Re: RFC: flavored provides, i.e. emacs25ish
On 2016-10-30 15:32 -0500, Rob Browning wrote:
> Right now a tool like dh-elpa, which needs a "viable" flavor of emacsXY
> at build time has to do something like this:
>
> Depends: ... emacs24-nox | emacs24 | emacs24-lucid ...
>
> and if emacs24 ends up staying in stretch, that becomes:
>
> Depends: ... emacs25-nox | emacs25 | emacs25-lucid
> | emacs24-nox | emacs24 | emacs24-lucid ...
>
> which is immediately incorrect if/when we add a new flavor (say the
> emacsXY-min that's been proposed).
As long as emacsXY-min Provides: emacsXY, there is no real problem, the
user can install that flavor and still have the build dependency
fulfilled.
> So I'm contemplating adding a new provides to all of the Emacs binary
> packages (e.g. to emacsXY, emacsXY-nox, and emacsXY-lucid). For now,
> let's call it emacsXYish, and given that, dh-elpa's deps would become:
>
> Depends: ... emacs25ish | emacs24ish ...
Any idea which flavor "apt-get install dh-elpa" will then install? The
whole point of listing the -nox flavor as first alternative is to ensure
that time and disk space are not wasted by installing all the additional
dependencies of the other flavors in the build environment.
Cheers,
Sven
Reply to: