pondělí 1. listopadu 2010

Michael Mahemoff: You can't just learn by reading or listening (interview)

Michael Mahemoff works for Google as Chrome Developer Advocate. He has been writing for Ajaxian for many years and wrote Ajax Design Patterns for O'Reilly in 2006. He is the author of useful tools such as ListOfTweets.com and also a funny project, IE6IsOlderThanYourGrandpa.com . He blogs at Softwareas.com and tweets as @mahemoff.

Můžete si přečíst český překlad. There is Czech translation available.

You have been writing for Ajaxian five years already, so you watch this field very closely. What were the most important changes in JavaScript-and-AJAX-World during last five years from your point of view?

There have been plenty of incremental changes, like our understanding of JavaScript getting more sophisticated and the introduction of libraries and tools like jQuery and Firebug to make life easier for developers. We've also benefited from massive performance improvement. But I think the most disruptive change, post-Ajax, is what we've seen recently with the new capabilities of browsers, i.e., HTML5, CSS3, and related technologies such as Geolocation (all of which I collectively refer to as "HTML5" for the sake of convenience). No matter how fluent we became with JavaScript and the like, it was still not possible to do things like video and rich graphics. To achieve the kind of capabilities modern apps required, we had to resort to browser plugins, hacks, and workarounds. Now with HTML5, we have dedicated APIs for many of these capabilities, defined as open standards. Compared to the previous techniques, they are based on open standards and they are typically faster, more secure, more powerful, and easier for developers to work with.


Michael speaking at JSConf 2010 (from Flicker)


You studied Psychology in addition to Software Engineering. How do these subjects go together?

In many ways. Artificial intelligence has always been the obvious intersection and remains a fascinating topic. But over the past couple of decades, User Experience, has come to the fore and emerged from a rather academic niche into a key sub-discipline of modern software development. You see product reviews now which include User-Friendliness as a factor; people expect products to be intuitive. You can only do that with an appreciation of human psychology, which means more than just speculation; psychology is an evidence-based discipline.

You wrote book Ajax Design Pattern. How you got to writing book from blogging and programming?

The book was the product of a blog post on the same topic, where I got inspired to collect ideas after seeing the Ajax term coined. People got excited, it hit a few of the right sites (Delicious Popular, etc.), and O'Reilly approached me about writing a book. I continued blogging excerpts as well as writing the entire book on a wiki. If I write another book, I would probably avoid the wiki approach and concentrate more effort on the blog, or at least a wiki with comments. It's a better way to solicit community feedback as people will rarely make edits to a long article that is mostly one man's voice. (And when they do, it's spam half the time!)

Michael speaking at JSConf 2010 (from Flicker)


You spent some time working with TiddlyWiki. It is a little bit strange project,isn't it? What do you like on it?

Yes, I worked on TiddlyWiki at Osmosoft, an innovation group inside BT which is run by TiddlyWiki creator, Jeremy Ruston, though I actually pored into the code much earlier, when I was writing Ajax Design Patterns. It's definitely different - I like to joke I'm one of the few people in the world who got paid to make web apps that run off a "file" URI. TiddlyWiki, at its heart, is a Single Page App - it contains all HTML, CSS, and JavaScript in a single file. This alone was innovative when it was first created, but what really makes it stand apart is its ability to save to the local hard drive, without any browser extensions and without using HTML5 offline storage APIs. It's possible using ActiveX on IE, native Mozilla APIs on Firefox, and resorting to a second applet file on other browsers, and it means people can easily build persistent web apps and even make "guerilla" multi-user apps just by sticking an HTML file on a share drive. The other cool thing is the plugin system. While TiddlyWiki is by default a personal wiki, you can quickly turn it into a blog, slideshow, or anything else. I recorded a screencast showing how to build a forum in 15 minutes. It also uses TiddlyWeb, so the forum is actually fully hosted on a server, even though its developed on a file URI.


You can meet Michael Mahemoff on HTML5 hackathon, on 15th November in Prague, register please (Google Translate).


You like hackathons. What do you see as the biggest benefit of hackathons?

Loves me some hackathon. As a discipline, software has strengths and weaknesses, and a major strength is that you can produce something completely awesome in a single day. Or of course, many people won't do exactly that, but they'll still build something and learn a lot in the process. So hackathons are a celebration of this amazing thing we do, where we can start the day with a seed of an idea and end up with a concrete manifestation of the idea. It's going with the flow, fitting in with the nature of software development. There's also a social aspect as well, people making connections, working with each other, learning from each other. And it can be as much or as little as the attendees want - some prefer flying solo, and that's fine too. Something else about software is that you can't just learn by reading or listening. We can debate the reasons, but most developers would agree you can only grasp the concepts by rolling up your sleeves and hacking out real code. Hackathons are an ideal environment to do that, a safe place where the output is not critical and there are plenty of opportunities to get feedback and assistance.

What was the longest hackathon you participated in?

At BT, we had an agile hothouse concept where the events went for three intense days.The idea was to pull together all the project stakeholders and put real users at the centre of attention, so we could work with them and ask them to test our emerging work. We could spike ideas and come out with a plan for the next iteration. It wasn't perfect. I felt like we sometimes needed more coders to get the balance right, in an event that was heavily oriented around building real products and presenting them at the end of each day. Some also argued that it's how software should work every day, not just an occasional big bang, but I think in a complex multinational corporation, such "big bang" events are the best thing realistically. I felt the principle was sound and found most of these events to be good value for the company.

So in Osmosoft there were hackathons inside the company? Why should companies try out their own hackathons?

Yes, they were incredibly effective and we ended up running them about one a month, as Osmosoft had a wide range of internal customers and external partners. Fond memories! We'd meet the customer in the morning, usually several of them present. We'd agree on user stories and prioritise them, and then we'd get hacking. Typically an hour or two per sprint, running from 10am and presenting our work at 7pm. Between sprints, a standup meeting to sync up and plan the next sprint. Ideally, we'd be checking into GitHub or similar repo the whole time, and tweeting the event. The ultimate plan was to not only present our work, but also host the resulting web app publicly by end of play (not that it always happened in practice).

There are two important points here. Firstly, we could move very quickly, in a way that often surprised our customer. The reason was our emphasis on reuse - every new project was an opportunity to build not just a new app, but a series of components that could be used on other projects. Not grand enterprisey whitepaper visions, but a 50-line TiddlyWiki plugin that actually does something useful, today. It could be a comments plugin, a graphical editor, or a word counter. Secondly, the openness may be striking in the context of a company building new products. We could do this because we were building a plugin-based framework. Osmosoft, as an open source provider, would make the open source components and an app which is mostly just a composition of those components. BT could then come along and customise it to their own needs, just as any other enterprise (such as our external partners) could pick it up and use it too. Of course, in practice, we were part of BT and we would offer them direct support in customising and deploying the apps. But we showed open source can work for a big company, and we showed the benefits of a hackathon.

You started working for Google in spring this year as a Chrome Developer Advocate. How do you got this job? Was it your idea or have you been asked by Google?

I was invited to apply for the role by someone in Google and I then went through a standard interview process.

Chrome has support for extensions as many other browsers today. But every browser has different implementation. Doesn't make sense to create some cross-browser interface for browser extensions? To make possible for developers to write extension once and run everywhere?

The obvious upside is write once, run many; which is a huge benefit of web development in general. The downside is the risk of stifling innovation too early. While Firefox's extension mechanism has led to wonderful things like Firebug, it's also been difficult for everyday web developers to jump into and left opportunities for improvement, so it was too early to standardise on something like that. Now that Chrome has an extension framework that's easy for web developers to pick up and use, and Mozilla has JetPack, and Safari and Opera also have extension mechanisms, hopefully people will start to draw Venn diagrams and work out where we have common ground. But only if it doesn't stop browsers from continuing to innovate in this area. Chrome has recently introduced the ability for extension commands to appear in the context menu; that's one example of the kind of ongoing improvement that shouldn't be hindered by a standardisation process.

How many GDD do you visit this year as a speaker?

I will speak at the three European GDDs (Munich, Moscow, Prague), in two sessions at each: Google Chrome Extensions and HTML or Native for Mobile Development. The latter will be me alongside Android advocate Reto Meier, so it should be a lot of fun.

Have you ever been to Prague or is it your first visit? What are you most looking for?

I've not been to Prague, but looking forward to it. The best thing for me will be the pre-GDD GTUG event. Czech Republic clearly has a strong web development community, so I'm looking forward to meeting the local developers and seeing what they come up with.

Martin Hassman was interviewing Michael Mahemoff before his visit to Prague.

If you want to meet Michael at hackathon on 15th November, please register.

3 komentáře:

Poznámka: Komentáře mohou přidávat pouze členové tohoto blogu.