Spotify halved build times with just one script
labs.spotify.com
Spotify halved build times with just one script
1–7 of 7 posts
Re: Spotify halved build times with just one script
#2Re: Spotify halved build times with just one script
#3Re: Spotify halved build times with just one script
#4Although the impact is incredibly minor, I'm sure consistently removing features also helps with build times.
Re: Spotify halved build times with just one script
#5Although the impact is incredibly minor, I'm sure consistently removing features also helps with build times.
Why do they do this? It seems every update the Spotify client becomes less and less useful.
Re: Spotify halved build times with just one script
#6Earlier quoted context omitted.
Why do they do this? It seems every update the Spotify client becomes less and less useful.
Money, most likely. Removing or obfuscating user choice allows them to push you towards content that they curate themselves, presumably to pay fewer royalties or something similar. I kind of ignored it until they started removing dislike options, at some point not even having hulu bundled will stop me from switching to a competitor.
Re: Spotify halved build times with just one script
#7All I could think of is, Rust already does this properly. It doesn't recompile libraries every time. It compiles them once, or again if the compiler is upgraded.
Additionally, the Rust community has a propensity (from what I've seen) to ship dependencies on v0.x.y libraries. I mean, just look at the version numbers on https://crates.io/! Below v1.0.0, otherwise-minor sem-ver updates are breaking, and can't be de-duplicated. (0.6 is breaking from 0.5, though Cargo does allow 0.6.1 from 0.6.0 upgrades)