= @person.address %>

At first delegate looks like the answer but and sure enough it would allow an address method to delegate to company but it still causes problems if company is nil. Also it gets kinda ugly when you want to delegate a whole load of fields as was the case with the app I was working with today.

Enter AR-Delegation

Plug-in time! Wouldn’t it be nice to do something like this?

class Person < ActiveRecord::Base
  belongs_to :company
  has_columns :from => :company, :except => "id" 
end

and get a fully nil safe set of delegated methods out of it so you can just call the fields like they belong to the model doing the delegation.

Hopefully these other examples will be self explanatory as to what this little plug-in can do:

has_columns :from => :source, :except => "id" 
has_columns :from => :source, :only => [ "title", "name" ], :prefix => "src_" 
has_column :col_name, :from => :source, :as => :new_name
has_column :col_name, :from => :source

It does allow fields to be prefixed which might be useful if you are bringing in fields from multiple other models or would just like to prefix the fields.

Source

I have popped it up on RubyForge so to install it would be:

script/plugin install svn://rubyforge.org/var/svn/ar-delegation

I would be really interested to see if anyone else finds this useful, I will be updating it with tests and stuff as and when I have the time but I would really love to hear some feedback on it.

So tonight I am speaking at GeekUp about this weeks buzzword, OpenID :) It isn’t going to be an in depth talk about all the technical details of OpenID but is going to focus on what OpenID is/does, how to best use OpenID (delegation!) and also I am going to talk about some of the weaknesses of it. It is only going to be a short talk, 10 minutes or so… so even if you are completely unconvinced and I don’t manage to turn you around, at least the torture won’t last long!

Just a quick note too, for those people who noticed the blip in the encoding of the BBC Backstage video, I have re-encoded it and uploaded it to a different video sharing site and replaced the version in that blog post so if you wanted to see the end of the talk, try again :)

See you tonight at GeekUp

Back in September last year at the first BarCamp in the UK, Paul Farnell and Matt Brindley created a super cool little web app that generated a Pecha Kucha slideshow from your last 20 del.icio.us links… it went down pretty well at BarCamp and has been used at several other events I have been to since then, it was really well used at the 2nd BarCamp London a couple of weeks ago which unfortunately neither Paul nor Matt could make it to :( but it was really nice to see it being used anyway.

Unfortunately, it did have some issues, it used iframe tags to load your bookmarked pages and so if any of them had javascript to break out of being viewed in a frame the whole presentation would come to a grinding halt or if the web page didn’t set a background colour you could sometimes see the next page being loaded underneath the current page… messy…

Anyway I have just found out that Matt has upgraded it to use their Site Vista back end to create the presentation so now what it does is it goes and takes a screenshot of your last 20 sites using the screenshotting engine from their product Site Vista and queues the screenshots up for viewing instead of trying to load the actual pages… JavaScript can no longer destroy your Pecha Kucha… Awesome ;)

I have been wanting Matt to change it to do it this way since he first talked about having an API for SiteVista and I am so pleased that it has happened, here is to many more uses of it ;) I hope it gets some wider recognition too. Bring on more Delicious Pecha Kucha, maybe at the April GeekUp?

So a couple of days ago Ian Forrester spoke at a special GeekUp event and I filmed the talk and so after many tedious hours of importing it into iMovie, here is the footage from it…

It came up in the pub afterwards that it might be a good idea to just do the GeekUp events in the pub like Ian’s Geek Dinners and as I noticed a lot of new faces at this event I think it is worth mentioning that we usually DO do this just in the pub! This was a bit of a special case because Ian was in town with a more substantial talk than we usually have at GeekUp but the regular event is mostly just in the B-lounge pub on Piccadilly in Manchester (more details on the GeekUp site) so if you fancy a night in the pub with some fellow geeks then I will see you there on Tuesday ;)

Like I said, there were a lot of new faces at this event so I am going to take this chance to do a completely blatant plug in case any of you are interested in other free events like this, check out the North West Ruby User Group or the Manchester .NET User Group who both meet at the same venue.

See you Tuesday…

So I am just about to leave for the BBC Backstage, New Media is Dead talk at the MDDA in Manchester.

I will be filming it and so that and the NWRUG video of Rob Lee should be up very soon… I have Rob’s talk all imported into iMovie, I just need to put an intro on it.

Not a great first blog post on daveverwer.com, hopefully some will be slightly less inane than this one though :D