I realy like the `pass prefix` idea. It's often way to much work to find the files for different programs.
-- hugo On Wed, May 27, 2020 at 07:02:46AM +0700, Allan Odgaard wrote: > On 22 May 2020, at 4:26, Nicolai Dagestad wrote: > > > If people can take a look at this and see if it breaks something we > > could > > maybe advance a little on this debat. > > The way you find the password store is inconsistent across these patches. > > Some will check `~/.password-store` even when `PASSWORD_STORE_DIR` is set, > others will not. > > Some will check `~/.local/share/password-store` even when `XDG_DATA_HOME` is > set, others will not. > > This idea about migrating to XDG Base Directory with backwards compatibility > seems messy: The great thing about `pass` is the simplicity of the password > storage, in this patch, the fish completion file now use a 12 line helper > function to find the location of where files are stored. > > I would suggest: > > 1. Write an exact specification of how one should obtain the password > storage. > 2. Add a `prefix` command to `pass` that output this directory, and use in > the helper scripts. > 3. For helper scripts that cannot call `$(pass prefix)`: write the simplest > possible code to obtain the directory (conformant to the specification) and > use that consistently across them all. > > Regarding the first item: It should also cover the case where no directory > is found, i.e. what would `pass init` use?
