Closed wolfgangw closed 10 months ago
9769fe4b0a670a02ad988e8a87badef61e53e9eb is a stab at that. By a) using fdfind
and b) going directly for AM candidates I'm seeing a significant speedup. Needs more testing over large trees.
This changes used and displayed asset paths from relative to full. Not quite sure yet how I like that. See related package()
and clean that up.
Run over a 40 TB tree with ~1000 AMs/packages survived. Initial AM search is way faster with fdfind
and the explicit search.
Looking ok so far. Needs more testing for completeness of AM search result.
Add an interim timer after AM search
Initial AM search over the 40 TB tree ~4 min. Look into stream processing of fdfind
's progress to get stuff going faster.
See e.g. https://mensfeld.pl/2016/12/ruby-stream-processing-of-shell-command-results/
Out of curiosity and as yet another tidbit: report number of AM candidates vs. number of actual AMs.
Startup over a large dirtree is unacceptably slow. Find a better way.