Aaron Patterson: it won't be easy to support HTTP 2

According to the developer of Rails, the community can improve the HTTP work in Ruby web servers, and improving the streaming performance can be used as a launch pad for HTTP 2
Sergey Siminskiy's picture
Sergey Siminskiy
08 June

Aaron Patterson

Aaron Patterson

Member of Ruby and Rails CoreTeam, top 1 Ruby contributor, Ruby Hero 2010. Speaker at RailsClub 2014.


Biggest Russian Ruby on Rails event

At RailsClub 2014 he had a speech about the future of Rack.

We had a chance to ask Aaron few questions after the speech.

What are you working on right now? 

Right now I'm working on a few things:

  • Speeding up view generation (including routing) in Rails;
  •  Speeding up tests in ActiveRecord by investigating parallel execution;
  • Working on Rack 2.0. 

What is the best and worst part of your job?

The best part is helping other developers, and of course deleting code. I think my favorite thing to do is delete code! The worst part is trying to figure out why some code is the way that it is. Working on Rails is sometimes hard because the codebase is old in some places, so it's difficult to track down where some code came from, or why a particular bit of code is the way it is.

What’s your main achievements at the moment?

I think my greatest achievement so far is the performance improvements I did on ActiveRecord (called AdequateRecord). It took me years to refactor the code to get to the point I could make this type of performance improvement. It feels good to work hard on something for so long and to see it actually ship. Also being able to make performance improvements without changing any user facing APIs makes me very happy.

On your opinion, how will Ruby and Ruby on Rails develop in the future? 

For Rails, I hope that we can improve the HTTP story in Ruby webservers. I want streaming responses to be a better supported, and use that as a launching pad for HTTP 2 support. As far as Ruby is concerned, I am excited to see improvements to Ruby's garbage collector. I think Ruby's long term future will be to add more concurrency constructs, maybe the actor pattern built in.

What’s the main problem which Ruby society faces at the moment?

On the far horizon for the Rails community, I think supporting HTTP 2 is very important and will be quite a challenge. We have lots of software that depends on HTTP 1.1, and I think it will be a challenge to upgrade.

What’s your favorite gem? 

I think minitest is a good example. It's small and easy to understand. But I have to say that no gem is perfect. Everything has some problems, you just might not see it yet!

What rules would you propose to establish for the successful development of the Ruby project with 250000 lines of code by many developers with very different skill levels?

I guess the best rule I could give would be to have good communication among the team. Other than that, stick to the SOLID design principles. I think they are good rules for new people to follow, and as you gain experience, you'll begin to understand when it's OK to break those rules.

Is there any chance that MRI will get rid of GIL in the nearest future?

Probably not in the near future, but I do think it will happen.

Do you read any Ruby\Rails blog? 

I usually read the ruby subreddit. Also I like listening to the Giant Robots podcast, but they don't really seem to talk about Giant Robots very often.

What do you like to do when not coding? 

Mostly cooking and playing video games. Mostly cooking though. I am curing my own bacon right now!

RailsClub conference on which we managed to communicate with Aaron will take place this year in Moscow 23th of September.

Get your ticket here.

A. Davydov: It's vital not to stop and find conceptually new solutions

An Open Source enthusiast and one of the core developers of the Ruby-framework Hanami shares ideas about Hanami, Ruby and the developer's life
Diana Ugay's picture
Diana Ugay
10 August

Anton Davydov
Anton Davydov

An Open Source enthusiast and one of the core developers of the Ruby-framework Hanami.

On the eve of RailsClub 2017, on which Anton will be one of the speaks, we questioned him about the job and his views on Ruby development.

What are you working on right now?

I am working for a Healthcare start-up, which can improve the lives of so many people in the US. Unfortunately, I can't tell you more about it as I have signed the NDA. Talking about Open Source projects, now we are working on the version of Hanami 1.1. Besides, Sergey and I create a library for Event Sourcing in Hanami. It already works without a global state as well as does subscribe & broadcast for events, supports multiple adapters and allows you to easily add your own ones. The library is to have much more cool features in the future, such as retries, the construction of the event tree and eventbox documentation.

Isn't there something similar in Ruby with the Wisper gem...

Yes, it is very similar, but Wisper implements pub/sub only in the memory of one instance, while we want to allow the developer to choose how to process such messages: in one instance or to scale a few.

What about new adapters? How will they be added?

You just have to resolve them into a container. For now, we have in-memory and Redis adapters which you can play with. Kafka, RabbitMQ and, probably, PostgreSQL will show up in the future. 

What is your view on the development of Hanami in the next four years?

As now there's definitely a monopoly of Rails, I would like to see a worthy alternative in Hanami. This will allow all frameworks to develop. Thus, Hanami already solves business problems and many developers are inspired by this tool. I hope that in the future we will manage to view on Ruby as a dying language and popularize new ideas.

Do you think Hanami can become a mainstream in the future?

I've noticed some significant changes in the attitude towards the framework and its ecosystem over the last year. To measure it you have to take a look at the gems downloading scale, the frequency of the posts publishing, the conferences'references. A year ago, when people heard about Hanami, they would circle a first finger by a temple, whereas now they come up and tell us that they enjoy the framework, and they are already using it.

Why do you think it was Rails that became the standard in Ruby development?

To my mind, there are several reasons for this. Firstly, everyone was tired of verbose Java- and PHP-frameworks, so there was a need to create a simple and operative solution. Secondly, the framework allows you to quickly launch products, the creation of which in other languages would take weeks and even months. For many of the developers it was like a breath of fresh air and Rails has quickly gained popularity.
Now, there is no other MVC-framework of full value except for Rails. There are Sinatra and other http-frameworks out there, but they do not provide the developers with the opportunity to start working on business tasks with just one command.

In your opinion, what are the problems the Ruby community is facing right now?

Well, a lot of my friends are saying or have said that they are too bored with Ruby. They do the same thing day after day, which, consequently, kills the motivation, that's why a lot of them begin to study other languages ​​and technologies. People lack something radically new, some ideas that will allow them to look at their work from the different perspective.
It is vital not to stop and find conceptually new solutions, which nowadays the guys from dry/rom are busy with. Unfortunately, any new approach - not only in Ruby - is almost always perceived with hostility. On the one hand, some criticality should definitely take place, however, it usually kills the motivation to move on and grow as a community.

What is the path of the Ruby developer for you in general?

There's no single answer for this question, I guess. In my opinion, a good developer is interested in the capabilities of different languages and systems. If so, the specialist himself is to decide what is good for him and what is not. Thus, they would be able to choose really useful approaches to be used in work in the future.

Which gem, in your opinion, can serve as an example of an ideal code?

I certainly would not point at Ruby core (laughs). it's  quite hard to talk about such ephemeral concepts as the quality and the beauty of the code as I myself can't say that I write a good, so I do not always like it. To me, the best gems are those that simply solve existing problems and do not create new ones.

How do you learn something new about Ruby? Do you read blogs or use other sources?

I like link aggregators, e.g. Reddit. I also use RSS, which allows you to get acquainted with different points of view on the same problem.
For example, an article appeared recently about why Hanami is bad. There was, in fact, the only one point: the call method is magically called in the framework. The majority of the community was trying hard to explain that this was normal: excessive explicitness hurts sometimes. It's very interesting to read such things as you start to look at things from a different angle.
I also have my own Telegram channel, and the followers often send me some interesting links.

You were doing a platform for a collaboration on the Open Source-projects of OSS Board, weren't you. Are you still working on this project?

That's correct, I continue working on it right now. It seems to me that I failed to promote the OSS Board properly. That's very difficult to find people who will create tasks there and do them, especially when the task is slightly more difficult than making changes to the form. The main problem of the project is the search for specialists who will be able to add a well-described task as you need to monitor it, update it and communicate with the developers, which is actually is a very great work. However, there are still people who help, which I am very grateful for.

What do you consider your greatest achievement in the career?

I received more than 80 comments with questions about each line on one of the pull request in Rails. I never finished it, but as a result I started take such things easier.
On top of that, I finished the project in Google Summer of Code, while a lot of people give up in the first month, even more - on the second, and only a few gets to the end.

What kind of project was it?

This was a plugin for Sidekiq, which shows the statistics for all the tasks. Unfortunately, I've abandoned it due to the lack of time, although it is a fairly popular library.

What are your main achievements in life?

To me, it is that I did not shy away from speaking in English for the first time in my life at the conference EuRuKo-2016 in Sofia in front of the audience of 700 people. It is way easier to make a report in Russian as this is your mother tongue. You can imagine that it is very anxiously when you do not know English well and do not have the experience of such performances.

In your opinion, who from the Open Source developers can be considered an example to follow?

It's definitely worth it to look at what other developers are doing. They generate and implement interesting ideas that you can help to develop or just take a good note. However, first of all you need to focus on yourself and not compare yourself with others.

What are you going to talk about at the conference and why is it worth listening to?

I am to talk about the experience. I've been working not only on the framework, but also on its ecosystem for the last year and a half. To some extent I can be called a developer advocate of the Hanami framework. Thus, I will talk about why this work is needed, what mistakes I've made along the way and how to take this experience and apply it to another framework or technology.