@ChrisK Extrapolating from the peak, showstoppers should be done in about... eight days. So, no release this week I guess, unless they're all closed in a single drop.
I wonder if all companies developing products (instead of projects) are stuck in sort of pathological optimism on what they can accomplish for the next release.
@CarlLange I also noticed that they tend to avoid Friday releases. Maybe in order to have better capacity to do paclet updates or .0.1 releases quicker on embarrassing issues?
@Kuba "Deploying" is a bit different from a software release, though. Surely bracing for download spike can still be a reason to not release on Friday...
I'm a somewhat suspicious of major customers adopting a new major version on the same day as it's released. I think major customers working with software/systems releases I work on have processes which are likely to take several months before a major version of our product is actually deployed in production.
Would anyone be able to check for me if they can reproduce the results of mathematica.stackexchange.com/a/191795/24936 ? Because for me I don't seem to get the same result, I just get an empty list
It's a bit weird and unfortunate that you need to watch WTC talks to learn about things like Developer`ToPackedArray. Have just sped up my code by about 4 hours.
@CarlLange I'm looking forward to the 12+ NumericArray because I can (hopefully) use it as a wrapper for all my packed array data types to protect them from unexpected unpacking.
Also allows you to control the exact storage type which is pretty nice. And supposedly it'll play well with the compiler.
@b3m2a1 Sounds really interesting. I was a bit sorry they didn't do any livestreams about the compiler, but then I suppose it might all get very in the weeds looking at LLVM bytecode
