[ad_1]
At its re:Invent conference, AWS on Tuesday announced a series of updates to Developer Sits own programming assistant platform that competes with the likes of GitHub Copilot. The focus here is to go beyond code completion and assist developers with a broader range of routine tasks involved in the overall software lifecycle.
The service, which you may remember under its previous name of “CodeWhisperer,” is part of AWS’s comprehensive Amazon Q generative AI platform, which also includes Q Business (which is also getting a slew of updates today).
“What developers need is that they want Q to be the friend to solve some of the heavy, undifferentiated tasks so they can actually have more freedom to innovate,” Swaminathan “Swami” Sivasubramanian, vice president of AI and data at AWS, told me. “That’s why having an assistant — or a buddy — who helps them do things faster and more streamlined is so important, and why we focus on it so much.”
Manage the software life cycle from start to finish
Sivasubramanian told me that he believes what sets Q Developer apart from competing platforms is its focus on the entire software development lifecycle. So far, this means helping developers troubleshoot issues and perform multi-step tasks to fix them (or build entirely new apps), as well as scanning code for vulnerabilities.
At re:Invent, the company takes this a step further. Q can now, for example, automatically generate unit tests. But perhaps more importantly, he can now do the one thing many developers hate most: writing and maintaining documentation for that code. To complete this cycle, Q can now generate the first code review when developers check their code.
“At Amazon, we have this rule that no code is verified without a code review,” Sivasubramanian said. “So, if you don’t review the code, you can’t verify the code. But many companies actually don’t have enough senior engineers to review or the chief engineer says, ‘I can’t handle this many reviews.'” Can someone Should he review it first before we do that?” Q will simplify the code review process by being the first line of reviewers and automatically checking code quality, vulnerabilities, etc.
Then, once the code is produced, Q’s new operations agent can now automatically pull data from AWS CloudWatch, the company’s monitoring service, and immediately begin investigating when an alarm goes off. “It uses[the knowledge it has about]an organization’s AWS resources and then sifts through hundreds of data points across different resources in CloudWatch. After its analysis, Q comes up with a potential hypothesis for the root cause and then guides users on how to fix it,” Sivasubramanian explained.
All you wanted for Christmas was help with Cobol and .NET migrations, right?
For organizations with legacy code bases, moving to the cloud often involves rewriting much of their existing code. One of the first distinct features of Amazon Q Developer was its code conversion agent. At the time, the focus was on this agent Update old Java applications. Today, the team expands this by also helping developers modernize their legacy .NET-based applications from Windows to Linux.
Although at first glance this may seem like a curiosity, AWS is also releasing a proxy for updating mainframe COBOL applications. Many large organizations still rely on this legacy code, which few developers know how to work with today. Sivasubramanian stressed that these are very complex migrations, so the goal here is not just to translate the existing code 1:1.
“Our goal isn’t really just to like the whole COBOL project and get it out of code,” he said. “The reality is that these projects are inherently very complex. You have to have a human in the loop to take advantage of them, but I’ve heard clients say, ‘Hey, this is taking several years and we’ve explicitly told clients that this is a game-changer and it’s going to drop that timeline dramatically.’
Sivasubramanian noted that although there was less COBOL code to train models to automate code migration, the team was able to leverage AWS’ extensive expertise in modernizing mainframe applications, as well as traditional approaches to code translation.
“Moving code from one language to another is arguably the easy part,” he said. “But the hardest part is: How do you know you’ve done it right? And how do you even know what the code is doing? And so the challenge with these (code bases) is that they’re usually poorly documented and the dependencies are not well understood. So what we’ve built It’s really very innovative, and it (the system) also understands, at a project level, what the goals of each module are, and then plans and creates a migration planning timeline to actually build the code, and then create the test – and bring humans into the loop to figure out how to validate it.
[ad_2]