Recently, I (Rachelle Palmer, Senior Product Manager for MongoDB Drivers) sat down with Kaitlin Mahar, Lead Engineer for the Swift and Rust drivers to discuss some of the exciting developments in the Swift space. This article details that chat.
Swift is a well documented, easy to use and convenient language focused on iOS app development. As one of the top ten languages, it's more popular than Ruby, Go, or Rust, but keeps a fairly low profile - it's the underestimated backbone of millions of applications, from Airbnb to LinkedIn. With its simple syntax and strong performance profile, Swift is versatile enough to be used for many use cases and applications, and we've watched with great interest as the number of customers using Swift with MongoDB has grown.
Swift can also be used for more than mobile, and we've seen a growing number of developers worldwide use Swift for backend development - software engineers can easily extend their skills with this concise, open source language. Kaitlin Mahar, and I decided we'd like to share more about MongoDB's commitment and involvement with the Swift community and how that influences some of the initiatives on our Swift driver roadmap.
Rachelle (RP): I want to get right to the big announcement! Congratulations on joining the Swift Server Working Group (SSWG). What is the SSWG and what are some of the things that the group is thinking about right now?
Kaitlin (KM): The SSWG is a steering team focused on promoting the use of Swift on the server. Joining the SSWG is an honor and a privilege for me personally - through my work on the driver and attendance at conferences like Serverside.swift, I've become increasingly involved in the community over the last couple of years and excited about the huge potential I see for Swift on the server, and being a part of the group is a great opportunity to get more deeply involved in this area. There are representatives in the group from Apple, Vapor (a popular Swift web framework), and Amazon. The group right now is primarily focused on guiding the development of a robust ecosystem of libraries and tools for server-side Swift. We run an incubation process for such projects, focused on providing overall technical direction, ensuring compatibility between libraries, and promoting best practices.
To that end, one thing we're thinking about right now is connection pooling. The ability to pool connections is very important for a number of server-side use cases, and right now developers who need a pool have to implement one from scratch. A generalized library would make it far easier to, for example, write a new database driver in Swift. Many SSWG members as well as the community at large are interested in such a project and I'm very excited to see where it goes.
A number of other foundational libraries and tools are being worked on by the community as well, and we've been spending a lot of time thinking about and discussing those: for example, standardized APIs to support tracing, and a new library called Swift Service Lifecycle which helps server applications manage their startup and shutdown sequences.
RP: When we talk with customers about using Swift for backend development, asking how they made that choice, it seems like the answers are fairly straightforward: with limited time and limited resources, it was the fastest way to get a web app running with a team of iOS developers. Do you feel like Swift is compelling to learn if you aren't an iOS developer though? Like, as a first language instead of Python?
KM: Absolutely! My first language was Python, and I see a lot of things I love about Python in Swift: it's succinct and expressive, and it's easy to quickly pick up on the basics. At the same time, Swift has a really powerful and strict type system similar to what you might have used in compiled languages like Java before, which makes it far harder to introduce bugs in your code, and forces you to address edge cases (for example, null values) up front. People often say that Swift borrows the best parts of a number of other languages, and I agree with that. I think it is a great choice whether it is your first language or fifth language, regardless of if you're interested in iOS development or not.
RP: Unquestionably, I think there's a great match here - we have MongoDB which is really easy and quick to get started with, and you have Swift which is a major win for developer productivity.
RP: What's one of your favorite Swift features?
KM: Enums with associated values are definitely up there for me. We use these in the driver a lot. They provide a very succinct way to express that particular values are present under certain conditions. For example, MongoDB allows users to specify either a string or a document as a "hint" about what index to use when executing a query. Our API clearly communicates these choices to users by defining our IndexHint type like this:
public enum IndexHint {
/// Specifies an index to use by its name.
case indexName(String)
/// Specifies an index to use by a specification `BSONDocument` containing the index key(s).
case indexSpec(BSONDocument)
}
This requires the user to explicitly specify which version of a hint they want to use, and requires that they provide a value of the correct corresponding type along with it.
(Note: mine is the MemoryLayout type. Being able to see the memory footprint of a class that you've defined is really neat).
Over the next six months, your top priority is rewriting our driver to be purely in Swift. For everyone who is wondering, why wasn't our official Swift driver "all Swift" initially? Why the change?
KM: We initially chose to wrap libmongoc as it provided a solid, reliable core and allowed us to deliver a great experience at the API level to the community sooner. The downside of that was of course, for every feature we want to do, the C driver had to implement it first sometimes this slowed down our release cadence. We also feel that writing driver internals in pure Swift will enhance performance, and give better memory safety - for example, we won't have to spend as much time thinking about properly freeing memory when we're done using it.
Top comments (0)