Видео

I'm working on People's credit files. Is there a bot to automatically put these PRS on the files instead of manually putting them in

Thank you, it was very interesting

Promo-SM 😄

How to determine if a particular release produced a failure can be ambiguous. It may be a latent feature or a defect that is not noticed for weeks.

The recommendateion for complete and thorough code reviews is both very specific if describing the Formal Inspection Process or very ambiguous.

Guess what? At some point to further reduce PR size and get deeper reviews in a short time, you just switch to pair programming and push to mainline.

Lower MTTR is correlated with more work on features and less Yak shaving and bugfixes.

Thank you for this video. A Junior software engineer

I once heard an interesting point while interviewing Principal Engineer - big PRs for sure might bring troubles and to be very complex to review. But very small PRs are also a signal - in particular scenario from that Engineer - it was actually a sign of some level of toxicity within the team, where they were so constantly throwing review comments that weren't actually helping to make the code better, that irritated devs started doing very atomic PRs thinking "ok now you bastard won't tell me a single word, finally"

06:59 A lot of organizations are being very fixated on speed and just looking at speed right like coding time cycle time all of those things and it definitely needs to be balanced with with quality indicators. For example developers might actually write more tests with the time that they're saving which increases the the code coverage but would then also dampen the impact on speed. 07:28 You might you might even get slower [using GenAI in development] if developers get more removed from the code because they didn't write it all, so their debugging time might get longer as at the same time there's risks to quality. There's there's lots been written about the quality risks of [GenAI] coding assistance. 07:55 It is important to not overdo it. I've seen one organization that very strongly looked at adoption and [had] very high expectations from the leadership about the lift that this is going to give in terms of speed so putting very high expectations on developers to use these tools to become a lot faster and putting a lot of pressure. Adoption should be neutrally monitored, like are people actually using it because it's useful and not put on an expectation beforehand of how much lift it's going to get them... obviously if you're if you stated the goal everyone should be using this then you probably pushing people in the in the wrong way 08:58 When measuring benefits it's also important not to have a fixed [goal] like this has to be 30% or more [saving]. The industry doesn't know yet actually what the benefits are how they look at in numbers and how they balance with risks. 21:45 As we're using coding assistance and potentially get higher coding throughput to create more code per time unit, if we look at our software delivery process then we have this one part of our process of our system where we're increasing the throughput. If some people are maybe like familiar with systems thinking, what will happen if we just increase throughput in one of you know one of the parts of our system? We'll get bottlenecks. In a lot of organizations code review or pull request review is already one of the bottlenecks, so what will happen if you create more code? 22:38 At thoughtworks were're big fans of pair programming as a tool to improve team collaboration and also team flow right and you know this could actually be like one of the things to also alleviate this this bottleneck, to have pairs that work with the coding assistant. Almost like a trio programming. 23:08 GitHub calls co-pilot your "AI pair programmer" which which annoys me a lot I have to say, you know because pair programming is about is a team practice to make the team better. 27:49 We are in a hype cycle right now at the peak of inflated expectations and the higher that peak is the bigger the hangover will be and the more people are going to say quickly oh my expectations were so high they were disappointed. So I think the key here is to adjust our expectations because this is definitely going to be useful in my opinion and going to be here to stay and we just have to figure out how to properly use it so that it. 29:01 I can also confirm I haven't seen anything really impressive there yet in that area so often when I would ask a coding assistant for example how can I improve this code or something, it gives me like, you could can add error handling, you know like very like generic kinds of things almost like a checklist of like a a newbie code reviewer. There's definitely still potential maybe it also can still get better. 43:27 I've been coding for more than 20 years at this point and even today you just go on the internet and you copy and paste things when you don't know yet what to do and you stitch them together and you just want to get it to work once. So I think in a way we've we've all been doing this right, there's just a tool now that can potentially make us even faster at doing that and that might make the problem worse. But I also have faith that maybe we will all adapt to this and we will still put our hands on the hot stove and get burned and then learn from that. So those cycles will still be the same so I'm trying to be not too cynical about it.

Really good presentation and Q/A. I read the report(s) but this gives much more flesh on the bones. To be honest I thought that DevOps was just another bussword but after learning more I really like what it represents. It aligns very well with all my experiences as a developer.

How do I access these reports?

Killer event!

💃 *Promo sm*

I really like this video, it not just helps me to under DORA metrics but also how to implement it.

Great summary! Thank you that was exactly what I was looking for.

sorry not even understanding the post above. can you dumb it down so a Program Manager can understand?

Thank you for this! Sorry I missed the webinar

To me the value of DORA comes when paired with the capability model and the 24 capabilities from Accelerate. Without that context I think DORA looses its meaning.

Lots of really good nuggets here!

ｐｒｏｍｏｓｍ

𝔭𝔯𝔬𝔪𝔬𝔰𝔪 😜