3 ms·
I actually had the same thought/idea just this morning. At $dayjob I consult with various teams, some of them developing new web apps to deploy on the cloud. S
by jiggawatts 4y ago
I actually had the same thought/idea just this morning.
At $dayjob I consult with various teams, some of them developing new web apps to deploy on the cloud. Some of these groups are scientists with minimal computer science experience, and no clue about cloud architecture, devops, or even basic software development best-practices.
In a mere hour of glancing at their code, I could provide them with valuable tips that cut days or even weeks off their development time, massively improved performance and security, etc...
There are many solo developers or small teams out there that could benefit from peer review of their code by someone much more experienced, but their org just doesn't have anyone available with that skill.
So "code review as a service" might be a pretty awesome thing. Grant the review SaaS access to your Git repo, approve access to some subset of the files, and it'll pick someone with the skills that match the contents. E.g.: for C# code it'll find an experienced C# dev.
The system could provide some safety nets such as automatically masking secrets. Similarly, it could perform a search & replace to hide product names or company names.
- overview 4y agoGenuinely very interested to hear an instance where you were able to cut weeks off development time.
- jiggawatts 4y agoMany instances, and in some cases it was more like months or even years. One large government department struggles to hire skilled application architects, because they can only pay public service rates in a highly competitive market. The only hires they can get are 6 month contracts for relatively inexperienced architects who use the job to pad out their resume and then immediately move on to private industry. To make their resumes look good, they throw the kitchen sink at every design. A recent project was a simple web app that's an extension to an existing web app. It implements trivial things such as adding sign-in to an otherwise static site, some email notifications, mobile push, reporting, and the like. Nothing fancy. They were a couple of months into development, and they already had: - Multiple web apps - Multiple API apps - A "mid-tier" behind the above - Dozens of lambdas/functions - Logic apps and workflow apps - Multiple databases - Multiple Kubernetes clusters despite most of the above running on a different PaaS platform already. - ETL jobs for copying data between databases that can directly access each other. - All of the above stretched across both Azure and AWS, just for funsies. Etc... It looks really impressive on a high level architecture diagram! Fantastic in a resume. But it's a ton of work for a developer to implement, and achieves nothing. Microservices and lambdas/functions have their place, but not for a single-person web dev project! Throwing out 90% of the "moving parts" from the design probably saved a year or two all by itself. Then I started demonstrating how to use storage emulators for local development, how to manage dev/tst/uat/prd configurations, how to get logs and debug dumps from the PaaS environments in the cloud, etc... That workflow stuff alone would save weeks or months, easily. All this took was just a couple of hours of chatting. Another meeting was less than an hour, and while the guy was scrolling through his code I yelled "Stop!" because I saw a reference to a low-level cryptography primitive. He had started trying to implement authentication and authorisation "from the ground up". I suggested using a popular JWT library instead. Just showing a developer how to use an APM tool like Application Insights or New Relic can cut weeks off the troubleshooting time over the lifetime of a project.