Should we disable editorconfig if .dir-locals is present?

[xuhdev]

.dir-locals.el often specifies a lot of style settings (and can be more detailed). Should we disable editorconfig in this case? Or, at least, leave this option to be set in .dir-locals?

[jgkamat]

See #141 for a little more in-depth analysis.

I personally like the existing behavior (let dir-locals overwrite editorconfig values), but I think disabling editorconfig entirely when dir-locals are present seems good too. It would be nice to have this configurable too :)

[10sr]

Really sorry I totally forgot to respond to this thread.

IMO we don't need to worry about this, in most cases the values in .editorconfig and .dir-locals.el should be consistent so it should be "fixed" if values in .editorconfig and dir-locals are different.
And checking the existence of dir-locals configs will make impl a bit complicated...: what should be done if configs for c-mode is only defined in dir-locals and configs for js-mode only in .editorconfig?

AFAIK currently editorocnfig-emacs does not touch variables if they are not defined in .editorconfig (so the values from dir-locals will be used if defined), and it is OK.

[anholt]

It would be nice if editorconfig in a subdirectory still applied if .dir-locals.el isn't in that directory. Mesa has a top level dir-locals and editorconfig, and then also editorconfig in some subdirectories to override the top level behavior. Now I think the top level dir-locals overrides the local editorconfig.

[10sr]

Thanks!

It makes sense, but unfortunately it is difficult because editorconfig command cannot tell where values are defined... (for example when output of editorconfig /path/to/file is key1=value1, we cannot know if it is defined in /path/.editorconfig, or /path/to/.editorconfig)
It is true that t is possible to implement editorconfig-core.el so that it can do this, but I don't like the idea of providing features that editorconfig-core-c does not provide.