Herding Code

Latest Episodes:

Herding Code 116: Eric Lawrence on Fiddler, IE Internals, and HTTP Jul 01, 2011

Show notes

Episode 116

This episode of Herding Code the guys talk to Eric Lawrence, the author of the popular Fiddler web debugging proxy. Eric's also a member of the Internet Explorer team and developer of several popular freeware tools.

Eric explains how he's been working on - and now runs - the team that works on the networking components for Internet Explorer.
Kevin asks Eric to clarify what portions of IE he works on. Eric explains that he's on fundamentals, which includes things like networking and security, not rendering or the DOM.
Jon asks Eric about his development focus. Eric says that his main focus is on C#.
Jon references an interesting bit of Eric's MIX talk about Fiddler - Fiddler is an HTTP proxy, so it works with all browsers and devices.
Scott mentions that he's used Fiddler for low-level network debugging. Eric talks about the broad range of Fiddler users.
Eric mentions that Fiddler's used for security testing, and there's a discussion of fuzz testing. Eric describes "dumb fuzzing" and "smart fuzzing".
Jon asks how Fiddler is used with mobile devices.
Jon asks Eric if he's taking advantage of any "internal" info or API's as a member of the IE team.
Jon talks about how the plugin system has really paid off for Fiddler over the years. Eric talks about how he's supported both a Javascript and a reflection based .NET plugin system.
Eric mentions how he's tested a "pure .NET 4" version of Fiddler, and talks about the Fiddler itself runs on Fiddler Core, which is really close to running on the .NET client profile.
Jon asks Eric about some of the interesting things he's heard built on top of Fiddler Core. Eric mentions some testing extensions, ELMAH use, and FiddlerCap.
Jon asks what IE9 features Eric is taking advantage of. Eric talks about IE9's support for the X-Download-Initiator header, which allows tracing why a resource was requested.
Eric talks about Fiddler now proxies Cassini traffic for ASP.NET developers, and Jon mentions how he saw from internal bug reports that Eric was driving ASP.NET debugging issues with pre-release versions of IE9.
Eric mentions that the issue with IE9 betas and Cassini was due to IPv6, and since the podcast was recorded on IPv6 day, the conversation shifts over to a discussion of IPv6.
Jon asks Eric about IPv6 use in the real world, and Scott K asks about issues with SSL over IPv6.
Jon asks about IE's and Fiddler's support for HTTP verbs beyond GET and POST. Eric talks about how he had to drop some strict enforcement of protocols for non-standard verbs since actual usage often didn't follow the specs.
Jon asks Eric about the 100 Continue response. Eric explains why it's there, and how he handles it in Fiddler, and how IE handles it.
Jon asks about the Accept header, asking Eric's opinions on its use and how IE9 handles it. Eric explains how he doesn't think Accept really works, because proxies and servers don't correctly handle them.
Scott K asks about the advantages of being able to analyze aggregate HTTP traffic for a large organization like Microsoft. Eric explains that there's not really a lot of opportunity for IE, but he does get advanced notice on Fiddler issues from internal Microsoft use against pre-released software.
Kevin asks about the "Fiddler has detected a protocol violation" error. Eric explains that it's helpful in debugging issues which browsers will attempt to hide due to being very liberal and forgiving with respect to protocol violations.
Eric explains that there was an HTTP 0.9 version which didn't have HTTP headers, and talks about how his awareness of protocol violations on major websites came in handy. He mentions that he's thought about an HTTP-Lint module, which would be a lot more strict with respect to protocol violations.
Question from Twitter - James Schmidt - "Will we see Fiddler features move over to IE Dev tools?
Jon asks about the common import / export format that IE F12 dev tools and Fiddler share... kind of.
Question from Twitter - Luke Foust - Hear about developing a side project inside Microsoft.
Jon asks Eric about some of his other freeware applications, including SlickRun and a popup blocker (big in Brazil!).
One of Eric's freeware applications was a utility to tweak the number of simultaneous downloads IE would use, which prompts Jon to ask about how simultaneous browser connections have evolved over the years.
Question Jarrod Dixon - "Possible to open source SlickRun? I use it a lot and would like to add some features?"
Kevin asks if Eric would consider open sourcing Fiddler at some point.
Eric wraps up with a description of what's in the works for Fiddler. Kevin asks for auto-update.

Show Links:

Eric's IE Internals blog
@ericlaw on Twitter
Fiddler (Web Debugging Proxy) website
Fiddler blog
The Devil Went Down to HTTP: Debugging with Fiddler (MIX11)
MIX11 release notes for Fiddler
Post: Fiddler is better with Internet Explorer 9
IE and the Accept Header
IE9 Standards Mode Accepts only text/css for stylesheets (HTTP 406 post Eric mentioned)
Some of Eric's freeware at Bayden Systems
SlickRun
Producing Open Source Software (book)

Download / Listen:

Herding Code 116: Eric Lawrence on Fiddler, IE Internals, and HTTP

Herding Code 115: RESTravaganza with Darrel Miller, Glenn Block, and John Sheehan Jun 22, 2011

Show notes

Episode 115

This episode of Herding Code the guys talk REST with Glenn Block (who's driving the WCF Web APIs), Darrel Miller (a REST expert with a lot of real world production experience), and John Sheehan (author of RestSharp) about what REST really is and what practical value it really offers in real world, production applications.

Jon asks Glenn for a quick overview of how WCF Web API fits in with REST.
Darrel talks about how he got into REST in support of desktop systems.
John asks Darrel about how HttpListener is working for him.
K. Scott asks Darrel why he's doing all the work to plumb RESTful services rather than just going with something like SOAP.
Glenn asks Darrel how his RESTful services are more "evolvable" than previous technologies he's used.
Jon asks Darrel what REST means to him. Darrel says it means that there are just two things that the client and server couple on: media types and link relations.
John says that very few people he interviews describe REST as how Darrel just did. Glenn talks about how his understanding of REST evolved when he more closely studied Roy Fielding's original dissertation.
Question from Twitter (@stevenproctor - Steven Proctor): "Do nice http paths really make an architecture RESTful? Wasn't there something about next available commands too?" Glenn and Darrel talk about how this is the fundamental concept of hypermedia.
Jon ignites a firestorm by asking why people who care about REST dislike how OData is implemented. Chaos ensues.
Darrel explains how OData's format doesn't match with some important RESTful principles like link relations and metadata discoverability.
Glenn points out that OData is an API that takes a constrained view of of HTTP, which offers a tradeoff which many developers find beneficial.
Scott K asks the guests how many non-demo OData feeds are actually available.
Scott K asks why not just use JSON instead of OData, and Glenn explains how the important difference is around metadata - JSON is just untyped data.
Darrel talks about the concept of serendipitous reuse, and how common media types offer better reuse than untyped JSON data.
John asks if anyone is actually putting the client re-use case into practice in the real world, and Darrel plugs the REST Fest. Glenn points out that just having multiple versions of a client working against a spec is a significant advantage, talking about HTTP 1.0 and HTTP 1.1 as an example.
Jon asks if a microformat approach could work, and eventually an RDF discussion breaks out.
Glenn talks about how many people view REST by mapping HTTP verbs to CRUD (create, read, update, delete) operations, and they're missing the importance of linking and hypermedia.
Kevin asks where people "lose the path" with REST, and what are the most important concepts to stick with. Darrel comes back to the hypermedia constraint as the most important concept that's commonly missed.
Glenn points out that your architecture is up to you, and you don't have to follow RESTful principles, but there's a problem if you don't and claim your API is RESTful when it isn't.
John points out that there are few examples of RESTful systems really paying off in practice. Glenn and Darrel point to Jon Moore's Oredev talk about real business value to Comcast, and Jon mentions Glenn Block's MIX presentation about device support based on content negotiation.
Glenn explains that opportunities are emerging as we're moving beyond the browser, and he sees a lot of opportunity for WCF Web API's to shine here.
Darrel describes another example of how a RESTful API could guide a common user experience across platforms, using Twitter as an example.
Kevin asks for public examples of good RESTful API's. Darrel mentions Sun's cloud API and SteamCannon; Glenn says that ATOM PUB is the best public example.
K. Scott asks about building RESTful clients.
Kevin asks about the practicalities of clients navigating hypermedia.
Question from Twitter (@kellabyte - Kelly Sommers): "I'm curious how REST might fit with an application that is wanting to store events and event sourcing. Is REST common for this?"
Question from Twitter (@gsogol - Jeff Sogolov): "How about Rest in the enterprise? Securing Rest services with Saml or oAuth? Also impersonation."
Jon mentions how WS-* defined methods for securing portions of message for different access and asks if REST handles that kind of scenario. Scott K, Glenn, and Darrel discuss.
The show wraps up with a discussion selecting architectural styles based on concrete benefits.
Darrel pimps REST Fest 2011, August 18 - 20.
John mentions his upcoming talk at DevLink on August 17, and Monospace July 23-25.
Glenn mentions the Portland Code Camp and //build/.

Show Links:

Glenn Block (MSDN blog, CodeBetter blog, @gblock)
John Sheehan (blog, @johnsheehan)
Darrel Miller (blog, @darrel_miller)
http://wcf.codeplex.com
http://restsharp.org/
Roy Fielding's dissertation on REST
http://microformats.org/
ALPS http://amundsen.com/hypermedia/profiles/
RDF (Resource Description Format)
Jon Moore's talk at Oredev 2010: Hypermedia APIs
Sun Cloud API
SteamCannon
REST Fest 2011
DevLink
Monospace
BUILD conference
http://www.odata.org/producers

Download / Listen:

Herding Code 115: RESTravaganza with Darrel Miller, Glenn Block, and John Sheehan

Herding Code 114: Trevor Burnham on CoffeeScript Jun 14, 2011

Show notes

Episode 114

This episode of Herding Code the guys talk to Trevor Burnham about Coffeescript, "a little language that compiles into JavaScript."

Kevin asks Trevor to explain what CoffeeScript is. Trevor explains how CoffeeScript helps you to write the same code you would have in JavaScript, but more quickly and with less effort.
Kevin ask if people like CoffeeScript because they hate Javascript. Trevor talks about the strong reactions people had to the addition of CoffeeScript support for Rails, and how CoffeeScript is not a dumbed down Javascript, it's really just a cleaner syntax for exposing the functional power in the that underlies Javascript.
Scott K talks about the time and political pressures that shaped Javascript, in some ways hiding a powerful language behind a mandated Java-like syntax.
Kevin asks about the language inspirations behind CoffeeScript. Trevor talks about how it's drawn inspiration from Ruby, Python, Haskell, and Erlang.
Jon asks about how CoffeeScript support is technically added to Rails - where is the compilation happening? Trevor explains how the coffee-script and execjs gems select the Javascript runtime.
Jon asks a question from Twitter by @darrencauthon about using CoffeeScript on Windows and .NET.
Scott K asks about other languages that compile down to Javascript, and what will happen when Javascript as a language. Trevor talks about Objective-J and Traceur.
Kevin asks about how you can debug CoffeeScript code.
Jon talks about the Try CoffeeScript interactive compiler on the CoffeeScript.org site and asks resources for learning more about CoffeeScript.
Scott K asks if CoffeeScript leverages more advanced Javascript features if they're available, and if there are language facilities for extending the language. Trevor talks about how CoffeeScript is just Javascript, so you can leverage the extensibility features that are already available in Javascript.
Kevin drops back to asking some basics: why is CoffeeScript so special? What's with the new function syntax?
Jon talks about how nice it is to just delete parentheses, braces, and semicolons. Trevor talks about how in general CoffeeScript code is 2/3 as verbose as the equivalent Javascript code.
K Scott asks about how CoffeeScript simplifies scope issues, and Trevor talks about how CoffeeScript is very opinionated about scoping.
Scott K asks about type coercion. Trevor mentions the wtfjs.com site, talking about how Javascript's type coercion can be surprising at times.
Scott K asks about how DOM manipulation is handled. Trever explains that CoffeeScript works great with jQuery, and any other Javascript library available.
Jon asks if it's possible to do CoffeeScript compilation in the browser, and asks if that's possible / practical for standard applications.
Jon mentions the SassAndCoffee project for .NET.
Jon asks if there are any Javascript to CoffeeScript converters. Trevor says he doubts there will ever be a good one, but after the show sent Jon a link to js2coffee (link below).
Scott K asks about JSLint support. Trevor talks about how CoffeeScript works with JavaScript Lint.
Kevin asks if CoffeeScript has been used as a DSL, and Trevor mentions CoffeeKup.
Jon asks what happens if CoffeeScript hits a compiler error.
Kevin asks about split between front-end vs. back-end use for CoffeeScript.
Jon some a question from Christopher Deutsch (@cdeutsch) about how to sell CoffeeScript to a team - is this today's flavor, tomorrow's legacy headache?
Kevin talks about how he likes the => function, and Jon mentions how he likes the @ operator as well.
Kevin asks about the object orientation features in CoffeeScript, and Trevor mentions how the class keyword is used.
Jon says the he sees a lot of similarity between CoffeeScript and SASS. Trevor talks about how both reduce repetitions.
Jon asks if jQuery could take advantage of CoffeeScript.
Trevor mentions how you can buy his book and mentions a recent article he published in PragPub.
Trevor's upcoming talks: O'Reilly Web 2.0 Expo in NYC in October and Oredev in Sweden in November.

Show Links:

Trevor Burnham (Site, @coffeescript, @trevorburnham)
CoffeeScript site
Hacker News: Rails 3.1 shipping with CoffeeScript
StackOverflow: CoffeeScript on Windows
StackOverflow: How can I compile CoffeeScript from .NET?
Objective-J
Traceur
GWT (Google Web Toolkit)
The CoffeeScript Wiki
List of languages that compile to JS
RailsCast: CoffeeScript Basics
PeepCode: Meet CoffeeScript
Trevor's book: CoffeeScript: Accelerated JavaScript Development (Amazon, direct from PragProg)
wftjs.com
js2coffee (Javascript to CoffeeScript compiler)
MiddleMan.rb - Middleman is a static site renderer that provides all the conveniences of a modern web stack, like Ruby on Rails, while remaining focused on building the fastest, most-professional sites possible
SassAndCoffee: project, NuGet
JavaScript Lint
CoffeeKup
Sass
PragPub: A CoffeeScript Intervention
O'Reilly Web 2.0 Expo in NYC in October
Oredev in Sweden in November

Download / Listen: Herding Code 114: Trevor Burnham on CoffeeScript

Herding Code 112: Josh Arnold and Jeremy Miller on FubuMVC May 24, 2011

Show notes

Episode 112

This episode of Herding Code the guys talk to Josh Arnold and Jeremy Miller about what's new with FubuMVC.

Jeremy Miller explains why FubuMVC "deserves to exist" and explains how compositional architecture and conventions help in building complex systems.
Josh talks about how FubuMVC diagnostics help in understanding how the conventions are being applied how FubuMVC is working.
Jon asks about how behavior chains work, and how they relate to routes. Jeremy and Josh explain how behaviors work and how they allow you to extend policies and conventions when you need to.
Jon asks how routes work, and Jeremy explains how they can be configured at a few different levels.
Jeremy talks about how FubuMVC is built to leverage static features in .NET through strong typing and leveraging the type system as much as is possible.
Josh and Jeremy talk about the advanced diagnostics which have recently been added to FubuMVC.
Jon asks how FubuMVC diagnostics compare to Glimpse.
Jeremy talks about the new packaging system, and how it can be used to apply complex and extensive changes just by dropping them into your application.
Scott K asks how the new packaging system relates to NuGet and OpenWrap, and Jeremy explains how the two are complimentary.
Jeremy and Josh talk about how their complex requirements in their active projects have driven FubuMVC's features.
There's a discussion of view engines - what's supported, what they're currently using in their projects.
Jeremy talks about how FubuMVC uses HTML conventions, how HtmlTags work, and how you can use jQuery-like chaining to reuse conventions.
Jeremy talks about how authorization works with the behavior chains. Scott K asks if this can be applied at the action level rather than at the UI level, and Jeremy explains the endpoint service.
We wrap up with a mention of Pablo's Fiesta, this Sept 30 - Oct 2 in Austin, TX.

Show Links:

Download / Listen:

Herding Code 112: Josh Arnold and Jeremy Miller on FubuMVC

Herding Code 111: John Papa on the Open Source Fest at MIX11 May 04, 2011

Show notes

Episode 111

This episode of Herding Code the guys talk to John Papa about the Open Source Fest he put together at MIX11.

Jon asks how the whole thing got started, and if John encountered any friction within Microsoft in getting this set up.
John describes the event and calls out some of the winners from the event.
There's a discussion of the Glimpse project. Scott asks what it is, and Jon tries to give the sales pitch for it.
John talks about how many of these really cool project are hampered by marketing mistakes like poor project pages and unmemorable project names.
John mentions some of the areas for improvement - less background noise, bigger space. Some of that was due to overwhelming response - stopped counting at 500 attendees, ran out of food 3 times, etc.
Scott asks if a next step should be an open source conference for .NET. Jon mentions that there are some benefits to piggybacking with a "real" conference so the bosses will pay for us to go.
Scott asks if there's any point to having sessions at a conference, since the real value at the conferences is in the networking and conversation. There's a discussion about how an open space is cool, but something of this scale isn't likely to self-organize.
Scott talks about how the ALT.NET Seattle event in Seattle is including open source hacking, proposing that larger conferences do this as well.
John mentions the Twitter list he's created for all Open Source Fest participants.
We take a question from Tony Champion, asking what John would do differently in future events.
John and Jon discuss the difference between consuming and participating in a conference.
John pimps the Silverlight MIXer event he runs at MIX.
Jon asks if there should be venture capital folks at future open source fests. John said said that it was important to keep clear of any ulterior motives at this first event, but it's possible that may happen in the future.
John and Jon talk about the difference between "official" events and sponsorship driven events.

Show Links:

Download / Listen:

Herding Code 11: John Papa on the Open Source Fest at MIX11

Herding Code 110: Geoff Dalgas and Jarrod Dixon take us behind the scenes at StackExchange Apr 25, 2011

Show notes

Episode 110

This episode of Herding Code Kevin and Jon sit down with Geoff and Jarrod at MIX to talk about their experiences from helping to build the first StackOverflow site up through today's fast paced world of StackExchanges and gold plated Lamborghinis.

Note: We recorded in quietest spot we could find - there's some background noise, but it's worth it.

Geoff and Jarrod talk about their past job experiences, including building (gasp) 911 software with Visual Basic 6.
Jon asks how things worked at the beginning - did people work on separate areas? Was there a plan?
How do you share data when developing with a remote team? Geoff talks about how they started with the Database Project type, but moved to SQL scripts, ending with a migration tool. Jon gets to say "idempotent".
Geoff and Jarrod talk about how they've moved form Subversion to Mercurial.
Jon asks how code moves from local development to a production server. Geoff talks about the build and deployment process.
Kevin asks how much the process has changed over the years.
Geoff talks about how features first hit Meta, then the "others" tier (everything but StackOverflow), then to StackOverflow.
We talk about how they're using Redis, including the newly open sourced redis-sharp library, and some of the tricks that are used to keep the cache performant, including gzipping cached data.
We talk about how they do performance tuning, how costly queries are tracked, etc.
Geoff talks about how search was moved from SQL Server Full Text indexing to Lucene.NET in order to move the load off the database server.
Jon asks about how they've integrated the ASP.NET stack with open source front-end and utility software.
Jon asks how the the IT and collaboration works with the growing distributed team.
Jon asks how bugs are tracked, and the talk about how they use the Meta sites for tracking issues. Jon is sold on the concept and asks if he can install a local instance of Meta for his own bug tracking concepts, but the guys just laugh at him.
Jon asks for more specific on the performance monitoring systems the team uses, and Geoff gives him the rundown.
Jon asks how things have changed as the team has grown and buckets of funding money keep rolling in.
Geoff explains how they handle multi-tenancy, now that they're hosting lots of sites. Jon is amazed to hear that it's just one big application which switches data structures and display based on the url.
Geoff mentions that they're using Less to keep the CSS sane.
Kevin asks how much work is involved in spinning up a new site. Geoff explains the tool they've got set up to generate the scripts to add a new StackExchange site.
Jon asks about MVC 3 is working for them. They like the Razor.
Jon asks what annoys them most about ASP.NET MVC. After some thought, they lament that the routes are defined separately from the actions, but mostly it's just a lot of love for MVC.

Show Links:

Geoff Dalgas @superdalgas
Jarrod Dixon @jarrod_dixon
Redis
BookSleeve (Redis client library)
Lucene.NET
HAProxy
Skype
Nagios
Cacti
Dapper
Sam Saffron @samsaffron
Benjamin Dumke @balpha
less.js

Download / Listen:

Herding Code 110: Geoff Dalgas and Jarrod Dixon take us behind the scenes at StackExchange

Herding Code 109: Harmony Hackathon Apr 20, 2011

Show notes

Episode 109

This episode of Herding Code the guys talk to the organizers of the Harmony Hackathon: twelve developers coding madly for 48 hours, trying to build an application for the non-profit Harmony Hill cancer retreat center.

Eric talks about the Harmony Hackathon came together and what they were trying to accomplis.
Jon asks about how things were coordinated. The guys talk about how they managed the "organized chaos" of throwing a lot of developers in a room for 48 hours.
Scott K asks how a group of alpha geeks got past selecting an IoC container in a weekend, let alone write some code.
Jon asks for a rundown on the dependencies (Fluent NHibernate, Fluent Migrator, ASP.NET MVC 3, and some testing tools).
Hooray for build servers!
Discussion of machine.specifications and NUnit in order to test both high and low level code.
Scott K asks how much actually got done.
We talk to Vic to get the real scoop - sure the tech was cool, but did they build anything? How was it like to be the project owner with this kind of project?
Discussion of extremely short sprints - down to 30 minutes at the end! How do you handle Pomodoro length sprints?
We talk about Chewie, which adds some Bundler features to NuGet.
Scott K gets the rundown on psake (build automation in PowerShell) from Eric.
Scott K asks about what kind of interaction patterns they saw emerging.
Scott K asks about the future plans for this project.
Scott K asks why ALT.NET people would want to do something nice, and Jon mentions his mild surprise at the happy, friendly vibe he'd seen at the ALT.NET Seattle events he's attended. There's a discussion of the general spirit of the ALT.NET Seattle group.
Jon asks for lessons learned and recommendations for other people who want to put on a similar event.
Scott K asks about friction points with the tools the team selected.
Razor was one of the new things that none of the developers had used before. Jon remarks that learning Razor was kind of a non-event for him, and the guys agree.
Kevin asks about how the project lifecycle looks like when compressed to a 48 hour period.
Eric explains what Harmony Hill does, explaining why everyone was so motivated to help out. The guys talk about how it's nice to go beyond talking about beautiful code and actually doing something.
Scott K asks if there were technological things the team would do differently in the future.
The Open Source Track at ALT.NET Seattle (May 5-8) is mentioned.
Coding alone? You must wear the Coding Solo Cowboy Hat of Shame.
Pimp your stuff time: Harmony Hill, ALT.NET 2011, Giles, Simple.Data
The guys ask Scott K when we'll see the MVC generator thing he's been working on.
Everyone loves Simple.Data.
Developers, developers, developers!
Woof, woof, woof!
Does anyone read these notes? I don't think you do.