[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: virtual package names for web browsers (was Re: Uninstalling Chromium)



On Wed, Jul 21, 2021 at 11:45:07AM +0100, Jonathan Dowland wrote:
> On Wed, Jul 21, 2021 at 01:34:50PM +0300, Reco wrote:
> > One would think that gnome-www-browser virtual package would fit such
> > role perfectly. I mean, if GNOME DE has some special requirement for a
> > browser, and Debian already has such aptly named virtual package -
> > surely it can be considered as a suitable dependency?
> 
> It'd be confusing for people not using GNOME.

I lost you here.

A user installs GNOME DE, one way or another. A user gets a browser,
which is suitable for GNOME DE. That's the way gnome-core currently
works, there's nothing wrong here.

Another user does not install GNOME DE, but installs a browser, which
just happens to be either firefox-esr, chromium or epiphany-browser.
First two are popular, popcon should show this.

Currently (as of stable) firefox-esr, chromium and epiphany-browser
provide gnome-www-browser. gnome-core directly depends on either on
these three packages.

Hence the GNOME user will get at least one of these packages, and can
even install all three at the same time.

Non-GNOME user will get gnome-www-browser by installing either of these
three packages, because of Provides. And if non-GNOME user wants this,
a user can install even all three at the same time.

Again, the current stable behaves like this, so I do not understand how
exactly replacing dependency on gnome-core will cause a confusion for
non-GNOME user. I'd expect some breakage for GNOME users, but I cannot
imagine where exactly it could happen.
After all, Provides merely does what it's supposed to do, it does not
force additional packages on a user.


> This is an area of interest for me (virtual package names, what Policy
> dictates, how we describe what they mean, semantically; how we do so in
> a way such that we can check their usage in the archive mechanically,
> etc.) so I might try to pick up my work on improving it post-bullseye.

As always, any attempt on improving Debian is welcome, regardless of the
outcome. It's not like Debian needs an improvement IMO, but
nevertheless.

Reco


Reply to: