Thomas Gummerer <t.gummerer@xxxxxxxxx> writes: > One thing that came up in the latest reviews, was to keep the stash > script intact throughout the series, and to not re-introduce it after > deleting it. I did however not do that, as that would make the > range-diff quite a bit harder to read. Of course, if you start from a suboptimal ordering and reorder to make it right, the range-diff that tries to match and compare the steps will become larger and harder to follow. What else is new? That is refusing to think clearly hiding behind tautology. > In addition removing the > script bit by bit also allowed us to find the precise commit in which > the missing 'discard_cache()' bug was introduced, which made it a bit > easier to pinpoint where the bug comes from imo. In an ideal ordering, you would do a command line parser first and dispatch the rewritten commands piece by piece to the C reimplementation while diverting the remaining unwritten parts to scripted "legacy" one. That would allow us to find the precise commit that was faulty the same way. Having said all that. I do not care too deeply about seeing this particular series done in the right order anymore. Everybody's eyes are tired of looking at it, and I do not mind to see you guys declare "nobody can review this, so let's keep it in 'next' and patch breakages up as they are found out", which seems to be happening here. We can divert our review cycles to other topics that haven't faced reviewer fatigue; they still have chance to be done right ;-).
