Hi! On Fri, 2018-11-23 at 13:09:13 -0500, Chris Lamb wrote: > [Adding guil...@debian.org to CC] > [Adding 473...@bugs.debian.org to CC]
> > In order for this bug not to sit idle for yet another five years, > > attached is my attempt at a patch. > > Alas, I don't think that this will merged simply because it hard- > codes vimdiff. Exactly. > I tried applying one of the configurable variants > of patch from: > > https://bugs.debian.org/120152#16 > > ... with some success but I am a little bit unsure about the > approach and, of course, that code is now extremely old, crusty > and as you can see it had some issues after review. > > Guillem, would you be amenable to applying any configurable patch > if one was to magically appear? If so, would you have any other > requirements or objections, either philosophically and technically? So, I think there are several related problems that have been conflated in this mass of merged bugs, with different requirements, which I'm now not sure have been spelled out adequately. Let me try to quickly clear this now, and during this next week I'll untangle the mess and probably try to salvage/prepare a patch for the first one. - There is a desire to make the diff generation tool configurable. This would ideally require, being able to pass a command-line with some placeholders for the old and new filenames. Or assume that the user will use a wrapper that honors the DPKG_CONFFILE_<type> environment variables, which would be slightly more cumbersome, but require less code. - There is a desire to get 3-way merge support. This is currently just unreliable, as we do not keep all the information required. I'd rather not spend time on this until the conffiledb work is merged, which I just need to sit down and do. :) Thanks, Guillem
