Sorry about that; looks the mirrors have caught up now. Here are 5 iterations of that same test on my machine,
# 1 bash 0m0.27s mksh 0m8.47s ~/builds/mksh/mksh 0m0.24s # 2 bash 0m0.29s mksh 0m8.55s ~/builds/mksh/mksh 0m0.27s # 3 bash 0m0.32s mksh 0m11.28s ~/builds/mksh/mksh 0m0.27s # 4 bash 0m0.34s mksh 0m9.92s ~/builds/mksh/mksh 0m0.25s # 5 bash 0m0.27s mksh 0m8.18s ~/builds/mksh/mksh 0m0.24s where ~/builds/mksh/mksh is built from https://github.com/MirBSD/mksh/commit/81a25f2. Not only is mksh much faster than R57... it's also consistently a fraction faster than bash 5.0.11(1)-release (on this box, with my build options, etc.) Thanks for the fix! -- You received this bug notification because you are a member of mksh Mailing List, which is subscribed to mksh. Matching subscriptions: mkshlist-to-mksh-bugmail https://bugs.launchpad.net/bugs/1855167 Title: Comparatively poor += performance Status in mksh: Fix Committed Bug description: Heavy use of += notably impacts script performance. Consider the following micro-benchmark (pattered after real script content): i=0 s= while ((i < 30000)); do ((++i)) s+=$i done which creates 138,894 character long string. On my system (macOS 10.14.6, 3.5 GHz i7), this takes ~8 seconds in mksh, compared with ksh's ~0.1s and bash 5's ~0.3s. Here're `real' timing figures from my most recent run (these figures are quite stable): - mksh r57: 0m8.17s - ksh 93u+ 2012-08-01: 0m0.10s - bash 5.0.11(1)-release: 0m0.30s It's no surprise that ksh93 is much faster, given its heavy optimisation. Striking, though, is the poor performance of mksh relative to the latest bash. To manage notifications about this bug go to: https://bugs.launchpad.net/mksh/+bug/1855167/+subscriptions
