Bug 20373 – Line counter with async Buffer

I tried to use the asyncBuf to speed file processing as it is described into the documentation: https://dlang.org/phobos/std_parallelism.html#.TaskPool.asyncBuf.2 Thus I use one script to generate file to a given size: - https://paste.fedoraproject.org/paste/0zCnwLcPLpalAE7q0BDnyQ usage: file_generator -o test_11k -w 11k file_generator -o test_401k -w 401k And another one which count line using an async buffer: - https://paste.fedoraproject.org/paste/HW8Ti4rqLBVyvDQOD~GyMw usage: counter_async_buffer -n 1 -t 1 -i test_11k counter_async_buffer -n 1 -t 1 -i test_401k The problem comes when I process the test_401k file, as the line counted is wrong (I checked using wc -l) after a closer look it seem that comes from the reused buffer of asyncBuff which at last iteration the result do not own a lesser size than the requested buffer size as explained into the documentation. Indeed in this file the end is: 1 190121746114132251381321230342516302196252336238211523943272873744285119323293314107 316322221221132661353262123081115418570291330356278322215013742329426 714213310231111593822146521912312869120169289362332157427352432313112226373403123825 6812101511112462691294263101232 90182312212511430133514352114282271133753782360462351124233948222161956731321 11822481725231121323330910521376234322119392811262411335432102108273 463633112212153255811679207 while the code give: 1 190121746114132251381321230342516302196252336238211523943272873744285119323293314107 316322221221132661353262123081115418570291330356278322215013742329426 714213310231111593822146521912312869120169289362332157427352432313112226373403123825 6812101511112462691294263101232 90182312212511430133514352114282271133753782360462351124233948222161956731321 11822481725231121323330910521376234322119392811262411335432102108273 4636331122121532558116792071512 1553203330299234738282167126033321272154232912111312416461868182323242 111932223509162312223104310231321116573254736811479599513441112318312221230321154 11363210249282132717260536372386522748224512596323581311 121932131303243861212327470532121636029110222323531121763 2499315312114672213683218207122451351984311032612832096363463812 .... so they are an extra content

> auto asyncReader = taskPool.asyncBuf((ref ubyte[] buf) => file.rawRead(buf), I think this should be: > auto asyncReader = taskPool.asyncBuf((ref ubyte[] buf) { buf = file.rawRead(buf); }, See https://dlang.org/library/std/stdio/file.raw_read.html

Thanks Vladimir Indeed this fix imply a correct number of line counted. It is still much slower than wc but it works thanks

Marking as invalid as there is no bug in any D components.

ok Vladimir, maybe to provide documentation more clear?

The documentation looks good to me as it is, but if you can think of some way to improve it that would have avoided this mistake, please do submit a pull request.