#general
- c
cuddly-accountant-94274
09/03/2015, 2:59 AMIn the context of the build cache I presume? - b
bored-art-40741
09/03/2015, 3:00 AMYeah - e
enough-analyst-54434
09/03/2015, 3:00 AMIs pruning not good enough? - e
enough-analyst-54434
09/03/2015, 3:00 AMie is there that much churn / that much going back in time? - b
bored-art-40741
09/03/2015, 3:00 AMThough with a trivial layer of indirection you could convert an LRU blob store to solve the same problem I was talking about earlier - b
bored-art-40741
09/03/2015, 3:00 AM(Just map fingerprints to canonical content hashes) - b
bored-art-40741
09/03/2015, 3:01 AM@enough-analyst-54434: In scala land we put out a lot of artifacts and they are transitively invalidated quite quickly - c
cuddly-accountant-94274
09/03/2015, 3:01 AMNaive reaction: For user purposes it would seem likely that you basically have a working set and you either fit everything and you're happy, or you don't and you're unhappy. For a CI like environment a hit rate > 0 but still small might still be ok. - b
bored-art-40741
09/03/2015, 3:02 AMNon-snarky question: have you ever worked on a large scala codebase? - e
enough-analyst-54434
09/03/2015, 3:02 AMto me or peter? - b
bored-art-40741
09/03/2015, 3:02 AMPeter - b
bored-art-40741
09/03/2015, 3:02 AMI know your lineage to at least one important degree - e
enough-analyst-54434
09/03/2015, 3:02 AMha! - e
enough-analyst-54434
09/03/2015, 3:02 AMScala is great, scala is ridiculous - e
enough-analyst-54434
09/03/2015, 3:02 AMIts a long live the king sortof thing - c
cuddly-accountant-94274
09/03/2015, 3:03 AMI've been exposed to large amounts of scala, but I haven't personally written much aside from one library. Mostly done Java on the JVM. - c
cuddly-accountant-94274
09/03/2015, 3:03 AM(at twitter) - b
bored-art-40741
09/03/2015, 3:03 AMAh okay - c
cuddly-accountant-94274
09/03/2015, 3:03 AMOh, you mean there's a lot of churn due to small changes butterfly-effecting? Sure. - b
bored-art-40741
09/03/2015, 3:03 AMI'm mostly referring to the productivity constraints of compiling scala as efficiently (and little) as possible - c
cuddly-accountant-94274
09/03/2015, 3:03 AMOh yes. I'm aware of the "speed" of the Scala compiler. simple smile - b
bored-art-40741
09/03/2015, 3:04 AMThough my LRU rant also comes out of the variety of relatively unrelated situations in which I've had this need - b
bored-art-40741
09/03/2015, 3:05 AMStoring python wheels, various binaries - b
bored-art-40741
09/03/2015, 3:05 AMThings that I can identify once and even provide a hash for, and I don't want my user to have to redownload when I push a flag that indicates to the build tool that things need to be rebootstrapped - c
cuddly-accountant-94274
09/03/2015, 3:06 AMdefinitely - c
cuddly-accountant-94274
09/03/2015, 3:07 AM(my wet dream used to be a software packaging system that would combine the best of source based packaging systems with binaries, using content addressible storage etc. nixos is the closest thing i've seen.) - b
bored-art-40741
09/03/2015, 3:11 AMI went down a long, long rabbit hole of how to "relativize" python wheels that have platform specific so's in them - b
bored-art-40741
09/03/2015, 3:12 AMI ended up concluding that I could make an OS for every wheel or I needed a way to consistently identify the target ABI for a binary dependency - b
bored-art-40741
09/03/2015, 3:12 AMAnd hilariously this is the one thing that OSX clearly won at: the ABI is like two dynlibs and everything else is relatively safe to relativize - b
bored-art-40741
09/03/2015, 3:14 AMAt the end of the day scope restriction and careful target environment management wins, though