#general
- h
happy-kitchen-89482
02/06/2016, 11:41 PMBut Go is far more expressive than C - h
happy-kitchen-89482
02/06/2016, 11:41 PMEven though it’s a fairly lightweight language - b
bored-art-40741
02/06/2016, 11:41 PMI think it's nice to think of all of these languages as having 4-5 measurable axes of usefulness - b
bored-art-40741
02/06/2016, 11:42 PMScala is really nice on most of the axes, but it's enormously expensive to compile and analyze. And when you slow down iteration time, you lose time and money, fast - b
bored-art-40741
02/06/2016, 11:42 PM4 A's and a record-shattering F disqualify a language, no matter its merits - h
happy-kitchen-89482
02/06/2016, 11:43 PMCase study: Foursquare’s infra team rewrote one of their most critical services in Go, and the benefits were massive. - b
bored-art-40741
02/06/2016, 11:44 PMTo be fair, that was really a case study in JVM vs native - b
bored-art-40741
02/06/2016, 11:44 PMAt least, the benefits were - b
bored-art-40741
02/06/2016, 11:44 PMAnd it wasn't possible without mesos or some other container based framework - h
happy-kitchen-89482
02/06/2016, 11:44 PMFor example, Go’s low runtime overhead (~30MB instead of hundreds of MB for the JVM) meant that they could run a lot more of them per machine, and let Aurora schedule them in a fine-grained way. - h
happy-kitchen-89482
02/06/2016, 11:44 PMSo? - h
happy-kitchen-89482
02/06/2016, 11:44 PMScala requires the JVM - b
bored-art-40741
02/06/2016, 11:44 PMNot contesting the point - b
bored-art-40741
02/06/2016, 11:44 PMJust saying that it extends beyond scala as well - h
happy-kitchen-89482
02/06/2016, 11:45 PMTrue, but a Java version of that server would have been a lot lighter weight than the Scala version. Still heavier than the Go version. - h
happy-kitchen-89482
02/06/2016, 11:45 PMMaybe not lightweight enough to push scheduling to Aurora, but possibly. - b
bored-art-40741
02/06/2016, 11:46 PMIt could be argued that scala, if you pretend the compilation and tooling issues don't exist, suffers more from targeting the JVM than it benefits - h
happy-kitchen-89482
02/06/2016, 11:46 PMAnd the other benefits (performance, compile times, tool stability) are unrelated to the JVM comparison. - b
bored-art-40741
02/06/2016, 11:46 PMHaving to generate a class for every closure is crazy town - b
bored-art-40741
02/06/2016, 11:46 PMAn interesting argument that I've heard is that scala is good for recruiting and retention - b
bored-art-40741
02/06/2016, 11:47 PMNot so sure about the latter, might be an old guard sort of thing. But if you're going to go that route, I feel like Haskell will treat you better - b
bored-art-40741
02/07/2016, 12:15 AMDo we have a standard way of defaulting an option in terms of the workdir? - b
bored-art-40741
02/07/2016, 12:16 AMI thought that I could somehow grab this off of
as passed intoregisterregister_options - b
bored-art-40741
02/07/2016, 12:22 AMregister.bootstrap.pants_workdir - b
bored-art-40741
02/07/2016, 12:22 AMAha - b
bored-art-40741
02/07/2016, 12:40 AMTricky, for what seems like it should be a first class and visible feature for a plugin author - h
happy-kitchen-89482
02/07/2016, 4:06 AMI have a TODO somewhere to replace the register function with an instance of a class👌 1 - h
happy-kitchen-89482
02/07/2016, 4:06 AMRight now bootstrap is a property on a function, which is icky - w
wide-energy-11069
02/07/2016, 4:12 AMGiven the new option deprecation cycle, shall we mark the option deprecated inCopy code
as well?./pants options - b
bored-art-40741
02/07/2016, 8:54 PMWould appreciate some eyes on this change: https://rbcommons.com/s/twitter/r/3436/