On Thu, Dec 10, 2020 at 02:18:27PM +0100, Ricardo Wurmus wrote: > > zimoun writes: > > > Hi Chris, > > > > On Wed, 09 Dec 2020 at 19:36, Christopher Baines wrote: > > > >> In the future, I'd strongly recommend not adding packages to the bottom > >> of modules, unless you really want the package definition to be > >> there. If every new definition gets added at the bottom, merge conflicts > >> become very likely. Related to this, I also moved the package definition > >> up off the bottom of the module. > > > > What do you mean? From my understanding, we always add new R packages > > at the botton of the files cran.scm or bioconductor.scm; mainly. > > The problem is 100% with tooling. Git needs some context to apply > patches. When you add package definitions to the bottom and that bottom > context keeps changing you will always need to tell Git how to apply that > patch. > > By picking an arbitrary location somewhere in the file you avoid this > problem because it’s unlikely that other people will pick that very same > location. > > > The only fix to avoid boring conflicts is to reduce the time between the > > submission and the merge, IMHO. > > Not even that would be a fix, because you can have two different people > submitting patches for modifications at the bottom of the file at the > same time. Applying them one after the other will result in the same > problem, no matter how fast we are. > > Of course it’s always better to reduce time between submission and > application. > > It would be *great* if we could find another way to appease git and do > the right thing for the most common case of simply adding a package > definition to the bottom of the file, no matter what context there might > be right above. > This is why I like sorting them alphabetically :) -- Efraim Flashner אפרים פלשנר GPG key = A28B F40C 3E55 1372 662D 14F7 41AA E7DC CA3D 8351 Confidentiality cannot be guaranteed on emails sent or received unencrypted