On Tue, Oct 01, 2024 at 11:02:53AM +0100, Phillip Wood wrote: > I'd argue that users who are not familiar with (e)dit are more likely to > make mistakes when editing hunks and are less likely to be able to fix them. This series is not about making errors more descriptive or making (e)dit more (or less) accessible. Editing the original hunk is already quite challenging and prone to errors. This series is about regaining the possibility for the user to see and correct their mistakes. > > > This is still missing "n q". Apart from that the test is looking good. > > > > I've been resisting the idea of "completeness", because I think "e y" > > should also be fine. But I'm not going to resist anymore here :-), > > since I don't think the test has much more value without "n q". So > > I'll add it. > > The reason I think we should have it is that the tests ought to be testing > realistic user input and not rely on getting EOF which is unlikely to happen > in real life. Sometimes, I use ctrl+d instead of 'q'. So, some of my interactive "add -p" sessions can be described better with "y" than with "y q" or "y n... q". And I'm real ;-) Of course, non-interactively: "e y" or "y" work as expected as the test in this series and others in t3701 demonstrate. Since we already have other tests, I didn't mind adding "n q" in an attempt to move the series forward.
