Need DHCP client scripts for NIS

[Paul Smith]

Yesterday morning I filed a bug https://bugs.launchpad.net/bugs/246322
requesting that the NIS package include a dhclient-exit-hooks.d script
to set up NIS based on the nis-domain and nis-servers (and
nisplus-domain and nisplus-servers) DHPC options.  All the other
enterprise Linux distros I'm familiar with (Red Hat and SuSE in
particular) include dhclient-scripts which do this automatically
already.  This is a critical component of deploying Linux in many
enterprise environments.

I guess I'm not used to how Ubuntu manages things like this: all the
F/OSS projects I'm involved in, if they have a bug tracker, tend to
manage requests like this via the bug tracking system to make them easy
to remember/find/track/maintain.

However, my bug was closed as "Invalid" and the person closing it
suggested that I find some appropriate mailing list on lists.ubuntu.com
to discuss it.

Looking through the lists this seems like the most relevant one.

So... what now? :-)

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Need DHCP client scripts for NIS

[Paul Smith]

On Tue, 2008-07-08 at 11:15 +0100, Mark Brown wrote:
> Note that this can't be enabled by default on upgrades since doing
> that may break existing installations by overwriting existing
> configuration. It also needs to be optional for new installations for
> similar reasons, probably doing something similar to what the samba
> package does for WINS.

I'm not sure whether it's better to discuss this in bug comments since
the discussion is then easily tied to the request.  I added a note there
saying I'm not so sure about this.  All other distros I'm familiar with
contain a dhclient-script that handles this automatically and it's
enabled by default (on Red Hat for example you can disable it by adding
NISPEERS=no or similar to something in /etc/sysconfig).

In order for overwriting to occur, the following must be true: the user
must have installed the NIS package (it's not installed by default on
Ubuntu), and the user must use DHCP for their network settings, and the
DHCP server must be configured to provide nis-domain and nis-server
settings as options (this is rare since only larger enterprises with
lots of UNIX systems bother with NIS), and finally the user must, for
some reason, want to use a different value for either/or the NIS domain
and the NIS servers than the one the DHCP admin suggested.

Those are a lot of "ifs" and I really don't believe they justify a lot
of complicated configuration and potentially confusing questions on
install.  I don't have any problem with a way to disable it if the user
wants to (a setting in /etc/defaults/nis maybe?  Or maybe they just
remove these entries from the dhclient.conf request?

There is a change in behavior, potentially, but Ubuntu changes behaviors
every release.  Finally, remember that the dhclient exit hook preserves
the user's setup when the lease is obtained and restores it when it is
released.  Typically the replacement configuration (in this case
the /etc/yp.conf file) has a comment saying it was replaced by the
dhclient script based on DHCP server settings, so users will know what
happened.

> For this to be fully useful there would also need to some facility for
> adding the NIS entries to the passwd, group and shadow files during
> installation.  This is outside the scope of the NIS package since it
> affects the global configuration of the system.

I don't think any change is needed here.  The base system nsswitch.conf
uses "compat" for the passwd, group, and shadow entries which is fine.

All I've had to do to get my Ubuntu systems to be on par with my Red Hat
Enterprise systems (in this respect) is add the dhclient exit hook and
the dhclient.conf entries, and it works well.

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Need DHCP client scripts for NIS

[Paul Smith]

On Tue, 2008-07-08 at 16:23 +0100, Mark Brown wrote:
> > In order for overwriting to occur, the following must be true: the user
> > must have installed the NIS package (it's not installed by default on
> > Ubuntu), and the user must use DHCP for their network settings, and the
> > DHCP server must be configured to provide nis-domain and nis-server
> > settings as options (this is rare since only larger enterprises with
> > lots of UNIX systems bother with NIS), and finally the user must, for
> > some reason, want to use a different value for either/or the NIS domain
> > and the NIS servers than the one the DHCP admin suggested.
> 
> The main use case this breaks is people using NIS to distribute
> information to laptops which then work off-line, potentially on other
> networks (I am aware of people doing this, usually with a slave
> configured on the laptop for disconnected operation).

I'm not sure I see the breakage, unless you mean you don't think such a
user would want to connect to the master server while on-line.  If
that's OK, then the user-configured /etc/yp.conf would specify the local
slave, then when the system was on-line DHCP would provide a new server
(the master or another slave) instead, then when the system went
off-line the local slave info would be restored to /etc/yp.conf.

ypbind would have to be restarted of course but that's true in all
cases.

> This is what low priority debconf questions are for - they're there if
> they are needed but not enabled by default so they don't confuse users
> in the normal course of affairs.

This sounds good to me.

> > There is a change in behavior, potentially, but Ubuntu changes behaviors
> > every release.  Finally, remember that the dhclient exit hook preserves
> 
> The reason for being especially conservative here is that breaking the
> NIS configuration can render the user unable to log in to the system -
> the consequences of getting it wrong are extremely serious.

True, but it takes a lot of customization to get there.  There's no way
to integrate NIS with the initial Ubuntu account since Ubuntu doesn't
allow you to choose their own UID at install time.  You're basically
stuck with a massive "chown" operation after the fact (which gets pretty
tricky when you throw in odd things like gvfs FUSE filesystems in $HOME
etc.), or else you create a local user account with a different username
then configure your NIS account to be an admin--in this case you always
have that local user account to fall back on (of course, people with UID
1000 won't be able to log into your box).

I do agree that losing your account is a big problem (I lost sudo access
due to the Hardy bug regarding sudo and hostnames--had to boot from CD
to recover.  That sucked.)  I am still skeptical about the _real_
threat, though, since if you are using NIS as your sole login to the box
you're really playing with fire anyway.  Nevertheless it's better safe
than sorry.

> > I don't think any change is needed here.  The base system nsswitch.conf
> > uses "compat" for the passwd, group, and shadow entries which is fine.
> 
> compat is only half the story - you also need to add the appropriate
> + entries to the relevant files for it to have any effect.

Oh right.  I wonder why Debian/Ubuntu uses "compat" instead of "files
nis" like most other distros?

Don't get me wrong, I like "compat" as it's more powerful than "files
nis".  But, as you point out, changes are required to base files in
order to enable it.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

My personal hopes for Intrepid

[Paul Smith]

I realize that Intrepid is already well on its way and it's late for any
change in direction.  However, I still would just like to get my hopes
for Intrepid out there.  Maybe they'll have to wait for Jazzy Jackrabbit
but hopefully not.

My main goal: Enterprise Readiness.  I want a great desktop that I can
start to seriously convince people to migrate to at work, away from Red
Hat Enterprise.  RHEL is a fine distribution but as a desktop it's
simply too dated (even the Workstation versions).  On the other hand,
they really "get" the enterprise requirements and their systems are very
simple for IT departments to deploy.  For the most part they just work.
I don't want Ubuntu to copy RH etc.; I want us to go beyond them!

Note I don't care if I have to install a bunch of extra packages.  I
realize the CD has limited space and it needs to be geared towards a
wider audience; that's fine.  I can give people an "aptitude install"
line they can cut and paste, or whatever.  What I _don't_ want is a
bunch of "after-market" customizations made by editing config files by
hand, etc., which is what we have to do today.

1. Proper multilib
By "proper" I mean LSB compliant, so it's portable/interoperable with
all the other major distributions of Linux (RH, SuSE, etc.)  Only Debian
and, by extension, Ubuntu get this wrong.

2. Switch to autofs5
It's more than time to make this switch.  For Hardy, this version isn't
even packaged!

3. Fix NetworkManager
I've ranted about this elsewhere and I'm tired, but if we insist on
modifying every network-based application on the system to work with NM,
rather than fixing NM to work well with all the other applications, then
lets get on with it already!  For example, autofs is still utterly
unreliable with NM in Hardy if your maps are distributed via network
protocols.
https://bugs.launchpad.net/ubuntu/+source/autofs/+bug/213574

4. Fix DHCP client
The DHCP client in Hardy is missing a number of important DHCP options
for proper enterprise readiness.  Bugs are filed:
https://bugs.launchpad.net/ubuntu/+source/dhcp3/+bug/74164
https://bugs.launchpad.net/ubuntu/+source/dhcp3/+bug/246343

5. Fix NIS for DHCP
NIS currently can't be configured by DHCP.  More and more enterprises
are moving to DHCP-based configuration for all clients, and Ubuntu is
behind other Linux distros in this area.
https://bugs.launchpad.net/ubuntu/+source/dhcp3/+bug/246322


Then there are a few other things which I'm sure are being worked on:
for example the fact that you can't browse Samba servers that require
login anymore in Hardy, like you could before (some deficiency in the
new Gnome vfs backend or something?).

Anyway that's my personal wishlist.  I'm willing to propose patches for
the bugs in #4 and #5, although for #5 I can only produce the hook
scripts, while the comments seem to imply that more detailed dpkg work
is needed for a complete solution.

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Can Evolution calendar reminders pop-up automatically?

[Paul Smith]

On Sat, 2008-07-26 at 05:57 -0700, Tim Zakharov wrote:
> Have Ubuntu developers modified this behavior as someone on the
> Evolution mailing list suggested?  Can someone help me get Evolution
> to behave in the manner I am looking for?

I think the behavior you're seeing is the default behavior for
Evolution.

If you want a direct popup, I think you need to install separate
packages for that on Ubuntu.  Maybe those other distros install those
packages by default.

Try using Synaptic to add the "mail-notification" and
"mail-notification-evolution" packages.  Note that these are part of the
Universe repository so if you haven't enabled that, you'll need to do
so.

If you have problems let us know.

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: OpenChange

[Paul Smith]

On Thu, 2008-08-07 at 14:52 -0600, Kevin Fries wrote:
> We are in the process of changing our Exchange server here to 2007.
> This breaks the Evolution in Hardy.  There appears to be a fix and it
> appears to be a plug-in as part of the OpenChange project.  It looks
> like it is supposed to make it into Evolution 2.24, and Fedora is
> already reporting that it will be part of version 10.  Does anyone
> know if this is going to make it into Intrepid?

One of the basic tenets of Ubuntu was that it ships with the latest
version of Gnome.  In fact, the release schedule of Oct/Apr was chosen
mainly based on the Gnome release schedule, to allow Ubuntu to be one of
the first distributions out the door with each new version of Gnome.

All that to say, Intrepid will ship with Gnome 2.24, so if this plugin
makes it into Gnome 2.24 it will be in Intrepid.

Personally, following the Evo mailing lists, I'm not confident that this
plugin WILL be ready for 2.24 (note that it relies on libraries provided
by the Samba folks, so it's not all up to Gnome devs).  But, you never
know!


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Bugs introduced in GCC?

[Paul Smith]

On Mon, 2008-08-18 at 18:18 -0400, Andrew wrote:
> > Are the statements below true?
> > Is the Ubuntu patch really 5 MB?
> > If so, what is it meant to change?
> 
> The diff  of the current gcc in Intrepid ,as found here [1], seems to
> only be 672 KB
> 
> gcc-4.2.3's [2] seems to be 1.1 MB

You're looking at the compressed size.  Uncompressed, this diff is
actually 5.2M.  About 4M of that is patches to GCC itself (the rest is
patches for the Debian packaging files).

There are 138 individual patches in that 4M.  Not sure where they came
from.

Looks like this was reported on Launchpad a week ago:

https://bugs.launchpad.net/ubuntu/+source/gcc-4.2/+bug/256797

and it was confirmed, but there's been no other response.

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Backtracing, Invalidated Bugs and Quality

[Paul Smith]

On Wed, 2008-08-20 at 16:06 +0200, Markus Hitter wrote:
> please stop this marking-as-invalid-mania.

+1.  It's a bad idea to hide problems, even ones that cannot be
replicated.

On the other hand I do think it's worthwhile to somehow mark bugs which
are not sufficiently documented as to be reproducible/fixable.  The
reality is, as Markus says, there are a LOT of bugs out there.  Should
we spend precious developer resources working (almost invariably
futilely) for hours to try to reproduce a bug like this, or should we
spend the same time fixing a bunch of bugs where we do have sufficient
information?  The answer is obvious.

So, if there is no suitable bug state existing already we need a new
state for these kinds of bugs.  We can call it "watching", or
"insufficient information", or "can not reproduce", or whatever.  But it
should not be "invalid"... or at least not have the current behavior of
invalid, where once the bug is marked that way it, and all dups of it,
drop off the edge of the world.

Also, it seems to me that if you tell someone their bug is "invalid"
that doesn't inspire them to come back with more information or send
more bugs in the future.  On the other hand, if you mark the bug as
"need help" or similar, then they know what they need to do next time.
It's more encouraging than discouraging.


My $0.02.

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Rejecting the Firefox EULA in Ubuntu

[Paul Smith]

On Wed, 2008-09-17 at 14:14 +0100, Sean Hodges wrote:
> It seems no'one has looked into this yet, so I will have a look this
> weekend.

There has recently been an announcement from Mozilla that they're going
to fix this.  See, for example, the coverage in Groklaw (even if you
don't like Groklaw they do have links to blog postings etc.)

The discussion about whether we want to keep FireFox with the attendant
risk of "brush fires" such as this one should continue, but I don't
think we need to rush out and adopt Iceweasel or whatever before
Intrepid ships (assuming Mozilla actually does what they say they're
going to do: fix the problem).

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Ubuntu and language packs

[Paul Smith]

On Thu, 2009-02-05 at 15:47 -0500, Mackenzie Morgan wrote:
> On Thu, 2009-02-05 at 20:57 +0100, Surfaz Gemon Meme wrote:
> > If they know English, do not need English support of Ubuntu.
> > If they do not know English, I do not think the English support of
> > Ubuntu could change something.
> 
> There are different values of "know."  They might know enough English to
> file a bug saying "Firefox crashed" but then the logs they post will be
> in, for example, Croatian.  Most triagers don't speak Croatian, so we
> can't read the logs.  And the reporter might not know enough English to
> translate all of the tech jargon in the logs, besides the amount of time
> it'd take.  At least the icons on the buttons can help when they use an
> English version to gather English logs.

I thought I'd just post this so everyone is on the same page WRT
internationalization in GNU/Linux systems.  For those that already know,
sorry for the noise :-)

GNU/Linux uses the gettext package for i18n.  There are other packages
used by other UNIX systems, which use large catalogs of numbers to look
up the right translations for each string, but this is a big pain and is
especially painful for distributed development/translation teams such as
you're likely to have in free software.

Gettext uses a really interesting trick to avoid the need for predefined
catalogs, at the expense of some runtime effort.  Basically, it uses the
string to be translated itself as the key for looking up the translated
string.  At runtime, when a translation is needed, the gettext library
is invoked with the static string to be translated as the argument.  It
generates a hash based on the contents of that string, then looks up the
hashed value in the translation catalog.

If there's an entry for that hashed value, then it is returned to the
caller.  If there is no entry, then the string that was provided is
returned.

What does this mean for the current discussion?  It means that the
"default" string is encoded right into the program.  So, even if NO
language packs are provided, all gettext programs will still print the
default string.  This default is typically called the "C" locale, for
hysterical raisins.

Virtually all default strings are written in English.  So, if you're
happy with the default strings you really don't need to install any
language packs whatsoever.  And certainly, if you want to use a
different locale than English, you can get away with ONLY that locale
being installed; if you don't have a satisfactory translation of the
string in your chosen locale it will use the default string... and
that's in English.  Or if it isn't it's definitely in whatever language
the developers of that utility wrote the code in :-).


IOW, there's really no reason to specifically install the English
language packs, if you don't want to use a locale of en_US or whatever.
Certainly you don't need to do it for any sort of debugging purposes.
If you want to force "no locale" to generate screenshots or whatever you
can just use "LC_ALL=C firefox".



-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Removing single program from multi program packages

[Paul Smith]

On Mon, 2009-03-23 at 08:53 -0400, Mike Jones wrote:
> If Ubuntu said "Ok Boss" and pretended to remove one specific
> program from a package for me, through the single "add or remove
> programs to your system" interface of add/remove, that would be...
> better, but not really the best solution due to the inherent
> dishonesty. 

I think this is not better, at all.  As others have mentioned if you
want items to be removed from your menus then you can do that already
with the menu editor.  Having your computer lie to you is, IMO, much
more unpleasant than having it truthfully tell you "I can't do that for
you".

> Is there just no way for a package maintaner to not have extra
> work piled on their already hefty load while at the same time we allow
> a user of Ubuntu to remove most traces of a program in a package with
> multiple programs without having to also remove the rest of them? Is
> it worth doing even if its possible? I think I'm in a somewhat unique
> position of having extreme distaste whenever my system tells me I
> can't do something in a counter intuitive way. It telling me it can't
> make my chair float on an anti-gravity sled is fine... it telling me
> that if I don't want one program of a package, I'm not allowed to have
> any of them in the package... thats not so great.

Of course you can remove anything from your system that you want.  You
have complete and absolute control over the contents of your disk,
should you choose to exercise it.

The package manager manages packages, it doesn't manage individual
files.  However, there are plenty of tools on a Linux system that do
manage individual files, such as cp, mv, and, most interestingly for
you, rm.

So, if you have specific files that you don't want to appear on your
system, simply use a shell and "rm" them, or use the file browser to
remove them if you prefer a graphical interface (you will need to get
administrative privileges before either of these will be successful).
The package manager has no say in this.

Of course, if at the end of this you discover you have a broken
system... well, then, as the saying goes: you get to keep both
halves :-)

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Why Ubuntu is not ready for prime time

[Paul Smith]

On Thu, 2009-08-27 at 19:23 +0300, Dotan Cohen wrote:
> > I guess Jonathan Taylor is joking or trolling here, one of the main
> > advantage of using Linux distributions for desktops is that every
> > application is automatically categorized where, on Windows, every
> > application is NOT categorized...
> >
> > Start->Programs->Adobe->Photoshop v.x->Photoshop
> >
> 
> How is one supposed to know to look under "Adobe" for Photoshop? Why
> isn't it under "Graphics" or "Photo Editors" or some such menu?

Exactly.  That's an example of how Windows organizes its menus (in case
you didn't notice the "Start" menu reference), and it's obviously bogus.

The other example (that you clipped) is how Ubuntu does it, which is
much simpler to understand.

Yet the OP raised menu organization as a thing that Ubuntu does _badly_.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Why Ubuntu is not ready for prime time

[Paul Smith]

On Thu, 2009-08-27 at 18:48 +0200, Vincent Arnoux wrote:
> On Thu, Aug 27, 2009 at 18:23, Dotan Cohen wrote:
> >> I guess Jonathan Taylor is joking or trolling here, one of the main
> >> advantage of using Linux distributions for desktops is that every
> >> application is automatically categorized where, on Windows, every
> >> application is NOT categorized...
> >>
> >> Start->Programs->Adobe->Photoshop v.x->Photoshop
> >>
> >
> > How is one supposed to know to look under "Adobe" for Photoshop? Why
> > isn't it under "Graphics" or "Photo Editors" or some such menu?
> 
> After the application is installed, a bubble notification guides you
> to the newly installed program by highlighting the path in the start
> menu. This part would be nice to see in our DE's.

This enhancement would be nice, but I'm sure you'll agree it's in no way
comparable to having sane menus in the first place.  It's great the
first time you install something but what about an application you use
only once in a great while?  The second time you need it the
highlighting will be long gone, and yet you can't remember where in the
heck that program went to!

Also, Windows has it's own fair share of programs whose names are not
very evocative.  If it wasn't one of the most well-known programs in the
world would you guess that "Quicken" was a program for handling your
finances?

Cheers!


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

RE: Why Ubuntu is not ready for prime time

[Paul Smith]

On Fri, 2009-08-28 at 18:30 +0100, paul.mcma...@met.police.uk wrote:

> Quicken, Outlook, Powerpoint, they're all bizzare names.
> Doesn't mean that Ubuntu can't do one better ;-)
> 
> 
> I believe the menu system is better in Ubuntu than Windows, however
> all apps could do with having their function added to their name.
> As like "Firefox Web Browser". We know what Firefox is, but in my
> office (still working on Windows 2000!) I don't think I'm the
> majority.


I'm not disagreeing with your statement, I just think that your wish is
already largely a reality.  I took a look through my menus (9.04) and as
far as I can tell ALL the standard applications have their function
shown, either by itself ("Terminal") or in addition to the name
("Rhythmbox Music Player").

There are some apps I've installed from multiverse, etc. that are not
standard parts of Ubuntu where this pattern is not followed (e.g.,
"Emacs 22") but things you've explicitly installed yourself are less
critical in this respect.

If you find apps where this isn't the case I think filing a bug in
Launchpad is perfectly reasonable.
-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: PulseAudio Managers

[Paul Smith]

On Mon, 2009-10-19 at 23:50 -0400, komputes wrote:
> >> And has anyone seen this little blurb from Lennart? Did Ubuntu really
> >> f**k this up or 
> >>
> >> http://0pointer.de/blog/projects/pa-in-ubuntu.html
> >   
> Hasn't this been fixed for over a week?
>
> + 0053-fix-sigsegv-module-bluetooth-device.patch: Don't strcmp
> uninitialized memory (LP: #437293
> )

This is actually the patch that Lennart calls "an outright insult"...
and based on the link in his blog he's absolutely right; that change is
ridiculous and, as a maintainer of a free software package myself, if
someone modified my code like that in a major distribution like Ubuntu
I'd be P.O.'d as well.  I'm sure Daniel does great work most of the
time, but this change looks like a dud to me.




-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: PulseAudio Managers

[Paul Smith]

On Tue, 2009-10-20 at 00:39 -0400, Daniel Chen wrote:
> On Tue, Oct 20, 2009 at 12:14 AM, Paul Smith  wrote:
> > time, but this change looks like a dud to me.
> 
> Clearly the right way to debug this is to comment out the patch in
> debian/series and see why pa_streq() is being passed crap. Anyone with
> bt hardware and valgrind up for it?
> 
> http://launchpadlibrarian.net/32493850/Stacktrace.txt

The question is what's the point of the patch at all?  If the contents
of address are bad, what's gained by allocating memory, copying the bad
contents into it, and comparing that (then freeing the memory)?

Either the address pointer itself is pointing into bad memory, in which
case the act of sprintf'ing it into the newly allocated memory will fail
just like the strcmp does, or else the string that address points to is
"bad" in some way, in which case what's the point of making a copy of
it; the copy will just be bad the same way anyway.

Plus, unless pa_sprintf_malloc() can handle NULL pointers properly
(possible, I didn't check) this patch actually INTRODUCES a bug that
wasn't present in the original code, by not testing address for NULL
before using it.

The patch should be reverted and the bug should be reopened: this change
has no chance of solving the problem.


Looking at the stacktrace above, it seems that address is a perfectly
legal pointer, pointing to the string "(null)" (not a NULL pointer!)
Maybe there was some confusion about this.  It seems that this string
was generated in the pa_hook_fire() function.  Maybe someone passed a
NULL pointer to a sprintf() variant here; on some systems (Solaris for
example) if you pass a NULL pointer to *printf() for a %s character,
rather than dumping core, it just prints a token like "(null)".
Personally I think this is a stupid idea; I'd much rather get a core
dump I can debug than have random data generated by my code with no
errors.

Since address is known to not be NULL, that must mean d->address is NULL
since that's apparently why the core dump was generated.  Since the
prior if-statement seems to ensure that d is not NULL, that means that
d->address itself is NULL.  I have no idea how/why that might happen.

If you want a change that will inhibit the core dump, changing the test
from:

if (address && !(pa_streq(d->address, address))) {

to:

if (address && d->address & !(pa_streq(d->address, address))) {

would help a lot more than the current change, as far as I can see.

Cheers!


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Ubuntu Domain Server

[Paul Smith]

On Thu, 2009-10-22 at 11:56 +0800, Christopher Chan wrote:
> > It doesn't matter how much work is involved. Do you think the 
> > Linux/Ubuntu community would be willing to change the way system 
> > logons work if it meant bug #1 could be completed?
> 
> Let us see. To change the way system logons work would mean changing 
> pam, the C library and just about anything that has to do with system 
> accounts. You are welcome to try to convince the Ubuntu community to 
> maintain a fork of all these essential system libraries and offer some
> form of backwards compatibility to avoid having to also modify who
> knows how many other packages like sendmail, apache,
> bind, ..., ..., ..., everything.

You guys need to step back a bit.  There's absolutely no reason whatever
that this _feature_ cannot be implemented on UNIX/Linux.

Yes, obviously the _implementation_ that relies on changing the UID/GID
scheme is a complete non-starter and cannot even be considered.  There's
no chance that anyone "upstream" will be willing to break that behavior
and as you say, Ubuntu cannot essentially rewrite the entire GNU/Linux
operating system to do away with it (don't forget that UID/GID is
heavily embedded in the kernel, too, so Ubuntu would have to rework the
kernel itself extensively).  If this is Ryan's question then the answer
is definitely no, not even if it meant bug #1 could be completed.  Let's
all remember our goal here is NOT to beat Microsoft by becoming a free
version of Windows.  Our goal is to produce a better product, while
still staying true to the UNIX roots and philosophy (which we believe
will lead to better software).

However, luckily for us we do not HAVE to change or do away with UID/GID
in order to implement automatic joins of a workstation.  There's
absolutely no reason that user "paul.smith" cannot have UID 1000 on one
system and UID 2000 on another system: you just need to implement a
mapping mechanism.

But there are so many things to be considered before you even get here
that impact directly on this.  For example, obviously security is
critical and so you'll need a secure way to do AAA.  How do you add
users?  How do users authenticate?  Etc. etc.  All critical questions.
Most likely you will need to base this on Kerberos, just because there's
nothing else out there with the requisite features + security, that I
know of anyway.

Once you have that figured out you must end up with some secure token
which represents a user that you can present to other systems as proof
of identity.  Then all you have to do is have each host map that token
to a locally relevant UID/GID.  UID/GID cannot be used between hosts,
anyway, in any secure fashion.  That's just one idea.


I'm certainly NOT saying it's not a lot of work.  I'm saying that it can
be done, and it doesn't require throwing out 30+ years of UNIX/POSIX
history to do it, so let's not dismiss the big idea based only on one
possible bad implementation.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: karmic trashed in Tomshardware.com

[Paul Smith]

On Tue, 2009-12-08 at 11:07 +0800, Tim Hawkins wrote:
> I dont know if it is relevant, but looking at the specs of the two
> "unstable" machines listed in the article, both had nVidia chipsets 
> 
> The one machine (netbook) that they said worked perfectly had an Intel
> Graphics chipset.  
> 
> Given the amount of discussion relating to problems with the nVidia
> drivers in Karmic, could this be a factor in this review.?

One never knows, of course, but FYI I have one system at home with an
Intel graphics chipset, and two systems at work both with Nvidia cards
using the proprietary drivers (both have dual monitors attached).

All three work just fine in Karmic, and have from the beginning (no
upgrade issues to speak of).

We've upgraded probably about 15 systems at work and only two problems:
one was that the screensaver enabled during the upgrade and, since that
person was mounting their home directory over NFS and it tried to
restart NFS, badness happened so that we couldn't get rid of the
screensaver by hand, and it was putting up a dialog asking about
overwriting some file.  I had to C-A-F1, login there, and kill it; the
upgrade finished just fine from there.  The other I'm not sure what
happened: they did it while I wasn't around :-)


I did see a problem on my system at home (Intel graphics) when I had the
screensaver set to "Random": my kernel panicked twice in one day while I
was away from my desk.  I switched back to "Blank" which is what I
always use anyway, and haven't seen a single glitch since.  I'm assuming
there're one or more bugs still in the Intel driver WRT GL graphics or
similar.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: is anyone ever going to fix this major bug?

[Paul Smith]

On Tue, 2009-12-08 at 12:30 -0800, Brendan Miller wrote:
> This bug more or less makes apt-get unusable on 64 bit systems and has
> been in for a long time now:
> 
> https://bugs.launchpad.net/ubuntu/+bug/402833

Is this still a problem in Karmic?  I don't have that file on my system,
and a search of packages available doesn't give any package named
"ia32-libs-tools".

Not saying it shouldn't be fixed in Jaunty, I'm just wondering if
there's an issue in Karmic as well--if so I think it must be different
(different package names, etc.?) than the Jaunty issue.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Bug#559761: ITP: release -- provides information about the current releases

[Paul Smith]

On Mon, 2009-12-07 at 00:14 +0100, Benjamin Drung wrote:
> Package: wnpp
> Severity: wishlist
> Owner: Benjamin Drung 
> 
> * Package name: release
>   Version : 0.1 (native)
>   Upstream Author : Benjamin Drung 
> * License : GPL v3+
>   Programming Lang: Python
>   Description : provides information about the current releases
> 
>  This package contains information about all releases of Debian and
> Ubuntu. The
>  release script will give you the codename for e.g. the latest stable
> release of
>  your distribution. To get information about a specific distribution
> there are
>  the debian-release and the ubuntu-release scripts.

I wonder what the difference is between this and the existing, standard
lsb_release command.

I'm not saying we shouldn't create a new package, if lsb_release is not
sufficient for some reason.  I'm just not sure what the reason is.  Can
someone write up a few sentences about what this package does that makes
it necessary, instead of using lsb_release?

Cheers!


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Evolution & Ubuntu 10.04 LTS

[Paul Smith]

On Fri, 2010-03-05 at 11:23 +0200, Peteris Krisjanis wrote:
> Well, as far as I have heard from developers, Staying with Evolution
> 2.28 decision was made because 2.30 will have too big sweeping
> changes, like D-BUS instead of Bonobo, etc. So it is too much for LTS.

What?!?!  Is that really true?  This is the first I've heard of this!
Evo 2.30 is an integral part of Gnome 2.30--it's bogus to try to leave
it behind.

Evolution is at the same time one of the most important applications for
deployment of Linux in a typical corporate environment (read:
Exchange-based), and also one of the most problematic historically.
That means we need to push _forward_, though, not hang back.  There are
real problems with 2.28: I can't believe anyone would prefer to stay
with that over 2.30 regardless of LTS.

Not all those problems are fixed in 2.30 but the Exchange MAPI support
in 2.29 is far ahead of 2.28, and the new capabilities added in
2.29/2.30 for other backends solve lots of bugs and clean up all kinds
of issues.  Plus, if Ubuntu moves to 2.30 they will get the advantage of
fixes made for 2.30.1, 2.30.2, etc.  2.28.x will be dead (is already
dead, from Gnome's perspective) and backporting changes across that
barrier will be extremely difficult for exactly the reasons quoted above
(sweeping changes).  Does Ubuntu really want to get stuck with broken
Evo for the entirety of the LTS?


Have the decision makers been following the development lists and trying
the new version?  Surely they must have some factual basis for such a
decision, in terms of experienced instability, rather than just running
away from the bullet list of changes.  However, I don't see any problems
and I've been building Evo 2.29.x from the latest git source every few
days and using it "in anger" on all of my systems for daily email (and I
get/send a LOT of email), with both IMAP and MAPI, for the last 3 months
or so.  It works MUCH better than 2.28.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Evolution & Ubuntu 10.04 LTS

[Paul Smith]

On Fri, 2010-03-05 at 17:17 +0100, Sebastien Bacher wrote:
> The current version of evolution is not broken and probably not over
> an unstable version which got some much rewrital and so few testing.

Sorry, but that's wrong: 2.28 is unquestionably broken.  To quote you,
"Glad that it works for you it doesn't mean that everything is bugless".

The Exchange MAPI support simply does not work, except maybe for the
most trivial cases.  Color me unimpressed with statements to the
contrary: within a few days of using it the first time last fall I ran
across a bug (which I discovered already filed in bugzilla for months,
and not fixed yet!) that caused my entire INBOX on the Exchange server
to be completely deleted.  I had to request from IT that it be restored
from backup.  Good times.

That one got fixed (eventually) but it was just the first one I ran
into, and those are just the bugs.  There are MAPI _features_ that
simply don't work in 2.28, especially related to calendaring.  I'll tell
you how broken it is: I still run Outlook in Crossover rather than Evo
2.28--and I loathe Outlook more than just about any software I've ever
used.

> Glad that it works for you it doesn't mean that everything is bugless.

That's a straw man.  The question isn't whether it's bugless or not: the
question is whether it's better than Evo 2.28 or not.

> The Ubuntu team decided to stay a known version because jumping on a
> rewrite not done yet would be risky and would requiring having
> ressourcing working chasing bugs and updates for that new version.

Have the Ubuntu devs actually tried using the new version?  Have they
run into problems?  Has anyone on the Ubuntu team contacted the Evo
developer's lists to ask their opinion and discuss the stability with
other users of 2.29.x?

Is the decision based on anything more concrete than reading the
roadmap?

> You can also note that the next RHEL version will use 2.28 and that
> Debian might be doing so in their coming stable, it means this version
> will keep being stabilized and worked.

Let's be honest.  No one is going to follow the RHEL packaging of Gnome
2.28 and be porting Evolution fixes into the Ubuntu repository.  Ditto
for Debian.  And certainly the Gnome devs won't be making new releases
upstream, although Ubuntu rarely packages them anyway.

In my experience (using Ubuntu since 2006 or so, including LTS versions)
only the most egregious bugs get addressed in the LTS and of course, no
new features will be available.  Anything that was broken in 9.10 as
well will not be considered a regression, and so won't be addressed at
all unless it's a security fix.



-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Evolution & Ubuntu 10.04 LTS

[Paul Smith]

On Mon, 2010-03-08 at 23:02 +0100, Sebastien Bacher wrote:

There are plenty of stability problems with 2.28; I subscribe to the Evo
mailing lists and people are constantly posting about this or that
database corruption, missing mail, etc. etc. (note this is with standard
IMAP or POP support, not Exchange--either OWA or MAPI).  The general
consensus seems to be that ever since the sqlite change back in 2.24
there have been definite issues, and these have not yet been worked out
in 2.28 either.  I think the 2.28 glasses are a little rosy in this
case.

No one is working on those issues at all that I can see, except insofar
as some stability issue is fixed in 2.29 and backported (pretty rare
these days).

> There is no perfect solution and either ways some users will be unhappy,
> if your real issue is the mapi support why don't we try to see what we
> can do about this one?

Indeed.  That would be lovely.  What do you propose?

> > Let's be honest.  No one is going to follow the RHEL packaging of Gnome
> > 2.28 and be porting Evolution fixes into the Ubuntu repository.  Ditto
> > for Debian.  And certainly the Gnome devs won't be making new releases
> 
> On what do you base those comments? In fact upstream just roll 2.28.3
> less than a week ago and agreed to keep commiting changes to gnome-2-28
> git when it makes sense to make collaboration between the distributions
> which will stay on those versions for a while easier. We do often look
> to fixes from other distributions too and lts versions do get new
> versions of GNOME usually too.

I base my comments on having used Ubuntu for 4+ years now.  I've never
known a single Ubuntu release to ever roll out a complete Gnome service
pack released after the Ubuntu release.  Occasionally there will be a
single app or two bumped, typically if a security fix is discovered.
Even for the last 8.04 LTS, which used Gnome 2.22, the final Gnome
service pack was 2.22.3 (released July 2008) and the large majority of
the Evolution, in particular, packages are still (today) using 2.22.1.
A couple of Evo packages are at 2.22.2, for security fixes, and nothing
is at 2.22.3.

As for Gnome, a 2.28.3 service pack is part of their official roadmap
and completely expected.  Are you saying that they are committing to
2.28.4 etc.?

Since you are staying with the LAST Gnome version of Evo rather than
starting with the next one you can take the latest 2.28 service pack,
but I see no reason to suppose this LTS will be different than the last,
or any of the other Ubuntu releases, in terms of updates.  If the
problem is extremely bad (like my "deleting the entire inbox" bug)
something will get done, I suppose.  If it's just a regular old bug, the
devs will be off working on 10.10 and won't have time to deal with
10.04.

And, I have no problem with that at all: I completely understand.  To
get a release out every 6 months you have to focus on the next one and
limit your efforts on the earlier ones, or nothing gets accomplished.
I'm fine with that normally... but in this case it'll be a year, not 6
months.  Which is much less pleasant.



-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: lucid and 2.6.33?

[Paul Smith]

On Thu, 2010-03-25 at 15:41 -0500, Patrick Goetz wrote:
> Flávio Etrusco wrote:
> > Seriously? Ubuntu is not only about techies, it's about general
> > use(rs) and businesses too. They have to have a solid and well-tested
> > base.
> > If you really wanna know what you're actually getting, you have the
> > sources and the changelogs.
> 
> It's not clear to me how a handful of folks at Canonical or RedHat 
> splicing and dicing kernel code from one version into another 
> necessarily gives you greater stability than an officially released 
> kernel that has been thoroughly tested by thousands of kernel hackers

First, it's not just a "handful of folks".  You should read up on the
testing labs that both Canonical and Red Hat have in place.  I wonder,
these days, how many people _really_ compile their own kernel directly
from kernel.org anymore?  Obviously the kernel devs do it, and probably
some folks using more "tuner" distros like Gentoo.  But how many is
that?  I used to do it all the time but it's been years since I needed
some new feature enough to spend the effort of compiling my own kernel,
rather than just using what came with my distro and waiting a few months
for the next distro release.

Second, the 2.6.32 kernel has been declared to be a long-term stable
kernel candidate by the kernel.org development team (Greg K-H), which
means that it will have important fixes regularly backported to it from
newer kernels by the kernel.org folks as well as RH/Canonical/etc., for
much longer than normal kernels.

Third, all indications are that Red Hat is going to base Red Hat
Enterprise Linux 6 on the 2.6.32 kernel, which means it will get very
long-term support (Red Hat EL 5 is based on 2.6.18, for example), just
like Lucid's LTS designation.

Actually 2.6.32 is, from a scheduling standpoint, just about the most
perfect choice of a kernel for Lucid there could be.

> But you're right, I can always just compile my own kernel, and the most 
> stable kernels I've used are ones I compiled from kernel.org source. 
> But then I have worry about doing all my own security patches and 
> maintenance until such time as I upgrade the distro.  Or I can pour over 
> the Ubuntu kernel source package diffs in my spare time hoping to figure 
> out if some feature has been backported.  Allow me to cut to the chase 
> and declare my original question to unsatisfactorily addressed.

You've had bad luck with distro kernels then.  As above, it's been years
since a kernel failed on me badly enough to want to build my own.

Every new kernel improves many things, it's true: but it's equally true
that it breaks many things.  To commit to providing the latest kernel
always is to simply get on a treadmill that never ends: every upgrade
gives you an entirely fresh new set of problems.

Every distro I've ever heard of understands this and that's why they
never upgrade major software versions, including the kernel, until the
next release.  Any serious problems discovered have fixes backported
instead: that way you are always improving in quality (until the next
release anyway) instead of jiggling up and down around the same level.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: No mouse or keyboard on 10.04 no GNOME

[Paul Smith]

On Fri, 2010-04-30 at 14:28 +0100, Joao Pinto wrote:
> I can't work without my computer and I'm in deep trouble! 
> Who someone help me??
> 
>   This is not a support mailing list, please check:
> http://www.ubuntu.com/support/communitysupport 

When you post there, please be SURE to provide very detailed information
on your hardware.  That includes the form factor (desktop, laptop, etc.)
as well as vendor and model number, amount of RAM, type of video card,
and any other "interesting" things about your hardware.

Also, be sure to mention what version of Ubuntu you were running before
you tried to upgrade.  And, a description of exactly what you did (I
don't know what the "save-upgrade method" is).

Asking good questions is the single most critical factor in getting
usable help.

Good luck!


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Aptitude included in Maverick by default

[Paul Smith]

On Sat, 2010-06-12 at 08:40 +1000, Chris Jones wrote:
> We already have apt-get, dpkg and gdebi. And between the 3 of those,
> all bases are already covered.

That's actually not so, I don't believe.  The only one of these three
that is in any way competitive with aptitude is apt-get.

And, aptitude has an advantage over vanilla apt-get in that it
"remembers" the difference between packages you have installed directly,
and those which were installed by the system in order to satisfy
dependencies.  If you later remove the packages that were installed
directly, then aptitude can also remove the other packages that were
installed to satisfy prerequisites (as long as nothing else is using
them, of course).

Aptitude also has a better "search" capability than apt-get.

Personally I always use "aptitude install xyzzy" whenever anyone
suggests using "apt-get install xyzzy".


Anyway, an unpacked aptitude package is less than 5M (aptitude itself is
just 2.2M).


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Aptitude included in Maverick by default

[Paul Smith]

On Sat, 2010-07-03 at 10:18 +1000, Chris Jones wrote:
> Thanks for dragging this up again Paul. I have moved on from this
> discussion but felt the need to reply since you seem to have
> misunderstood what I meant by my original comments.

Er... I don't know why you're moaning at me.  I posted on this thread
exactly one time, on 12 June (and I CC'd you).  I didn't bring anything
up again.

https://lists.ubuntu.com/archives/ubuntu-devel-discuss/2010-June/011755.html

> To clarify; I was not stating that apt-get is a substitute for
> aptitude. I was not stating that dpkg is a substitute for aptitude.
> And I was not stating that gdebi is a substitute for aptitude. I was
> stating that the between the three combined as a team of apps, you can
> pretty much do everything possible that aptitude can do. Yes, with
> more hands on work and more commands. But it's all there for those who
> want it make use of it.

I think you misunderstood MY comment.  For Ubuntu I'd rather see a
single tool that has a simpler, powerful user interface than multiple
tools that require you to perform "more hands on work and more commands"
to get the same behavior.

But whatever.  That ship has long since sailed since all the various
Ubuntu help pages already direct users to run "apt-get install
" rather than "aptitude install ", and I'm
perfectly happy running "apt-get install aptitude" on my systems.


Cheers!


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Discussion about bug #615504

[Paul Smith]

On Tue, 2010-08-17 at 11:56 -0300, Leandro de Oliveira wrote:
> I thought not having a versionless symlink was a bug, but it seems to
> be an explicit decision, if that is the case, then java apps using
> bluetooth will always need the user to install libbluetooth-dev
> instead of being able to work with the latest version. Since backwards
> compatibility should be a given, I don't see why there coudn't be a
> versionless symlink.

But, backward compatibility is NOT a given; that's why we have versioned
libraries in the first place.

In a POSIX-y system like Linux, shared library versions work like this
typically: the library filename is "libfoo.so.x.y" where "x" is the
major version and "y" (which might have multiple parts) is the minor
version.

The rules are that any changes that do not modify the library ABI can
change only the "y" version but leave the "x" version the same, and
these can be used without recompiling code.

Any changes the modify the library ABI _must_ change the "x" version,
and all applications must link against the new version explicitly.

To support this, the linker (usually) puts a reference to "libfoo.so.x"
in the binary so that's the library that the runtime linker looks for,
and the package will install a symlink "libfoo.so.x" which points to the
currently-installed/newest "libfoo.so.x.y" shared library.

In order for new applications to be compiled, the development package
installs a symlink "libfoo.so" which points to the
currently-installed/newest "libfoo.so.x", or major version.  The
compile-time linker, when it sees "-lfoo" on the link line, will search
for "libfoo.so" and link that, putting (as above) "libfoo.so.x" into the
binary for the runtime linker to find.


Some incorrectly-built applications force the compile-time linker to add
"libfoo.so", only, to the binary: this is wrong because it doesn't
follow the conventions above and the "libfoo.so" won't be available on
systems which do not have developer packages installed (as above).

Note that shared libraries can reference other shared libraries and
typically follow the same rules as above.

You should file a bug against the Bluecove library: it's not being
created properly.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Where can I find more docs/help on upstart?

[Paul Smith]

Esp. how it interacts with Ubuntu/Debian packaging.

For example on my Ubuntu 10.04 system I installed tftpd-hpa and find
that it's been converted to upstart.

No problem, except that I don't want it to start when my system boots, I
want it to be started via xinetd (I need it to work this way in order to
test compatibility with other software; please don't suggest "just don't
do that"... I have to do it).

What's the proper/recommended way to change the configuration of this
package so that (a) it doesn't conflict with new packages installed, and
(b) upstart does not try to start tftpd when my system boots.

In the "old" SysV way there were fancy tools that managed the SysV
symlinks and a method for modifying them that was, be consensus,
preserved across upgrades/etc.  How do I do the equivalent thing with
upstart?

I (finally) found the init.5 man page so I see the format of the config
file, but is there some way to customize the startup of these packages
other than hand-editing conf files?

So far my Google/etc. searching for admin-level details of upstart
hasn't netted me very much.


Thanks!


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Where can I find more docs/help on upstart?

[Paul Smith]

On Tue, 2010-09-07 at 11:28 +0100, Matthew Paul Thomas wrote:
> -BEGIN PGP SIGNED MESSAGE-
> Hash: SHA1
> 
> Paul Smith wrote on 04/09/10 20:46:
> >
> > Esp. how it interacts with Ubuntu/Debian packaging.
> >...
> > So far my Google/etc. searching for admin-level details of upstart
> > hasn't netted me very much.
> >...
> 
> Have you found <http://upstart.ubuntu.com/wiki/>?

Yes, thanks for that.  Most pages there are specifications for future
improvements.

The documentation about today's implementation doesn't give any hints
about how a service can be disabled, a la update-rc.d disable.

All I can find is the description of the init file.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

libc profiling... can anyone use this?

[Paul Smith]

Hi all; I've been trying to use gprof to profile some of my code and the
profiling results for my code only seem to account for about 5 seconds
of runtime, out of a total of 40 seconds.

So I installed the profiled libc library installed and I'm trying to use
that to try to figure out where the rest of the runtime is going.  When
I attempt to link with -lc_p as described in the libc6-prof README file,
etc., I get linker errors and/or immediate segmentation faults when I
run it.

I've filed a bug a few days ago with details:
https://bugs.launchpad.net/ubuntu/+source/glibc/+bug/667470


Has anyone else been able to get this working?  Am I just missing
something here?


Cheers!


PS. Yes I know about valgrind but I was hoping to use gprof in this
case.
-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Evolution in Ubuntu (was: Re: Evolution & Ubuntu 10.04 LTS)

[Paul Smith]

Last year I expressed concern over the fact that Ubuntu was leaving
behind Evolution in Lucid (that is, shipping Gnome 2.30 but leaving
Evolution alone back at 2.28).  The explanation at that time was that
the change from Evo 2.28 to 2.30 involved a lot of significant rework
(which it did) and that the LTS release was not a place to experiment
with that sort of change.

OK, fine.

I was definitely perplexed, then, to see that in Maverick, which is not
by anyone's standards a long-term support release and is running Gnome
2.32, Evolution was left at version 2.30!

And now I've heard (not officially) that Natty, which (as I understand
it) will contain Gnome 3 as the base version, will ship with Evolution
2.32!!

It's one thing to leave behind this package due to the confluence of LTS
and a major amount of rewriting, but what is the justification for
continuing to ship obsoleted versions of Evolution with every release of
Ubuntu?

Is this some sort of new policy decision, that Evolution will forever be
left back in every release?


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Evolution in Ubuntu

[Paul Smith]

On Sat, 2011-01-22 at 16:16 +1000, Chris Jones wrote:
> > Is this some sort of new policy decision, that Evolution will
> > forever be left back in every release?
> 
> Hi Paul. I can understand what you are saying, but is it really that
> important to you? I'm curious as to why it's bugging you this much.
> Does Evolution 2.32 have a particular feature that you require?

Evolution is now, thanks to the tireless efforts of Matthew Barnes
(among others), undergoing impressive amounts of feature addition and
bug fixing every single release.  With every release it increases its
integration with Gnome facilities, reduces bugs, gains new capabilities,
etc. at a pretty rapid pace (compared with a few years ago).  The code
is being cleaned up, dialogs fixed, hangs removed/parallelism
introduced, etc. etc.  Bonobo and other ancient Gnome technologies are
being excised, dbus and gconf uses are being fixed, etc.  The IMAP+
backend is being enhanced to work better with newer IMAP server
capabilities.  Other backends have similar enhancements.

Honestly I can't remember exactly what features/enhancements went into
exactly which version of Evo.  But in general, yes, it is that important
to me :-)

> If you're keen enough, you can probably sneak in the updated Evolution
> into your own system from a third party PPA or something. I'm sure the
> deb files are available from somewhere on the internet. It might be
> worth your while doing a Google search for them.

Well, I have myself created and maintained a makefile that allows people
to build Evo (and its various parts) and install it in a separate
directory, so it doesn't interfere with the package versions; I prefer
this method.

There are a lot of reasons this Makefile (or a PPA) are not optimal,
which I can get into if anyone REALLY cares.


I guess I would turn around your question and say, what's Ubuntu's
justification for choosing this one particular application to "leave
behind" at an older version?  I understand the confluence of "major
changes + LTS" that resulted in the 10.04 decision, but this seems like
something that would need to be individually revisited for each release.

Anyway, it sounds like everything is good for Ubuntu 11.04, sticking
with Gnome 2.32.  I'm presuming that the Evo 3 will be in the Gnome 3
PPA for those who want to try it, along with the rest of Gnome 3.

Cheers!


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Evolution in Ubuntu

[Paul Smith]

On Sat, 2011-01-29 at 10:37 +1000, Chris Jones wrote:
> I still don't see any real necessity to kick up such a fuss Paul.

*shrug* That's nice.  I do, or I wouldn't have asked the question.
Since you don't even use Evolution and I do, every single day, I'm not
sure why you feel the need to give me grief about this.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: The Dell Latitude reality check

[Paul Smith]

On Wed, 2011-02-16 at 15:49 -0600, Patrick Goetz wrote:
> A better solution would be to include the most common proprietary 
> drivers in some kind of encrypted sandbox on the install CD and let
> the user choose whether or not he/she would like to use the
> proprietary drivers at the time of the install.  I'll let the GPL
> legal experts figure out precisely how this could be done, but it
> seems to me it should be possible.

Hi Patrick.  I'm not familiar with the specifics of the broadcom
licensing per se, but if it's like the other proprietary driver
licensing problems we have then you're misunderstanding the situation.

The problem has nothing to do with the GPL vs. LGPL vs. whatever; in
fact it has nothing to do with licensing of any part of Ubuntu at all.

The problem is with the licensing of the _proprietary driver_.  These
drivers typically allow no-charge download directly from the vendor
website, but the license includes terms that make REDISTRIBUTION illegal
(without permission).  That means that Ubuntu _cannot give you_ the
driver, unless they go to the vendor and pay $$ (presumably you're not
asking that a distro you downloaded for free should pay $$ for the
privilege of shipping these proprietary drivers) or work out some other
arrangement.

In short, there's simply nothing that Ubuntu can do about this if the
vendor won't agree to allow redistribution.

For things like nVidia graphics drivers or the Microsoft truetype fonts,
etc. we can get away with this because they are not critical to bring
the system up, so we can get the system up then there's a little trick:
there's a package available that will automatically go to the vendor
site and download the package directly from there (no redistribution by
Ubuntu here) and install it.

Of course, this trick does not work with a network driver for obvious
reasons!


I know this doesn't make anyone, including me, happier but at least we
can focus our ire in the correct direction and not shoot the hapless,
and helpless, messenger.

Cheers!


PS. Hopefully I've not completely misinterpreted the situation with the
broadcom drivers... I probably should have gone to look at the license
before writing all this.  Oh well!  :-p :-)


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Ubuntu One needs cloud encryption like LastPass does it

[Paul Smith]

On Thu, 2012-04-05 at 18:33 -0500, Jordon Bedwell wrote:
> On Thu, Apr 5, 2012 at 5:42 PM, Sam Smith  wrote:
> > The point is that SpiderOak (and Lastpass) never know the user's password.
> > And never receive the encryption key. The key never leaves the user's
> > computer. The server never gets it. The only thing that ever lands on the
> > server is an encrypted blob.
> 
> From their website "Retrieve files from any internet-connected
> device", "Access all your data in one de-duplicated location"... I
> know to the easy consumer that doesn't spell lies but to me it reads
> "We do know your encryption key, if we want to and little do you know,
> we do have the ability to get the key that encrypts the encryption key
> too."  Companies lie all the time, or they tell pieces of a story and
> never tell the entire story.  Though I don't know if it's more of a
> lie then an assumption on their end and maybe even they themselves not
> even understanding what could possibly go wrong, or they just don't
> care because the user doesn't pay too much attention after "WE NEVER
> KNOW."
> 
> The key to knowing the full story is read "Retrieve files from any
> internet-connected device."  To add to it, let me point out this:
> "Easily access all of your data from any device within your SpiderOak
> network or on the web" which contradicts this: "SpiderOak never stores
> or knows a user's password or the plaintext encryption keys which
> means not even SpiderOak employees can access the data" and it's no so
> much a direct contradiction as much as an arrogant assumption that we
> (or I guess only I in this conversation) don't realise that their
> employees do have a way to access it, they just need to do a couple
> minutes worth of work, that is what makes it contradict.

None of the statements you quote above are proof of lying (or
incompetence), or even indicative of it.

The crux of the issue is simply that SpiderOak is a proprietary program
and so you don't know what it REALLY does.  The model that SpiderOak
documents on their web site IS secure.  It's definitely more secure than
ubuntuOne.  The passphrase is never sent to the server at all and the
content cannot be (reasonably) decrypted without the passphrase.  They
have a downloadable application that runs on your local system, and if
you use that and never use their web interface to browse your files then
your passphrase is never transmitted over any network at all, encrypted
or not.

If the software behaves as documented, then they are right: SpiderOak
employees cannot decrypt your files.  Period.  Phrases like "retrieve
files from any internet-connected device" don't matter: it just means
you enter that passphrase into the application running on the local
device to decrypt the files after they're downloaded from the servers:
it doesn't require the passphrase to be transmitted to the servers.

Of course the problem is "IF", above: the _documented_ model is secure,
but that doesn't stop a SpiderOak employee with sufficient access from
adding a back door to the application, which will grab the passphrases
and send them along.  That's a risk with ANY encryption software that
you didn't write completely yourself, of course, even ssh etc., but it's
much more risky with proprietary software for obvious reasons.

If that's what you meant, then you should have just said so clearly
instead of couching it in ominous-sounding hints and accusations.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Ubuntu One needs cloud encryption like LastPass does it

[Paul Smith]

On Fri, 2012-04-06 at 01:41 +0100, Dale Amon wrote:
> I do not know the details, so I will ask: is it the case that:

All we can know for sure is the way the system is DOCUMENTED to work, as
I said in my other email.

>   * The user crypto key is generated on the
> the user machine.

Yes.

>   * The password for the user key is set on
> the user machine and never leaves it.

Yes.  Well, the passphrase is in your head obviously, so of course it's
available wherever your head is.

>   * The user crypto key never leaves their
> machine(s).

No, I believe they upload the generated key to the server, after it's
been encrypted with your passphrase.

>   * The user's password for their crypto key
> is never used outside the confines of their
> local machine.

Yes, as long as you don't use their website to access your content and
only use the local tool.

>   * The data is fully encrypted on the user
> machine and only encrypted data transits
> the net and sits on the storage server.

Yes.

>   * The encryption algorithm is such that 
> no key except the one on the users 
> machine can decrypt the remotely stored
> data.

Yes.

> If the statement made in the other reply is true, and you can
> 'retrieve your data from any internet device' then it is patently
> obvious that data security *is* violated.

Why is it "patently obvious"?  I'm sure when they say "any internet
device" they don't mean devices that do not have access to the secure
tokens necessary to decrypt the content.  They mean a device that has
internet access (so it can retrieve the encrypted content from the
server), and where you can enter your passphrase to decrypt it.

Even if they did not upload the crypto key, that doesn't mean that you
couldn't have it with you on a USB key or something, and still access
your data from "any internet device".

I'm sure that they felt that forcing you to keep both the passphrase AND
the crypto key yourself was simply not a commercially viable solution
for the general public.  It would be nice if they offered an option
(with appropriate cautions) to not upload the keys at all, I agree.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: server went unbootable

[Paul Smith]

On Sun, 2012-04-08 at 13:55 +0100, Dale Amon wrote:
> There weren't many choices in 2001 or 2002 when I was
> looking into it. Do you have any suggestions on one
> that might be particularly useful? It doesn't have to
> be terribly fancy, just enough so you can get a shell
> and 'fix things'.

I used dropbear for a while.  Works well.

http://matt.ucc.asn.au/dropbear/dropbear.html

Cheers!


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Where is libgcr-3-common:i386?

[Paul Smith]

I'm trying to run a program under Wine in my Ubuntu 12.04 64bit fresh
installation.  When I run it I get this error:

p11-kit: couldn't load
module: /usr/lib/i386-linux-gnu/pkcs11/gnome-keyring-pkcs11.so: 
/usr/lib/i386-linux-gnu/pkcs11/gnome-keyring-pkcs11.so: cannot open shared 
object file: No such file or directory

I've discovered that the 64bit version of that file lives in
gnome-keyring and so I tried to "sudo apt-get install
gnome-keyring:i386".  I found some libraries it requires and those
require other libraries, all of which do exist, until I get down to a
requirement for:

$ sudo apt-get install libgcr-3-common:i386
Package libgcr-3-common:i386 is not available, but is referred
to by another package.
This may mean that the package is missing, has been obsoleted,
or
is only available from another source

E: Package 'libgcr-3-common:i386' has no installation candidate

However it's odd because if I go to the Ubuntu site to look it up, that
library seems to be packaged properly; for example:

https://launchpad.net/ubuntu/precise/i386/libgcr-3-common

Can anyone explain what the message means and how I can get a copy of
that library to install?  Am I missing some extra repo (I think I've
turned them all on at this point)?


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Problems with setuid app in Ubuntu 12.04

[Paul Smith]

Hi all.  I've recently installed Ubuntu 12.04 64bit.

I'm using a proprietary VPN utility from Juniper Networks on my Linux
system.  In previous versions of Ubuntu, it worked just fine.  In the
current version of Ubuntu, I'm getting failures.  For some reason it's
not letting me invoke a setuid application.

The way it is deployed is it unpacks into a hidden directory under
$HOME, then the main VPN program needs to be made setuid root so that it
can be invoked by me but do root-y things.

Then, you can either run the tool directly from the command line or you
can run a little Java control window which manages the VPN.  If I run
the program directly from the command line, the setuid works and the VPN
comes up and works fine.  But, I can't control it or see how long it's
been up.

Whenever I try to use the Java control panel the GUI comes up and tries
to run the setuid program, but it fails and then the whole thing
crashes.  I get this error:

  Failed to setuid to root. Error 1: Operation not permitted

But I have clearly set the right bits and it works when invoked
directly.  I've tried many different variations of Java including
downloaded ones directly from Sun/Oracle.

One note, the application is 32bit and so I need to run 32bit Java as
well.  Not sure if that matters.

Has the 12.04 release installed some new security measures that might be
keeping my setuid program from working properly?  I've tried putting
Java under /opt/jvm and also run "service apparmor teardown" to try to
be sure apparmor is not involved, but I don't know enough to know if I
succeeded.

I'm really stuck and could use any pointers or tips anyone has.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

NetworkManager VPN not usable: workarounds?

[Paul Smith]

Hi all.

I've posted a critical bug in Ubuntu 12.04 in Launchpad more than a
month ago:

https://bugs.launchpad.net/ubuntu/+source/network-manager-openvpn/+bug/990765

Because of this bug I cannot use my Ubuntu system to connect to work
anymore: I've had to create a virtual machine and load an older Ubuntu
release in it and use that while I wait for a fix or workaround.

This is simply not sustainable: my work-from-home productivity has
dropped through the floor using this method.  Something has to give.

At the same time there doesn't seem to be anyone interested in working
on this bug so far.

Back in the day all of this would have been controlled through simple
scripts and text file configurations and I could just go in and hack it
myself if the UI was misbehaving.  It's immensely frustrating to me that
something so stupid and trivial should cause so much pain.  Does anyone
have any thoughts about how to work around this issue, or where I should
go to get more traction for solving it?


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: The cloud directory name "Ubuntu One" has a space in it....

[Paul Smith]

On Wed, 2012-12-26 at 18:09 +0200, Dmitrijs Ledkovs wrote:
> On 26 December 2012 16:52, Fredrik Öhrström  wrote:
> > this causes problems when I try to use the directory for my source code
> > projects that use make or when I put configuration files for dosemu in 
> > there,
> > or a thousand other situations, when spaces in directory names cause
> > problems
> 
> Exactly what is broken? It's 2012 surely software must be able to deal
> with " " in file names / paths.
> Maybe we can fix the broken piece instead?

The "make" program, as defined by POSIX and implemented by every UNIX
system since the 1970's, cannot support pathnames containing whitespace.
The format of makefiles is fundamentally word-based with whitespace
separators and there is no syntax supporting escaping of special
characters.

Changing this is quite a significant amount of work (it's not just
introducing an escape character: make uses a "lazy evaluation" scheme
for its makefiles which means all the internals of the implementation
would need to change as well, not just the front-end makefile parser),
and would violate standards and force makefiles to be non-portable to
any other implementation of "make".

I'm not taking any position on what the cloud directory name for "Ubuntu
One" should be.  I'm just stating facts related to using "make".


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: The cloud directory name "Ubuntu One" has a space in it....

[Paul Smith]

On Thu, 2012-12-27 at 10:45 -0500, Rodney Dawes wrote:
> And you really shouldn't compile things within a synchronized
> directory.  If you have two different machines of different
> architectures, the compiled binaries being synchronized could cause
> problems.

I obviously have no idea exactly what the original poster was trying to
do, but it's perfectly possible (and reasonable) to have the source code
live in the shared directory so it can be used on multiple systems, but
write the makefile such that the results of the build (object files,
etc.) are created in ANOTHER directory, which is not shared.

In fact, every single GNU auto-tooled package supports this model out of
the box (unless the author messed up), and for some it's the ONLY
supported model.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Our Networking Story

[Paul Smith]

On Fri, 2014-03-07 at 09:39 -0500, Bryan Quigley wrote:
> On Fri, Mar 7, 2014 at 1:34 AM, Dale Amon  wrote:
> > The only feature I hold near and dear is that I be able
> > to ssh into a server in a rack 8000 miles away, fiddle
> > with /etc/network/interfaces if needed, and then reliably
> > ifdown/ifup one of god knows how many connections (I often
> > work with machines that have 8 or even more hardware ethers,
> > not to mention ethn:m's.
> 
> Could you detail what process you are exactly using to do this
> reliably?  Are you using bonds/vlans/bridging?

I'm not sure this level of detail is needed.

I've not tried this on Ubuntu, so maybe it already works fine, but
similar to the OP's request on remote Red Hat systems it is always
supported to run:

  service network restart

and know that your ssh session will not be dropped while the network
restarts (unless you screwed up the network config and it doesn't come
back up of course).

This also works:

  service network stop && service network start

Of course, trying to run "stop" followed by a "start" in a different
command line can't work:

  service network stop
... uhm ... oops ... time for a drive!!

I think the request is that there be a straightforward "restart" option
that is ensured to not drop the connection (assuming, again, that the
start works properly and you haven't changed details of the connection
you were using).


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Upgrade issues

[Paul Smith]

On Fri, 2014-05-30 at 12:25 -0700, Dale Amon wrote:
> Oh, and when I switch desktops, only one screen changes. These
> massively decreases the space I have available for windows. Even
> with the four desktops and two screens I typically have all of
> them completely filled and the lower tool bar half full and
> up to over filled.

In my experience the upgrade from one version of Gnome to another is not
always so smooth.  That is, the upgrade of the _operating system_ works
fine and dandy, but the configuration files and configuration in your
user directory from an older version of Gnome don't always upgrade
smoothly to the newer version of Gnome and cause user / desktop issues.
It's more pronounced the more Gnome versions you skip when you upgrade.

That sounds like the kind of thing that happened to you.  If you wanted
to test this you could just create a temporary user account and log in
as that account, and see if the UI issues are still there or not.


What I like to do when I upgrade between major versions is create a
brand new user directory, and simply move files I want to continue to
use from the old user home directory to the new one (music, pictures,
documents, etc.)  I also keep all my "work" files (code, etc.) outside
of my home directory in another directory (still in the /home partition)
so that I don't need to mess with moving that.

There are probably tools or instructions you can find for cleaning out
all the Gnome cruft from the previous versions in your home directory:
if you do that then log back in Gnome will recreate "pristine" copies.

Of course you lose your customizations but I typically only have a few
and they're easily reinstated.  YMMV.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Kernel releases

[Paul Smith]

On Thu, 2015-07-30 at 10:29 -0700,
christopher.carl...@zodiacaerospace.com wrote:
> I think having to upgrade the OS more than once a month is
> ridiculous.  One of the reasons I've always preferred Linux is because
> I don't have updates every week, but it seems we have kernel upgrades
> more often than that.

You don't say what version of Ubuntu you're using.  But, if you want to
know why the package was updated you can check the changelog; for
example:

http://packages.ubuntu.com/trusty/kernel/linux-image-3.19.0-25-generic

or whatever version of the kernel you have, then click the "Ubuntu
Changelog" link.  Or, you can read the changelog.gz file in
your /usr/share/doc/linux-image-$(uname -r) directory on your local
system.

Most likely the updates are due to security issues: you can find the CVE
numbers and look them up.  Or they are for other bugs; you can look up
the LP number.

You don't have take every update that comes along: you could simply
choose to update once a month instead.  Or, I often update immediately
but don't bother to restart my system (even though it says I need to)
until it's convenient for me.  For kernel packages there's absolutely no
reason you need to restart immediately; your system will work just fine
(modulo whatever bugs were fixed of course) running the old kernel for
as long as you want until you restart.  Most other packages are the same
way.

You can also reset your software updater settings to check for updates
only once every two weeks instead of daily.  Or you can set it to never
automatically check for updates, and just do it yourself whenever you
want to.

But many people prefer to get their security patches immediately and be
notified quickly when a security issue appears.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Kernel releases

[Paul Smith]

On Thu, 2015-07-30 at 11:49 -0700,
christopher.carl...@zodiacaerospace.com wrote:
> I also don't always restart, but I presume the only reason we got a
> kernel update is because of a bug or security, which encourages me to
> restart.  It nags at me (not the system, but my concern that there's a
> problem lurking). 
> 
> My whining is brought on by my strong dislike of Microsoft, and I get
> irritated that I have to reboot my Windows 7 machine every week.  To
> think that Ubuntu is following in their footsteps concerns me.

Well, all software has bugs.  It sounds like you're upset that Linux has
bugs...?  I don't know how to help with that concern :).

If that's not it, then it's just a matter of whether you want to have
access to the fixes to the bugs as soon as they're available, or have
the bugs all batched up and delivered in chunks once a month or so.
There's no one right answer to that, different people want different
things.  So... the best you can do is offer people a choice.


-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: Problem with update manager

[Paul Smith]

On Sun, 2016-04-24 at 15:17 +0200, Андрій Шелунцов wrote:
> Hello.
> Update manager does not show update to 16.04 

Please always specify which version and flavor of Ubuntu you're running
now whenever asking for help.

If you're running Ubuntu 14.04 LTS, that's expected.  Unless you
specifically request it, you won't be offered an upgrade until Ubuntu
16.04.1 is released.

If you're running Ubuntu 15.10, then you may need to wait a few more
days to be offered the upgrade.

If you're running any other version, upgrade directly to 16.04 is not
available.  You'll first have to upgrade to one of the above two
versions.

You can force an upgrade if you want with "sudo do-release-upgrade".  Be
sure you've completely upgraded your current version first!

Be aware that the sooner you upgrade the more likely you are to run into
issues...

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: LibreOffice Install and Use Problems

[Paul Smith]

On Tue, 2016-09-27 at 08:51 -0700, paulwhee...@cox.net wrote:
> I attempted to install LibreOffice (office productivity suite
> (metapackage)) on my mint Linux system.
> 1. Synaptic said the package was broken - Image 1. I changed
> repositories and got it to install.
> 2. However, it would not start. Found that it had been installed as a
> file owned by root, so user could not run it. - Image 2. I changed
> owner to user.
> 3. I tried to start it, again, and got message 3.
> 
> The last time I tried to report a bug in this package, someone told me
> I was in the wrong place. I am not sure how talking to the package
> maintainers could be the wrong place!

I think you're misunderstanding the complexity of the F/OSS ecosystem.

For a software package like LibreOffice, there are upstream maintainers
who actually write the code.  They provide source code to users.  Some
of those maintainers also provide binary packages that can be installed;
LibreOffice does this for example:

https://www.libreoffice.org/download/libreoffice-fresh/

However, most upstream maintainers don't do this, they only provide
source code.

Then there are the distributions: Ubuntu, Debian, Red Hat, Gentoo, Arch,
etc.  They take the source code from upstream maintainers for all the
different packages that make up a complete GNU/Linux system, compile
them to create packages that are appropriate for their particular
distribution (and do a huge amount of other work involved with
constructing a usable and useful system from a bunch of component
parts).  So you can't (generally) install a Red Hat package on a Debian
system, or a Gentoo package on a CentOS system, etc.  You have to get
the package that was built for your particular distribution.

That's why people are telling you that you're in the wrong place:
problems with the installation of packages are specific to the
distribution you're using, and you're using Linux Mint, not Ubuntu--this
mailing list is for people using the Ubuntu distribution.

Linux Mint in particular is a "derived distribution", which means it is
created by taking the results of another distribution (in this case
Ubuntu and Debian), and adding its own set of packages on top of that.
 So, a large number of the packages in Linux Mint _are_ straightforward
Ubuntu packages, unmodified.  But, not all of them, and the package
management on Mint is not identical to Ubuntu (as I understand it),
which means problems in package installation may not be related to
Ubuntu.

I'm sure that if people using Ubuntu were having this problem installing
LibreOffice on their systems we'd know about it by now, so presumably
the problem is specific to Linux Mint.

You'll have more luck asking on one of the forums here:
https://www.linuxmint.com/links.php

It may be that after discussion there you'll find that there really is a
problem with the Ubuntu packaging, but it seems unlikely as LibreOffice
is hugely popular.

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: 64bit Motherboards are a minefield of config problems

[Paul Smith]

On Sat, 2017-08-05 at 08:37 +0800, Jesse Steele wrote:
> Generally, installing Ubuntu on 32 bit machines has been no problem.
> However, different 64 bit motherboard manufacturers have different
> native BIOS settings, many of which create problems for installing and
> booting to Ubuntu.

Maybe you can give some examples of what kinds of problems you mean.

I've been running GNU/Linux distributions of all types exclusively on
64bit systems for probably 15 years or more and I've NEVER found a
motherboard or BIOS that gave me any problems.  Your message sounds like
many motherboards won't work with Linux and you have to search carefully
to locate a compatible one.  That's definitely not been my experience.

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss

Re: It is time to use MATE DE as default desktop on Ubuntu 18.04 LTS

[Paul Smith]

On Sun, 2017-10-22 at 13:41 +, Myself wrote:
> a) You become less powerful
> b) They become more necessary
> c) Everyone bows to the great Gnome developers.

Oh for...

I've unsubscribed from this list.  The level of ridiculous inanity has
overwhelmed my ability to deal with it... there's far too much of it in
the world at large these days; I don't need it in my mailbox as well.

I hope anyone who is honestly looking for help with their system
manages to find it, but it won't be from me.

Cheers!

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss