https://bugs.documentfoundation.org/show_bug.cgi?id=155087
--- Comment #33 from Gabriel Masei <[email protected]> --- (In reply to Mike Kaganski from comment #24) > (In reply to Gabriel Masei from comment #22) > > If there is a probability, however small, that the suggestion could be > > wrong or > > the existing form could be a valid one then no auto-correction should be > > performed. > > Please note that the following is just nitpicking on the "however small". > > Consider English replacement i->I. There *is* a non-zero probability, that > the author actually wanted to have the "i" in their text. One case is using > it as a Roman numeral; another is just showing an English alphabet letter in > the text, and so on. But the replacement rule is useful, because the > frequency when i was used incorrectly (I was intended) is *much* higher than > the expected use of i. > > So there is *some* margin of allowable errors here :) Shouldn't that be performed at grammar checking level ? As at that level this distinction can be made. Anyway, let's suppose that it should be made at auto-correction level. In this case I understand that there could be exceptions. But they should be treated as such: exceptions. This means that an exception of this kind should be discussed and agreed exceptionally. Otherwise, if we define a principle that will give room to interpretations then we'll face the same issues. So the principles should remain but maybe adding a fourth principle would be helpful: there could be exceptions but they should be treated as such and extra feedback and acceptance steps should be performed before accepting them. And those extra steps should be explicitly stated. -- You are receiving this mail because: You are the assignee for the bug.
