On Thu, Jul 25, 2024 at 10:45:04AM +0100, Phillip Wood wrote: > Hi Rubén > > On 24/07/2024 17:12, Rubén Justo wrote: > > On Wed, Jul 24, 2024 at 04:21:53PM +0100, phillip.wood123@xxxxxxxxx wrote: > > > > That was my initial thought [*1*] when the problem with "dash > > 0.5.10.2-6" appeared. > > > > Junio proposed [*2*] documenting the changes to address it as a separate > > patch, and I think it makes sense and it is valuable to capture the > > situation this way in the history. > > We normally avoid merging commits that are known to break our ci. Fair point. I'll reroll and let Junio decide, or ignore ;). > We can add > a comment about the dash problem to the commit message when this fixup is > squashed. Also the problem is now documented in > Documentation/CodingGuidelines which is more likely to be read by other > contributors. > > > Regarding the bisectability, I don't understand what stops from being > > bisectable. Except in a scenario with a shell like "dash 0.5.10.2-6" > > there won't be any problem. > > But we know that shell is in use in a popular Linux distribution so it is a > problem for those users. It's not an excuse; I just wanted to point out that Ubuntu 20.04 was updated with a version of "dash" that doesn't have the issue we're seeing in our CI for quite some time now. The version I mentioned in message [1/2] of this thread: dash 0.5.12-9. > > Best Wishes > > Phillip > > > And in one with it, which should be > > uncommon, the situation is well explained. > > > > So, I dunno. > > > > 1.- 2b57479c-29c8-4a6e-b7b0-1309395cfbd9@xxxxxxxxx > > > > 2.- xmqq7cdd9l0m.fsf@gitster.g Thank you for trying to make Git and its history better.
