FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Debian > Debian Development

 
 
LinkBack Thread Tools
 
Old 06-27-2012, 08:23 AM
Philipp Kern
 
Default duplicates in the archive

On Mon, Jun 25, 2012 at 12:38:42PM +0200, Svante Signell wrote:
> > Which wm does that? I know it isn't gnome-shell at least, as I've been
> > using it quite successfully without nm installed.
> Have you tried to use evolution without NM?

I didn't try but it only suggests network-manager. However some applications do
behave weird if you just deinstalled n-m (pidgin for instance), because they
assume that you're not connected. After a reboot (maybe dbus restart is enough)
they certainly connect again, though.

Kind regards
Philipp Kern
 
Old 07-08-2012, 02:48 PM
Adam Borowski
 
Default duplicates in the archive

On Wed, Jun 27, 2012 at 10:23:38AM +0200, Philipp Kern wrote:
> On Mon, Jun 25, 2012 at 12:38:42PM +0200, Svante Signell wrote:
> > > Which wm does that? I know it isn't gnome-shell at least, as I've been
> > > using it quite successfully without nm installed.
> > Have you tried to use evolution without NM?

Evolution seems to work just fine.

> I didn't try but it only suggests network-manager. However some applications do
> behave weird if you just deinstalled n-m (pidgin for instance), because they
> assume that you're not connected. After a reboot (maybe dbus restart is enough)
> they certainly connect again, though.

I tested a good part of Gnome today without n-m and it appears there are no
regressions at all. The only differences are:

* it gets rid of n-m icon in the systray (duh)

The dependency comes via gnome-core depending on network-manager-gnome.

>
> Kind regards
> Philipp Kern



--
I was born an ugly, dumb and work-loving child, then an evil midwife
replaced me in the crib.


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 20120708144801.GA24820@angband.pl">http://lists.debian.org/20120708144801.GA24820@angband.pl
 
Old 07-08-2012, 03:02 PM
Adam Borowski
 
Default duplicates in the archive

WHOOPS, SORRY. Meant to delete this old draft, not send it.
The issue is valid, but sorry for incomplete mail.

On Sun, Jul 08, 2012 at 04:48:01PM +0200, Adam Borowski wrote:
> On Wed, Jun 27, 2012 at 10:23:38AM +0200, Philipp Kern wrote:
> > On Mon, Jun 25, 2012 at 12:38:42PM +0200, Svante Signell wrote:
> > > > Which wm does that? I know it isn't gnome-shell at least, as I've been
> > > > using it quite successfully without nm installed.
> > > Have you tried to use evolution without NM?
>
> Evolution seems to work just fine.
>
> > I didn't try but it only suggests network-manager. However some applications do
> > behave weird if you just deinstalled n-m (pidgin for instance), because they
> > assume that you're not connected. After a reboot (maybe dbus restart is enough)
> > they certainly connect again, though.
>
> I tested a good part of Gnome today without n-m and it appears there are no
> regressions at all. The only differences are:
>
> * it gets rid of n-m icon in the systray (duh)
[was incomplete]
* "network settings" deep in the control panel will say the networking on
this system is not compatible


Since n-m breaks actually working software (udev, ifupdown) for non-obscure
uses (connecting a phone via USB, bridged setups, non-basic VPNs, etc), a
desktop environment hard-depending on it is bad, and this really needs to be
a Recommends: relationship instead.

N-M compared to ifupdown:
* makes things easier for new users (good! especially in a default install)
* is said to make wifi easier (when it works...)
And downsides:
* breaks usb0 completely (keeps raising and lowering the interface in a
loop, no apparent way to tell it to keep its grubby hands away)
* breaks a load of complex setups

"Breaks unrelated software" on the system is a RC severity, and there's no
way one can say a windowing environment is related to core networking.
Thus, I'd say, #542095 needs to be upgraded -- and changing Depends: to
Recommends: is a non-intrusive fix. It will cause n-m to be installed
unless explicitely refused, just like you want it to be.

On the other hand, breaking such setups is not a RC bug in n-m. Like any
non-core package, there is no requirement for it to be universal:
* not working with complex setups is at most wishlist
* breaking USB networking by flipping the interface is normal
It's just gnome-meta hard-depending on it what's wrong.

--
I was born an ugly, dumb and work-loving child, then an evil midwife
replaced me in the crib.
 
Old 07-08-2012, 08:13 PM
Noel David Torres Taño
 
Default duplicates in the archive

> WHOOPS, SORRY. Meant to delete this old draft, not send it.

> The issue is valid, but sorry for incomplete mail.

>

> On Sun, Jul 08, 2012 at 04:48:01PM +0200, Adam Borowski wrote:

> > On Wed, Jun 27, 2012 at 10:23:38AM +0200, Philipp Kern wrote:

> > > On Mon, Jun 25, 2012 at 12:38:42PM +0200, Svante Signell wrote:

> > > > > Which wm does that? I know it isn't gnome-shell at least, as I've

> > > > > been using it quite successfully without nm installed.

> > > >

> > > > Have you tried to use evolution without NM?

> >

> > Evolution seems to work just fine.

> >

> > > I didn't try but it only suggests network-manager. However some

> > > applications do behave weird if you just deinstalled n-m (pidgin for

> > > instance), because they assume that you're not connected. After a

> > > reboot (maybe dbus restart is enough) they certainly connect again,

> > > though.

> >

> > I tested a good part of Gnome today without n-m and it appears there are

> > no regressions at all. The only differences are:

> >

> > * it gets rid of n-m icon in the systray (duh)

>

> [was incomplete]

> * "network settings" deep in the control panel will say the networking on

> this system is not compatible

>

>

> Since n-m breaks actually working software (udev, ifupdown) for non-obscure

> uses (connecting a phone via USB, bridged setups, non-basic VPNs, etc), a

> desktop environment hard-depending on it is bad, and this really needs to

> be a Recommends: relationship instead.

>

> N-M compared to ifupdown:

> * makes things easier for new users (good! especially in a default install)

> * is said to make wifi easier (when it works...)

> And downsides:

> * breaks usb0 completely (keeps raising and lowering the interface in a

> loop, no apparent way to tell it to keep its grubby hands away)

> * breaks a load of complex setups

>

> "Breaks unrelated software" on the system is a RC severity, and there's no

> way one can say a windowing environment is related to core networking.

> Thus, I'd say, #542095 needs to be upgraded -- and changing Depends: to

> Recommends: is a non-intrusive fix. It will cause n-m to be installed

> unless explicitely refused, just like you want it to be.

>

> On the other hand, breaking such setups is not a RC bug in n-m. Like any

> non-core package, there is no requirement for it to be universal:

> * not working with complex setups is at most wishlist

> * breaking USB networking by flipping the interface is normal

> It's just gnome-meta hard-depending on it what's wrong.

*

First of all I'm not a DD but just a Maintainer of 2 packages and a long time user.

*

Since I fled away from KDE and felt into Gnome in Debian, I'm using it without N-M installed. It is only a matter of dpkg -force-depends -P two packages every time aptitude "corrects" my system when I install something, and I must say I'm more than happy by not having N-M: nothing messes with my network configuration (which is non-standard) and also users (my wife, or even myself using my normal account) can not disable networking nor break it.

*

I have not tried Evolution (I use kmail even in Gnome and my wife uses Icedove) but I can say that Pidgin works better without N-M than with it.

*

Regards (and thanks for all the time you spend that makes Debian my distro of choice)
 
Old 07-09-2012, 06:46 PM
Ian Jackson
 
Default duplicates in the archive

Adam Borowski writes ("Re: duplicates in the archive"):
> "Breaks unrelated software" on the system is a RC severity, and there's no
> way one can say a windowing environment is related to core networking.
> Thus, I'd say, #542095 needs to be upgraded -- and changing Depends: to
> Recommends: is a non-intrusive fix. It will cause n-m to be installed
> unless explicitely refused, just like you want it to be.

I definitely think this should be fixed for wheezy.

Adam earlier wrote on -devel:

I tested a good part of Gnome today without n-m and it appears there
are no regressions at all. The only differences are:

* it gets rid of n-m icon in the systray (duh)

The dependency comes via gnome-core depending on
network-manager-gnome.

To the Gnome maintainers: given that Adam has done this test, to check
that things are OK without n-m, would you be willing to make this
change to the gnome-core metapackage ?

Thanks,
Ian.


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 20475.10012.868938.877409@chiark.greenend.org.uk"> http://lists.debian.org/20475.10012.868938.877409@chiark.greenend.org.uk
 
Old 07-10-2012, 02:17 AM
Félix Arreola Rodríguez
 
Default duplicates in the archive

El lun, 09-07-2012 a las 19:46 +0100, Ian Jackson escribió:
> Adam Borowski writes ("Re: duplicates in the archive"):
> > "Breaks unrelated software" on the system is a RC severity, and there's no
> > way one can say a windowing environment is related to core networking.
> > Thus, I'd say, #542095 needs to be upgraded -- and changing Depends: to
> > Recommends: is a non-intrusive fix. It will cause n-m to be installed
> > unless explicitely refused, just like you want it to be.
>
> I definitely think this should be fixed for wheezy.
>
> Adam earlier wrote on -devel:
>
> I tested a good part of Gnome today without n-m and it appears there
> are no regressions at all. The only differences are:
>
> * it gets rid of n-m icon in the systray (duh)
>
> The dependency comes via gnome-core depending on
> network-manager-gnome.
>
> To the Gnome maintainers: given that Adam has done this test, to check
> that things are OK without n-m, would you be willing to make this
> change to the gnome-core metapackage ?
>
> Thanks,
> Ian.
>
>

A system without network-manager is still usable even for desktop users.

I mean, for example, when Pidgin opens, and n-m is not available, it
just tries to connect directly to internet. When pidgin opens, and n-m
is active pidgin waits to connect until n-m gets connected. Sometimes is
annoying because other network interfaces may be active with full
internet and pidgin waits until n-m reports ready.

A similiar experience happens with Evolution.

But, ignoring the "a desktop works fine without n-m" thing, n-m makes
more, much more easy connecting to wifi networks, espeacially for
laptops. I suggest make Laptop task depend on n-m, in this way, n-m
don't get installed on desktop systems, just on laptop systems.

--
Atte. Félix Arreola Rodríguez,
Firmado con GPG, llave 1E249EE4
 
Old 07-10-2012, 08:38 AM
Miles Bader
 
Default duplicates in the archive

Félix Arreola Rodríguez <fgatuno.123@gmail.com> writes:
> But, ignoring the "a desktop works fine without n-m" thing, n-m makes
> more, much more easy connecting to wifi networks, espeacially for
> laptops. I suggest make Laptop task depend on n-m, in this way, n-m
> don't get installed on desktop systems, just on laptop systems.

What's wrong with Recommends: in that case? It seems to perfectly
match the "makes life easier for <common but not universal use-case
XXX>" scenario you describe.

A hard package-dependency in a case like this, when there isn't
actually any hard functional dependency, and there are issues with the
depended-upon package, are decidedly user-unfriendly.

[Yeah, the various desktops tend to abuse hard-dependencies in a lot
of other ways, but ... in most cases this just results in bloat. NM
is a bit worse than that.]

-Miles

--
Brain, n. An apparatus with which we think we think.


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 878vest3i6.fsf@catnip.gol.com">http://lists.debian.org/878vest3i6.fsf@catnip.gol.com
 
Old 07-10-2012, 01:32 PM
Josselin Mouette
 
Default duplicates in the archive

Le mardi 10 juillet 2012 à 17:38 +0900, Miles Bader a écrit :
> What's wrong with Recommends: in that case? It seems to perfectly
> match the "makes life easier for <common but not universal use-case
> XXX>" scenario you describe.

Recommends is wrong for metapackages because it gets upgrades very
wrong. This is why it is used very marginally.

> A hard package-dependency in a case like this, when there isn't
> actually any hard functional dependency, and there are issues with the
> depended-upon package, are decidedly user-unfriendly.

It is unfriendly to the extreme minority of users who want a specific
selection of packages rather than the default metapackages.

Accidentally these are the users who also have the ability to make their
own package selection.

--
.'`. Josselin Mouette
: :' :
`. `'
`-


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: http://lists.debian.org/1341927163.10559.1.camel@pi0307572
 
Old 07-10-2012, 05:06 PM
Adam Borowski
 
Default duplicates in the archive

On Tue, Jul 10, 2012 at 03:32:43PM +0200, Josselin Mouette wrote:
> Le mardi 10 juillet 2012 à 17:38 +0900, Miles Bader a écrit :
> > What's wrong with Recommends: in that case? It seems to perfectly
> > match the "makes life easier for <common but not universal use-case
> > XXX>" scenario you describe.
>
> Recommends is wrong for metapackages because it gets upgrades very
> wrong. This is why it is used very marginally.

In the general case, yes. This needs to be fixed. But it's not relevant
here because squeeze had network-manager-gnome already. So we have three
scenarios:

* a new install. Recommends will pull in n-m unless explicitely rejected.

* upgrade from squeeze (or earlier sid). Network-manager-gnome will be
upgraded if present, and if it has been removed, that was not without a
cause.

* debfoster/aptitude/"apt-get autoremove". Recommends will keep n-m-g safe
from autoremoval.

The problem with recommends is that they fail to pull in *new* relationships
of an existing package, this is not what's the case here.

> > A hard package-dependency in a case like this, when there isn't
> > actually any hard functional dependency, and there are issues with the
> > depended-upon package, are decidedly user-unfriendly.
>
> It is unfriendly to the extreme minority of users who want a specific
> selection of packages rather than the default metapackages.

So you call people who want to connect a smartphone an "extreme minority"?
The last time I checked, it's hard to find a person without one. USB
cables seem to be more popular than wall chargers, and a good part of phones
can transfer data over them. It also gets you an order of magnitude better
transfer speed than wifi, and you don't have wifi everywhere.

Folks who want to connect more than one machine via a VPN are also not that
rare among Debian users. Or ones with a bridged setup. Those are more
technical, yeah, but:

> Accidentally these are the users who also have the ability to make their
> own package selection.

except that unless you sit deeply in Gnome development, you don't know which
exactly components you need. This is precisely what a metapackage is for.
Am I supposed to know by heart whether the file manager is called "nautilus"
or "caja" this week? Or what do I need to install to have clicking an image
show me its contents?

--
Copyright and patents were never about promoting culture and innovations;
from the very start they were legalized bribes to give the king some income
and to let businesses get rid of competition. For some history, please read
https://en.wikipedia.org/wiki/Statute_of_Monopolies_1623
 
Old 07-18-2012, 11:06 PM
Ian Jackson
 
Default duplicates in the archive

Josselin Mouette writes ("Re: duplicates in the archive"):
> Le mardi 10 juillet 2012 à 17:38 +0900, Miles Bader a écrit :
> > What's wrong with Recommends: in that case? It seems to perfectly
> > match the "makes life easier for <common but not universal use-case
> > XXX>" scenario you describe.
>
> Recommends is wrong for metapackages because it gets upgrades very
> wrong. This is why it is used very marginally.

Could you please explain this in more detail in the specific case of
gnome-core and network-manager ? I'm not sure I understand the
problem.

Thanks,
Ian.


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 20487.16739.688176.814402@chiark.greenend.org.uk"> http://lists.debian.org/20487.16739.688176.814402@chiark.greenend.org.uk
 

Thread Tools




All times are GMT. The time now is 07:31 PM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright ©2007 - 2008, www.linux-archive.org