Hi Massimiliano, On Fri, May 17, 2019 at 08:17:34AM +0200, Massimiliano Bellomi wrote: > Thank you Christopher. > > Is it fine for you if I send a 9th patch which fixes the 3rd one ? ( ...so > patches still be applied in sequence from 1 to 9, patch 3 included )
No, please never submit a patch that is known wrong and needs to be fixed by another one. It is important because developers spend a great share of their time dealing with complex bugs that require lots of bisects, which means commits must have the most correct behaviour possible. The worst thing (and it still occasionally happens) is when a bisect operation lands on a commit which doesn't even compile... I'm never going to blame anyone for integrating bugs or stuff that doesn't compile, it happens to all of us from time to time. Sometimes you perform a last minute cleanup after your tests, you get distracted, you submit and it's broken. OK. But distributing a bug on purpose, no, please never do that. Thus please fix your original patch to include this fix. Also, please make your commits a bit more descriptive, think about the "what, why, how" principle I often ask for (ie what this change does, why it does it, and how you chose to do it). And when it's a bug, it's "what is the problem, why does it do this, and how do you fix it". Actually reviewing patches that contain no info takes a lot of time because one has to figure based on the patch what was attempted to be done. Usually a well described commit is immediately applied (or commented) and a not well one can rot for several days until someone finds the time to do it, and has the willingness to edit the commit message to place in it all what they figured so that this info is not lost for the future. When in doubt, think you're trying to sell me your patch series and your role is to convince me to buy it. It's not far from the reality anyway since any patch adds a little bit to the maintenance effort. Thanks, Willy
