[ale] just installed LibreOffice in Linux, should have been easier
Don Lachlan
ale-at-ale.org at unpopularminds.org
Tue Mar 15 17:50:33 EDT 2011
On Mon, Mar 14, 2011 at 12:54 AM, Ron Frazier
<atllinuxenthinfo at c3energy.com> wrote:
> Hypothetically speaking, if I wanted to get rid of LibreOffice (which I
> don't) is there a way to identify the master package, so to speak? A
> Synaptic search yields 22 items. Nothing jumps out as being a master
> record. There's no mention of it in the application in Software Center.
I don't know LibreOffice and I think others have followed up with
pretty decent information.
Generally, Ubuntu (like Debian) will create a master package for
application "foo" and call that package "foo" - it may be a virtual
package, it might be the main binary, but if you install that via APT
then it will install all the other packages that it depends upon
(libfoo, libfoo-tuesday, foo-tuesday, foo-gtk, etc.). In the same way,
uninstalling "foo" should uninstall all those packages along with it.
> If I
> count each OS as a separate machine (even though some have multiple OS's
> on the same hardware), I have to do this conversion on 3 Linux machines
> and 5 Windows machines. So, you can see how that original procedure I
> found put me off substantially.
You later responded to someone else:
> I was praising the fact that I got LibreOffice up and running on Windows
> a heck of a lot faster and easier than I did in Linux by following the
> most obvious method of installation.
It wasn't "the most obvious" - it was the most obvious to *you* based
on *your* prior experience. You Did It Wrong. You made a mistake.
That's understandable, because it was the obvious/right decision based
on your experience, but I think you're still stuck believing that
GNU/Linux should operate according to your expectations, based upon
your prior experience with other systems.
It doesn't and it shouldn't. GNU/Linux has its own way of doing things
(based on a lot of old school UNIX conventions) and Ubuntu has its own
way beyond that - and GNOME has its own way beyond that. But it will
not be like Windows.
Some of your criticisms are very valid, but some are not.
>> Yes, I can come up with solutions which meet your requirements;
>> however, it's impractical when you're accommodating dozens of
>> distributions, architectures, versions, etc. for THOUSANDS of software
>> projects.
> I understand what you're saying. I'm a geek, so I can edit config files
> and go through elaborate procedures when necessary, if I have a motive
> to. I've been around the block for 27 years or so with Windows.
> Believe me, especially pre year 2000, configuring software and
> installing it or uninstalling it could be a real bear. There is a whole
> sub industry of "installers" for Windows. However, modern applications
> installations, for the most part, are double click it, answer some
> prompts, done, it works, use it. That's what Linux developers are
> competing with if they want mind share of average users who are not
You are factually wrong. Modern application installations on *Windows*
are mostly double-click but on MacOSX (and previous) it's
drag-and-drop and on GNU/Linux it's select-and-install from a GUI like
Synaptic or "install packagename" from a CLI. In fact, I think some
GNU/Linux distributions or desktops allow you to double-click a
package and it will run the package manager.
And that's just desktop operating systems. The iP([ao]d|hone) and
other mobile platforms use an "app store", while some allow you to
install from web pages.
Those are ALL modern application installations. Please, stop assuming
the Windows way is a) Right and b) Universal.
> ultra geeks. I'm sure many would say the diversity of Linux is the
> beauty of Linux, but as you pointed out, it also comes with a price. If
> I were to develop for Mac, I have maybe 2 major versions to worry about
> - 9 and 10. If Windows, maybe 3 - 5 versions. For Linux, dozens or
> hundreds. So, I can see how that would be a problem. If I were a
> developer, I think I would focus on the top 5 or so, as I mentioned. If
> each project could make their applications installers click and go, but
> compatible with the distro package manager, then that would really help
> average users get things going the way they want.
You are not speaking from knowledge or experience, you are
hypothesizing and conflating two separate issues: making it simpler
for users to install software and making it simpler for developers to
create a package.
Installing software is usually pretty easy; your experience w/
LibreOffice was harder than it needed to be because you were working
under false assumptions. Additionally, you're still pretty new to
Linux but you're trying to live a bit more "on the edge" of software,
so it makes sense that you ran into a problem that most users would
not. But generally, installing software in Linux is incredibly easy.
OTOH, building (and then maintaining) packages is a fair amount of
work and can be very difficult for someone who doesn't know the
platform they're building for - RPM is very different from dpkg, and
even building a RPM for both CentOS and OpenSuSE will require
significant work for each platform.
Currently, a developer writes their application, probably distribute
binaries for their choice distros, and leaves source code for anybody
that wants to build it for other systems. Then, after it gets a user
base, people ask their distribution to begin carrying it and someone
steps up to say, "Hey, I'll build foo for our distro!"
Each distribution is different. Each application is different.
Sometimes an app gets packaged up quickly, sometimes it takes years.
Sometimes the developer does the packaging, but usually not.
Your model would shift a significant amount of work onto developers
when the current system works pretty well and gives our users far more
software than Microsoft or Apple do.
My point is, I've written software and packages for GNU/Linux - for a
living - and you haven't, please stop telling me how easy it would be
to change.
> all. If they need to install something, it needs to be very obvious and
> simple, or something I can talk them through on the phone, or I end up
This comes back to what's "obvious". Apple makes some of the most
intuitive applications yet they are nothing like what you're claiming
is "obvious".
> By making Linux easier to use, and more accessible to average people, it
> will gain mind share and new users faster than ever. That way, people
> will see that they probably don't have to use expensive proprietary
> solutions.
GNU/Linux is much easier than you seem to think it is. It's not as
easy as OSX, but it's also not a restrictive sandbox.
GNU/Linux _seems_ harder than it is because it works differently than
what most people expect, either based on experience with Windows or
OSX. To fix that, I think we need to focus on teaching users the UNIX
way (and Ubuntu way (and GNOME way)). Yeah, we should improve some OS
level issues, but the impression "Linux is hard!" is about
expectations versus reality - if people expect GNU/Linux to act like
MS Windows, it will always fail. If they expect it to behave like
GNU/Linux, it wins.
-Don
More information about the Ale
mailing list