PHPSW: Test or else... Legacy
9 Jan 2019 at Runway East: Bristol

Good talk and I liked the overall message, essentially "test more than the happy path".

Good delivery, I liked how you engaged with the audience.

If you're going to live demo anything then TDD is the thing to live demo. For future talks see if you can change the colour, the live demo was not the easiest to read at the back.

I wonder if you could pick a different example to illustrate test to break. If I were doing TDD then I think you'd cover off most of the examples in normal 'happy path' testing.

However if you did something like the roman numerals kata, you could show working tests for the happy path range, say 1 to 4000. But then ask for your test to break, what happens if someone provides a negative number. Or a number like 1 million.

Good luck at PHP-UK.

Very entertaining talk. There was a good selection of advice.


The translation layer is a good approach. My only suggestion for improvement would be I'd have to have seen more practical tips like the translation layer, possibly at the expense of the more higher level suggestions.

In summary a great talk. I hope you submit to many conferences, others would benefit from it too.

I watched this talk on YouTube (https://www.youtube.com/watch?v=76C1TNI6Sdw). It's a really nice talk; there were lots of funny comments, which is great for the entertainment value.

There were lots of practical suggestions for working with legacy, many of which are actually applicable to green-field projects. I actually think the differences between these different kinds of projects are not so obvious after all.

Thanks, and just keep talking!

Rob, thank you for your talk - it was excellent. I have only a few minor items of feedback, and I put them them forward knowing I could not possibly have made a better job of it. I hope they are helpful.

There were several mentions of the word "obviously" when a point was being made. This is probably not necessary - some things that are obvious to senior engineers may not be obvious to juniors. Indeed, I think it is fine to state first principles and base assumptions, even if it is a basic as wanting your car key to start your car.

There was a section on working code where you mentioned "bug free code", and I was going to make a point that there is no such thing - and then a couple of minutes later you beat me to it - "bug free as humanly possible". That was very good. My only remaining feedback here is that the main sub-point may have taken a bit of time to emerge.

A couple of times the screen could have done with being higher, or the slides not reach all the way to the bottom - from the back, the material at the bottom was cut off by people's heads.

Finally, I experienced some amusement when I thought I heard you say "famous communists like Milton Friedman". Perhaps I should get a "herring aid"! I presume you said "economists" :-)

Huh. It's possible my comment got chopped off due to a malfunctioning filter.

I'd put "The slides don't need ' < ? php ' at the start of each of them", which presumably got chopped off as a hack attempt....continuing the other stuff that was lost:

Also, although I agree that brackets belong on the next line when writing code, for slides have the { on the end of the method declaration line saves a whole line of space.

Finally blue and green should be avoided as text colours. People's eyes find it much harder to resolve blue colours than red, and slightly harder to resolve green than red. Changing from colours that are appropriate in an ide, to very light pastel colours will make the text easier to read. e.g. even as bright as #dfdfff is noticeably blue in slides, but is much brighter and so stands out much better.

I think spending more time on why rewrites are bound to fail would improve the talk, as it's an important point, and although I agree with what was said, if I hadn't already heard the arguments in depth before the talk, I doubt I would have been convinced during the talk.

If you can, end the talk on a solid take-away point, rather than a list of stuff as it gives a 'stronger' impression.

The readability of the slides could be improved:

* For the non-code slides, just use all the space available rather than just the middle and decide if you want the background light, then have all dark fonts, or a dark background and light text.

* For the code slides, again use way more space but also optimise the code formatting for a presentation rather than editing the code.

* We're not staring at the code for ages, so it's better to use a black background rather than they dark grey of an IDE theme.

* The slides don't need '

Great talk.
It went in to just about the right level of technical detail, while still getting across a great set of principles.
I hope you get to try crumpets one day!

Really enjoyed the talk and definitely agreed with the points made. The pace and delivery were spot on in my view; the courageous live coding demo was very effective.

The only thing I'd add is to give a very brief definition of what a unit test actually is (just a few words rather than a lesson), if somebody hasn't seen one before they might be a bit lost.

Great talk, really funny!
The only suggestion, maybe explain a bit more about the different concepts for rookies like me.

Good talk, really well organized.
As a suggestion, don't be afraid of waist a bit more of time with the examples (or not use them at all).

I'm ready to break my car!