Omar Rizwan
.
@omar,
uint15_t
Omar Rizwan
.
@omar,
Omar Rizwan
.
@omar,
it's funny how this is one of the most jarring initial things about writing Verilog, that you right-size your integers to very weird specific numbers of bits, there's no built-in system word size like 16 bits or 32 bits or 64 bits to snap to
Omar Rizwan
.
@omar,
also reminds me of
- (my sketchy impression of) programming with dependent types or contracts, where you specify bounds on how big stuff is & what goes in and out of functions
- programming in C, where you usually build your software out of fixed-size arrays of 10 or 100 or 1024 or whatever -- again, encoding an expectation of the scale of your data -- because making dynamic structures is so bespoke and annoying
Omar Rizwan
.
@omar,
& this, on the value of hard-coded limits in data structures: (https://news.ycombinator.com/item?id=35212434)
andrew blinn
.
@disconcision, @types.pl
@omar guy hit the hardcoded limit in his commenting subroutine
Dave
.
@daveliepmann, @social.tchncs.de
@omar reminds me of an old common lisp mailing list thread where they crucified a guy for advocating the value of abstract data structures — so the concrete implementation of the hashmap would change to one efficient for the number of keys it has