Apply suggestions from code review

Co-authored-by: Jan Hrcek <[email protected]>

Author: nomeata

podcast/52/transcript.markdown

@@ -158,9 +158,9 @@ So, for technical reasons, we decided to update the version of GHC that we used
 
 The next hurdle was to improve the build system because the build times were getting out of hand as the codebase was growing. Even though it was using Shake, it wasn’t doing any caching across machines. So, we were the first adopters of Cloud Shake. We’re looking at what other companies and other teams were doing with Bazel, where you can have a CI build farm that will ensure that for every recent version of the codebase, building in e-developer machines, it’s instant because it all comes from the shared cache.
 
-So, Neil had started adding this feature to Shake, but no one was using it, I think. So, we were basically the first adopters. This was all my doing because, like I said, I was focusing on developer productivity, as that’s what I thought was important for the team. So, we were sending fixes and contributions to Shake, inviting *(0:31:34 unintelligible)* to get immersed and released promptly. It was really challenging to make this work properly because you need a completely hermetic with completely explicit dependencies set of build rules to accomplish cloud cache. And Shake wasn’t making this particularly easy. But once we managed to get working, the productivity improvement was amazing. So, you could build a codebase in seconds, maybe a few minutes compared to the several tens of minutes that you stick previously to this one. And this was one of team complaints of new team members. How difficult it was for them to be productive, how long it took to build a codebase, how long it took them to switch from one branch to another, and so on.
+So, Neil had started adding this feature to Shake, but no one was using it, I think. So, we were basically the first adopters. This was all my doing because, like I said, I was focusing on developer productivity, as that’s what I thought was important for the team. So, we were sending fixes and contributions to Shake, buying Neil beer to get that merged and released promptly. It was really challenging to make this work properly because you need a completely hermetic with completely explicit dependencies set of build rules to accomplish cloud cache. And Shake wasn’t making this particularly easy. But once we managed to get working, the productivity improvement was amazing. So, you could build a codebase in seconds, maybe a few minutes compared to the several tens of minutes that you stick previously to this one. And this was one of team complaints of new team members. How difficult it was for them to be productive, how long it took to build a codebase, how long it took them to switch from one branch to another, and so on.
 
-So, after our cloud build system was up and working, my next goal was to get an IDE for it. And remember, I said previously we were using Intero backup. But I think in 2019, Chrisdone stopped maintaining Intero, plus Intero had never been converted to any other build system other than Stack. So, it would have been an option for us here. So, I was looking around, trying to find something that worked. I mean, internally, we were using GHCi or GHCid as a *(0:32:59 unintelligible)* feedback loop. But I mean, our focus was getting more and more sophisticated. The types were getting more complex, and I could feel the need for an IDE, especially for the more junior team members. So, looking around, I found GHCide, which was a new prototypical very early-stage IDE that had come out of Digital Asset, where Neil was working. 
+So, after our cloud build system was up and working, my next goal was to get an IDE for it. And remember, I said previously we were using Intero backup. But I think in 2019, Chrisdone stopped maintaining Intero, plus Intero had never been converted to any other build system other than Stack. So, it would have been an option for us here. So, I was looking around, trying to find something that worked. I mean, internally, we were using GHCi or GHCid as a poor man's feedback loop. But I mean, our focus was getting more and more sophisticated. The types were getting more complex, and I could feel the need for an IDE, especially for the more junior team members. So, looking around, I found GHCide, which was a new prototypical very early-stage IDE that had come out of Digital Asset, where Neil was working. 
 
 *AL (0:33:24)*: Again, invented by Neil, I guess.