• "aylwyn" <as6 sanger.ac.uk> Apr 25 2012
• Dmitry Olshansky <dmitry.olsh gmail.com> Apr 25 2012

"aylwyn" <as6 sanger.ac.uk> writes:

Hi, I've just written a smallish program reading
lines from stdin and doing some parsing and other text
processing. I find it's about 3-4 times faster than a python
script I had doing the same thing, which is nice but not as good
as I hoped. After profiling, it seems almost half the total time
is being spent in

std.stdio.File.ByLine!(char, char).ByLine.popFront()

called once per line on stdin, and within that mostly in

std.stdio.File.readln!(char).readln(ref char[], dchar)


Before I do some benchmark comparisons with C etc, can I just
check that

foreach (line; stdin.byLine())

is the correct way to do this. Are there any lower-level methods
I could use for buffered access to a stream of chars?

(using DMD64 D Compiler v2.058.)

Dmitry Olshansky <dmitry.olsh gmail.com> writes:

On 25.04.2012 15:10, aylwyn wrote:
 Hi, I've just written a smallish program reading
 lines from stdin and doing some parsing and other text
 processing. I find it's about 3-4 times faster than a python
 script I had doing the same thing, which is nice but not as good
 as I hoped. After profiling, it seems almost half the total time
 is being spent in

 std.stdio.File.ByLine!(char, char).ByLine.popFront()

 called once per line on stdin, and within that mostly in

 std.stdio.File.readln!(char).readln(ref char[], dchar)


 Before I do some benchmark comparisons with C etc, can I just
 check that

 foreach (line; stdin.byLine())

 is the correct way to do this. Are there any lower-level methods
 I could use for buffered access to a stream of chars?


It is. It needs more optimizations though. There have been talks of 
byLineAsync() that will opportunistically fill next line in background 
and so on and so forth.

 (using DMD64 D Compiler v2.058.)


I bet straight fgets will beat stdin.byLine.

-- 
Dmitry Olshansky