:: Re: [Dng] recommendation for consid…
Góra strony
Delete this message
Reply to this message
Autor: T.J. Duchene
Data:  
Dla: dng
Temat: Re: [Dng] recommendation for consideration: keep as close to debian as possible

> possible From:    Luke Leighton <lkcl@???>
> To:    dng@???
> Date:    Today 01:44:25 PM

>


>
> i believe you may be severely underestimating the workload that the
> current debian maintainers handle. there are over 35,000 packages,
> and i believe something like 1,000 maintainers. there are something
> like 12 ports to different architectures, and the mirrors (of which
> there are around a hundred) require something like 160 gigabytes of space
> and tend to redline whatever network bandwidth they're allowed,
> particularly during upgrades.
>

Actually, with respect, I'm not ignoring the size of Debian. I understand
perfectly well what is involved. I do admit, I could have explained myself
better.

I'm all for using their upstream work, and even establishing a good working
relationship with them for passing patches back to them. I'm not suggesting
that Devuan start over. What I am saying is that Devuan should not concern
itself with the day to day Debian's problems, such as Debian's release
schedules or packaging decisions.

In that respect Devuan needs to go its own way. The less Devuan feels
pressured to keep in sync with Debian, the easier it will be for Devuan with a
much smaller developer base.



>
> to expect even a medium-sized team to cope with 10 to 100 times the
> workload which the current debian team handle, by dropping the entire
> debian repository onto them and expecting them to be able to recompile
> it and maintain it is... i think you'd agree, completely unrealistic.


Yes, I would.

If you will humor me, I would also point out that most of the people
interested in Devuan or even Debian will be aiming for servers, not desktops.
If they want something more than that, they will probably go elsewhere,
because Debian's (and by extension Devuan's) repository is not refreshed fast
enough for their taste.

In that sense, user applications like Gimp and Libreoffice become less
important. Personally, I think that Devuan could, even possibly should,
consign them to a rolling release repo to be updated whenever Devuan has the
time.

The official Devuan release could just be the core Linux system, and a
selection of the most reliable service daemons, which is something much more
reasonable to ask from a smaller team.


Furthermore, I know the decision has been made and I am not trying to change
it, but I'd like to express the opinion that Devuan being based on Debian
testing makes things much harder than they need to be. I would have chosen to
base Devuan on Debian stable instead. Before anyone protests, please
consider that:

1. If Devuan was based on Debian stable, then Devuan would not be hostage to
whatever release schedule Debian has. If Debian Jesse is a year late - which
has happened to Debian before - then Devuan is a year late, if only by proxy.

2. As Jesse stands right now, you have to excise systemd before Jesse is even
finished, which makes doing so something of a moving target, especially if
Debian changes chains of packages upstream, making you have to start the same
process all over again.

3. With the exception of a few things, generally speaking user applications
do not use systemd and are virtually agnostic, even in binary form. They will
usually run reliably on any Linux as long as the core system libraries meet a
minimum.