5 Steps To Fast & Reliable Software Development

[@ContinuousDelivery]

🎓 STUDY DEPLOYMENT PIPELINES ONLINE WITH ME | GET £100 OFF FOR A LIMITED TIME ONLY 🚨 Explore, in-depth, each stage of the Deployment Pipeline and learn best practices and techniques to turn new ideas into quality features for users quickly and reliably. FIND OUT MORE HERE ➡ courses.cd.training/courses/the-anatomy-of-a-deployment-pipeline

[@danielwilkowski5899]

This is gold! This looks like pure pragmatic knowledge, and zero of the noise, buzzwords, and bs.

[@danielwilkowski5899]

@ContinuousDelivery I reckon that with continuous integration, a team produces hundreds if not thousands of commits daily? I mean, if you should complete a small value to the user in say 10-20 minutes, which should be possible with TDD, and you do that all day, then surely after day of work of a team, the team will have created hundreads of commits. Dave, how important is it to name commits very carefully? Should I put as much or more care into naming commits than naming classes and variables? I tend to be picky when it comes to commits, but I'm wondering - at the back of my mind - if dave says continuous integration works well, maybe there's no need to name commits well, since you never have to go back? if we always commits changes that pass the tests, and we have near 100% coverage, and we do real tdd, then maybe I can just make commits like in Kent's Beck TCR game, where commits don't have messages, and it's "just fine"? What's your thoughts on that?

[@borjonx]

“Have we successfully managed to manage the complexity of The problem in our solution?” @3:45 - that’s some next level linguistic voodoo right there, and gets to the root of the challenge; managing complexity. Kudos!

[@ContinuousDelivery]

That is really one of the key themes of my book "Modern Software Engineering"

[@georgehelyar]

Fast feedback really is the basis of good software development, whether it's CI, CD, TDD, agile, etc. I don't get how people find this hard to understand

[@ayeo]

In particular circumstances, it is simply not feasible to introduce CI/CD. Some businesses have heavy processes, certification, deployment policies, data access levels, etc., in place.

[@ContinuousDelivery]

...and there are examples of other orgs in those same industries and same circumstances that work, more effectively, doing the things that I describe. I spent much of my career working in regulated industries, mostly finance, but some medical, some gambling, and CD works better in those circumstances. I have a blog post on that topic here: www.davefarley.net/?p=285

[@karlstenator]

I wish that everyone and anyone that I work with professionally could watch your videos. 💚

[@2bfrank657]

"Semantic diffusion" what a useful term! Your explanation immediately brought "root cause" to my mind. The true root cause of something is found by asking why repeatedly in response to explanations given. I find however that root cause is just used as a fancy word for "cause" by many people.

[@onebeartoe]

Thanks for sharing. I like how you re-did the build to fail sooner.

[@seetlive]

Great tips, thank you for this good quality advice

[@ForgottenKnight1]

Sometimes a sign off is needed because of bureaucracy before a release is done. I worked on some project, where I had to get an approval from my manager for every deployment....

[@WilliamPowerDental]

Is it compliant? Is it resilient, is it scalable? You mentioned these can be automated in the Acceptance phase. Any tips on how to do this?

[@ContinuousDelivery]

Yes lots of tips, there is a lot of content on this channel that goes into different aspects of that, or if you want a more guided tour, check out some of my training courses. Probably the most relevant to your question is "Anatomy of a Deployment Pipeline" which is currently on sale with £100 off till the end of January. courses.cd.training/courses/the-anatomy-of-a-deployment-pipeline

[@CarlosVera-jy7iy]

Great content I have to get your courses, but the currency exchange don't help me 😢

[@kayakMike1000]

Oh if this could be a reality in firmware motor controllers...

[@kayakMike1000]

It could be.

[@vitalyl1327]

In some places, full MISRA-C is a requirement, luckily.

[@almari3954]

Where significant concurrency is involved no amount of testing will help. Model checking and full formal verification would help, however this is still slow and very expensive...

[@vitalyl1327]

@@almari3954 ideally there should not be any concurrency whatsoever in a BLDC controller. Designs tbat utilise concurrency (such as ODrive) are just wrong. A proper real-time design will be an FSM with precise timings of all measurements and PWM settings, wile communications are handled entirely independently on a separate core.

[@ContinuousDelivery]

It can, and it is, this is how Tesla write the code for the ECUs in all of their cars, it is how many orgs develop embedded systems, here is a conference talk on the topic form a few years ago (not one of mine) kzbin.info/www/bejne/epSplmmZetuhfbMsi=OdoCbMUrZCHK8ENZ

[@disgruntledtoons]

Step 0: Management pulls its head out of its arse. The other steps become more like after this happens.