On 12/25/13 8:29 AM, Gordon wrote:
On Tuesday, 24 December 2013 at 23:52:49 UTC, Andrei Alexandrescu wrote:void main() { size_t[size_t] unions; foreach (e; "input.txt" .slurp!(size_t, size_t)("%s %s").sort.uniq ) { unions[e[0]] = e[1]; } }Thanks for this very elegant solution! For completeness (since we're dealing with timing): 1. Running the above code with Garbage-collection enabled, takes 1m45s. 2. Running it with GC disabled takes 50s . 3. Running with GC disabled and without sort+uniq (my data is already uniq'd), takes 40s.
Thanks for the numbers!
This is a beautiful idiomatic D code, but for now I think I'll stick with the more verbose code above. Ine reason is that I want to provide helpful and verbose error messages for invalid input (e.g. "Invalid numeric value found in field 3 line 1000") - and with the "slurp" method, any input error will result in a not-so-helpful exception (e.g. "std.conv.ConvException@/usr/include/dmd/phobos/std/conv.d(2009): Unexpected 'H' when converting from type char[] to type ulong).
Thanks, I added https://d.puremagic.com/issues/show_bug.cgi?id=11816 to keep track of this.
Andrei
