PHP Joburg 2018
26 - 28 Sep 2018 at Blandford Manor

This talk was very surprising for me. There were several surprising lessons learned. Clearly the serialization I have been doing is very basic in comparison, and the fact that you are able to marry Go with PHP was something I had never thought of.

Michelle seemed a bit nervous at the beginning, maybe because of the minor IT issues leading up to her talk, but she settled into the talk soon after starting and gave a great talk.

Going forward I am going to look at serialization very differently keeping an eye on performance and the fact that serialization can always be expanded to use features I have not required yet like annotations.

I think the only thing I would have enjoyed seeing was a basic summary of what is advanced serialization. There was a quick show of hands to see if anybody does not know what serialization is, but something I learned as the talk progressed was that there is every day serialization and then there is advanced serialization. Maybe when going into the more complex elements of the talk having covered all the possible features of advanced serialization at the beginning would have been nice and helped me follow along.

I knew there were problems with using floats and precision but I did not know to what extent, Bernhard really opened my eyes and the background history of why was just fascinating looking at base ten vs base two number systems.

I will never compare a float again without thinking about this talk.

It was a very dense Maths heavy talk and there was once when adding precision to the discussion in the middle that I got lost for a short while but caught up again later.

This talk left me wondering about my own code and thinking about how can I fix things going forward and how big the problem could be with lots of hidden surprises waiting to jump out in the future.

I think the only thing for me that was missing was maybe some tips on what to do with an existing system that uses floats. Bernhard did provide rules and guides of how to do it properly but to go from float based system to string based system is not that clear. What would help would be the ability to review the slides from the talk in combination with our existing code base.

Overall a difficult topic done really well.

Learning about Hadoop was just amazing, when I looked at it in the passed myself I often found myself confused and struggled to see how all the pieces fit together.

Michael really did a great job in providing the history and how all the different pieces fitted together and explaining how Hadoop is not one thing, but lots of different things arranged together in different ways.

Its unfortunate he had to leave early and could not be part of the speaker panel as I think his topics and experiences would have been a great addition to the discussion.

I think the only issue I have was that he unfortunately ran out of time for questions. The size and scope of the topic really leaves you with a few unanswered questions, but otherwise a really great talk.

I really enjoy statistics and ranking info like this and you could see Brad went to a lot of effort to try and extract and interrogate the data in many different ways to get the most relevant data.

Brad as a speaker is great as he brings a different kind of energy to any conversation that is actually infectious and gets you excited about the topic.

I think the only issue I had was the volume of information was sometimes difficult to follow combined with learning about BigQuery and the different data sources.

The nature of the topic it may be difficult to improve, but maybe if these sub topics could be broken up into more distinct sections maybe it would easier to follow or maybe Brad should consider using a laser pointer more so we can follow better want numbers he is looking at and talking about, and maybe this would reduce the amount people like myself get a little lost in the numbers.

PHP as a community has a very large ecosystem and stats like this I think are very important and could be included in every conference in some way or form so we can keep up with changes in the community.

Test for me was one of the big takeaways from the conference and this talk was central to that learning curve. I have used testing but its not part of my every day development processes and what this talk and the previous days workshop showed me is that it needs to be part of my normal development process rather than something that sometimes happens at some point.

Liam gave a great introduction talk cover at a high level the fastness of software testing and touched on several key points and tips. You could easily break this topic up in several talks and workshops focusing on a single topic.

My only comments would that Liam could have gone a little slower. It did allow for more question time at the end but some slides were shown very briefly. As far as I can remember, Liam also played it safe by not breaking away from his slides to do something dangerous like run a test suite live, like maybe PHP Unit or a Travis test build, that would have been great and would have help some people make the connection to where testing fits in and how it all fits together.

This was a great introduction into a world that is very foreign to most developers. It went into some details about the different ways data is grouped and analyzed touching on commonly used algorithms.

When you get to the detail of it, the topic is actually a very challenging topic to cover in such a short time and I think Michael did a really great job. It was a pity we could not ask questions as Michael clearly has a wealth of experience in this field that we could all learn from.

My only comment would be that most of the examples were abstract or theoretical with small data sets to try explain the concepts and that was great and made sense, but I would really have enjoyed an example that connected with our average developer worlds, like taking Apache logs or something similar and extracting some data in some cluster analyses to yield a real world example of using data and machine learning. I suppose the difficulty here is getting the data that legally Michael can display publicly and for it to be a useful enough example to demonstrate.

A really great workshop tackling a topic that I think needs to be drilled into every new developer and re-iterated on experienced devs on a regular bases.

For me the best part was learning how you could fit all the processes together from planning with stakeholders to implementing your interfaces and lastly testing the domain choices.

James is really a great speaker and clearly has a wealth of knowledge and experience, I could spend days with him running code examples or code reviews to see his point of view on things.

For me the only really difficult part was the fastness of all the topics. I felt we sent a lot of time on event storming and a lot less time on SOLID, PHPStan and Behat.

I understand why this was this way as one thing leads to the next and it makes sense discussing it all together. However, for me it was 2 workshops run as one. I think the planning with event storming was great as a single workshop maybe with Behat or maybe not (could go ether-way).

Then the code quality side of things done separately. I know they all go together in the end but I felt we could have spent more time talking about things devs could implement immanently and also take more time to understand things like "Why could else be a bad thing?", "What does it mean to reduce indenting?" and "Why Prototyping with poor planning can sink you?".

I find a lot of devs struggle to see good code from bad, and if it works, it works. We only did it once, but simple things like looking at code together asking the group is this good or bad code and why, helps a lot of devs change how they read code, something I find takes a few light-bulb moments.

This was really great, it introduced me to a world I knew very little about and never really considered and often brushed off as not relevant to me. Going forward I am definitely going to look at cloud based technology a lot closer and Serverless gives me a clear starting path for experimenting and learning.

The only negative for me was that I think it was little ambitious to cover so much in such a short time. My suggestion would be to distribute a participation brief of all the technology and setup steps including repo's that are needed and to be installed and setup beforehand. Then a very strict policy where we don't wait for anybody the same way we are strict about time and questions of conference talks. I don't know how practical this is in practice but if it works we would spend more time on the topic we were there for and spend less time fighting Internet and Windows platform issues.

Really great talk and the amount of effort that went into data collection was impressive. Really enjoyed it and the audience interaction was very well done.

Really awesome talk that covered a lot of complex topics in an easy to understand approachable manner. Thoroughly enjoyed it!