Why a linux version?

20 views
Skip to first unread message

Pierre Raybaut

unread,
Jun 10, 2008, 5:07:04 AM6/10/08
to python(x,y), cgo...@gmail.com
Hi all,

Python(x,y) was originally designed for Windows XP/Vista because
installing all necessary Python modules for scientific applications on
these platforms may be difficult for new users.

On the other hand, with Debian/Ubuntu for instance, the package
manager (apt) provides an easy way to get and install a package set
equivalent to Python(x,y) distribution on Windows.

So, it seems that a linux equivalent to Python(x,y) distribution is
not really needed.

If you have some issues with this kind of installation procedure on
linux, or if you don't agree with above statements, please send a post
here: Claude Gouédard (advanced linux Python user) and myself would be
pleased to debate on this.

Regards,
Pierre Raybaut

Johannct

unread,
Jun 10, 2008, 5:42:05 AM6/10/08
to python(x,y)
you might want to check that indeed an equivalent set of cross-
validated packages is available in a if not any linux distro. As for
me, I have FC8 and there are many missing, including of course ETS.
And anyway, isn't the global cross-validated set of package a plus for
any new linux users? I remind you that by default all these are not
installed on a new distro, and it would take trial and errors/time/
confusion/anger for a new user to iterate on her/his usage of python
until she/he comes around all the packages she/he needs.

So basically, I do not understand your point. In any case, you might
want to change your web page into "Free scientific and engineering
development software for Windows users"

Johann

m.s.

unread,
Jun 10, 2008, 7:46:01 AM6/10/08
to python(x,y)
> you might want to check that indeed an equivalent set of cross-
> validated packages is available in a if not any linux distro. As for
> me, I have FC8 and there are many missing, including of course ETS.
> And anyway, isn't the global cross-validated set of package a plus for
> any new linux users? I remind you that by default all these are not
> installed on a new distro, and it would take trial and errors/time/
> confusion/anger for a new user to iterate on her/his usage of python
> until she/he comes around all the packages she/he needs.
>
> So basically, I do not understand your point. In any case, you might
> want to change your web page into "Free scientific and engineering
> development software for Windows users"

As a long time Linux user, I don't see very much added value in
providing Python(x,y) as a Linux bundle instead of providing separate
packages in a cross-validated distribution.

But if you care about us (and I hope so!), you could help testing of
the packages in Python(x,y) on Linux, and provide advice to the major
distributions (Ubuntu, Fedora, Gentoo) on what versions bundle
together.

Such a bundle would instead be most useful for OS X, that is the
platform that gave me most headaches in installing python modules.
There are a few unofficial ones, but are pretty scattered and poorly
maintained.

I'd also like to see other bundles for Windows users. For example, I
personally use Scipy/Numpy/Matplotlib with WxPython/WxMPL instead of
Qt.

m.

mrozko

unread,
Jun 10, 2008, 7:50:28 AM6/10/08
to python(x,y)
Hi all,

I stay somewhere in the middle...

Pierre has true, that good Linux distro provides easy way to get and
install a package for Python users. On the other hand, the package
which concentrate (maybe most) useful python-packages makes a sense
for me. I think the Python(x,y) should have also a linux version,
certainly if authors would like to distribute this M$ WinXP/Vista-
fiendly package among the people.

I think the verdict for designing the linux version of Python(x,y) is
on its authors.

At the end, I would like to vote for "Python(x,y) for Linux"...

All the best,
Marian

Pierre Raybaut

unread,
Jun 10, 2008, 7:59:16 AM6/10/08
to python(x,y)
Johannct wrote:
> you might want to check that indeed an equivalent set of cross-
> validated packages is available in a if not any linux distro. As for
> me, I have FC8 and there are many missing, including of course ETS.

Is it not possible to install ETS via easy_install ?
My point is that linux users are not afraid of such things as command
line installation. On the contrary, Windows users installation skills
are -generally- limited to one-click installers. That's why we were
thinking that perhaps a linux version would not be as useful as the
Python(x,y) installer for Windows users.

> So basically, I do not understand your point. In any case, you might
> want to change your web page into "Free scientific and engineering
> development software for Windows users"

Please try to be constructive in your comments: of course, the web
page title would have to be changed but I don't think that this change
would represent in itself a technical challenge... Good suggestion
though ;)

Anyway, thanks for your comments, we are really opened on this topic
(hence the post...).

Pierre

Johannct

unread,
Jun 10, 2008, 8:03:23 AM6/10/08
to python(x,y)
> But if you care about us (and I hope so!), you could help testing of
> the packages in Python(x,y) on Linux, and provide advice to the major
> distributions (Ubuntu, Fedora, Gentoo) on what versions bundle
> together.
>
I do not understand what you are asking me to do here. Please provide
a specific example of a python(x,y) package to be tested on Linux.
Let's take an easy_install example : wouldn't it be nice to have some
bundle pythonXY that when used with 'easy_install pythonXY' downloads
all the correct packages that currently make up for the windows
pythonXY bundle? I agree that this is completely different than what
is been done currently, but I just want to point out that a clean easy
way to retrieve a cross-validated set of all these packages does not
exist on Linux, to the best of my knowledge. And I would add as well
that I work with some people who do not have root access, so that a
bundle with a separate python package would also be a plus for
them....
Johann

Pierre Raybaut

unread,
Jun 10, 2008, 8:06:28 AM6/10/08
to python(x,y)
m.s. wrote:
> I'd also like to see other bundles for Windows users. For example, I
> personally use Scipy/Numpy/Matplotlib with WxPython/WxMPL instead of
> Qt.

wxPython is already included in Python(x,y).
As for WxMPL (didn't know about it), consider it done!

Thanks for your comments.
Pierre

Johannct

unread,
Jun 10, 2008, 8:35:14 AM6/10/08
to python(x,y)

I thought I was trying to be constructive! :)
Anyway, if the reason for pythonXY is to provide a complete one-click
scientific python environment to windows users,
it is fine with me, and I even say kudos though I don't use windows :
the more scientific python users across platforms, the better.
But the first thing I read when I go to the web page is :
"Python(x,y) is a free scientific and engineering development software
for numerical computations, data analysis and data visualization based
on Python programming language, Qt graphical user interfaces (and
development framework) and Eclipse integrated development
environment."

Well, that whets my appetite! So I inquire as to the availability for
linux. Your answers seem to point to the fact that the scope of your
effort is not that large, or that you envision serious technical
challenges that I absolutely do not pretend to judge and am ready to
believe to represent too much of an effort.
So be it, I will continue to rely on my intrepid command-line
skills. ;)
Johann

Pierre Raybaut

unread,
Jun 10, 2008, 10:13:48 AM6/10/08
to python(x,y)
Johannct wrote:
> I thought I was trying to be constructive! :)
> Anyway, if the reason for pythonXY is to provide a complete one-click
> scientific python environment to windows users,
> it is fine with me, and I even say kudos though I don't use windows :
> the more scientific python users across platforms, the better.
> But the first thing I read when I go to the web page is :
> "Python(x,y) is a free scientific and engineering development software
> for numerical computations, data analysis and data visualization based
> on Python programming language, Qt graphical user interfaces (and
> development framework) and Eclipse integrated development
> environment."
>
> Well, that whets my appetite! So I inquire as to the availability for
> linux. Your answers seem to point to the fact that the scope of your

You are absolutely right. And these introducing words were not
randomly chosen: until now, we were seriously thinking of developing a
linux version (I mentioned it many times on scipy-user list).
But I completely disagree if you are implying that being available
only for Windows makes one software less "free" -- we do not sell
Windows licenses! (...ewk :) ) we are just helping Python users who
are using -not by choice, most of the time- the Microsoft OS.

This beeing said, if we choose not to develop the linux version, I
will immediately add "for Windows" somewhere on the website... even if
it is really common to mention supported platforms on the download
page only (and BTW, the word "cross-platform" has not been mentioned
on the website... so you are really trying to make me say what I did
not ! ;) ).

> linux. Your answers seem to point to the fact that the scope of your
> effort is not that large, or that you envision serious technical
> challenges that I absolutely do not pretend to judge and am ready to
> believe to represent too much of an effort.

These are not technical challenges (I'm not a linux expert though):
the main problem is the time needed to develop this linux version. It
seems that the linux version would be far more simple to do than the
Windows installer. But, even if all the related individual tasks are
simple, it takes time, and so we (legitimately I think) hesitate to
start this project.

Pierre

gabriel

unread,
Jun 10, 2008, 11:08:49 AM6/10/08
to python(x,y)
I can see that I will be a serious effort to support Python(x,y) for
Linux. Python(x,y) is essentially a distribution, and since different
Linux distros use different package management tools Python(x,y) will
have to be adapted to each of those. That costs time. I use Linux and
would be very happy with a Python(x,y) for Linux. But in the end I
think that indeed most Linux people will figure out how to install
packages themselves. Even though I don't use Windows I would be more
happy to see a fully functional Windows Python(x,y) than a less
functional one that is supported on many platforms. The reason for
this is that I worry that scientific software that I write on my Linux
box is in practice less cross-platform than I'd like, simply because
the barrier for many Windows users to install the myriad of packages
that my software depends on (matplotlib, pytables, mdp, numpy, scipy
and who-knows-what) is high. Ideally I'd like to advise Windows users
that if they don't know what they are doing they should just install
Python(x,y). For that reason I would give priority to maintaining a
perfectly functioning up-to-date Windows version.

Gabriel

m.s.

unread,
Jun 10, 2008, 12:30:28 PM6/10/08
to python(x,y)
> I do not understand what you are asking me to do here.

Nothing. I was talking to the python(x,y) developer.

> Please provide
> a specific example of a python(x,y) package to be tested on Linux.
> Let's take an easy_install example : wouldn't it be nice to have some
> bundle pythonXY that when used with 'easy_install pythonXY' downloads
> all the correct packages that currently make up for the windows
> pythonXY bundle?


No, it wouldn't be nice, because it would fight badly with the
standard packaging system(s).
It would lead to more problems than good.

> I agree that this is completely different than what
> is been done currently, but I just want to point out that a clean easy
> way to retrieve a cross-validated set of all these packages does not
> exist on Linux, to the best of my knowledge.

You already have most,if not all, packages of python(x,y) included in
standard Linux distributions. Installing five or ten of them from a
package manager is not harder than installing just one. You tick the
packages you want on the package manager, you click "install", all
packages are installed.

The best you can do probably is to create a python(x,y) *metapackage*,
that by itself contains nothing but just pulls down python(x,y)
packages as dependencies (say, like the "ubuntu-desktop" debian
package on ubuntu linux).

What could be useful (and I was asking) is collaborating with
distributions so to be sure that versions of the packages included in
standard distributions are in sync and work nicely together. This
would meet the python(x,y) goals without conflicting with the standard
package management.

> And I would add as well
> that I work with some people who do not have root access, so that a
> bundle with a separate python package would also be a plus for
> them....

This is something I hadn't considered and that may be worthwile, if
there is a significant number of users in this need.

m.

Claude

unread,
Jun 10, 2008, 1:10:43 PM6/10/08
to python(x,y)
Hi all,
I proposed Pierre to take care of the "Linux side" of Python(xy).
A difficult task indeed. I'm not a "linux gourou" nor a "scientific
computing gourou". I have just some experience in both ...

So ...
I agree with Gabriel in the sense :
> ... I think that indeed most Linux people will figure out how to install
> packages themselves.

I have myself installed a "Python(xy)-Linux-Equivalent" on an UBUNTU
distribution without any difficulty and, Johann, the apt packager take
care itself of the good interplay between all individual packages.
It's not that evident on an winXP machine ! Here is the challenge!

To answer Johann,

> But if you care about us (and I hope so!), you could help testing of
> the packages in Python(x,y) on Linux, and provide advice to the major
> distributions (Ubuntu, Fedora, Gentoo) on what versions bundle together.

Of course, we care about everybody here ... but :
I have myself some experience on Ubuntu where python is specially well
managed, on Fedora limited to C5, because of migration towards Ubuntu,
and RedHat, at work.
I can give some opinion about these, but for sure, I won't install
Gentoo to give an opinion. I would prefer someone here using Gentoo,
to give is own opinion about it.

Now, about Linux in general the real difficulty is to pick out a
minimum set of package to do something. This minimum depends on the
"something", and "minimum" is user-dependent also.
I can give some experience about driving matlab-users towards python,
as an example to fix the "something", the "minimum" would be ipython
at least . If your minimum is
an Eclipse workbench, I have no special experience but there is no
difficulty to install it ===> apt-cache search eclipse .

Regards
Claude

rese

unread,
Jun 12, 2008, 6:02:23 AM6/12/08
to python(x,y)
Hi

I am not sure I understand the reservation you have to have a linux
version of that bundle.
It is critical!

I am using Suse Linux (opensuse in fact), and it is a NIGHTMARE to
keep up with updating all these packages or even installing some of
them, EVEN IF some of these (or many of these) are available as Suse
compatible rpm versions. I am reinstalling all these things on
different computers (since I have to manage 3 or 3 PCs) and this is
really a nightmare to keep the same versions etc.

So a BUNDLE including all these packages with an update of the
versions of each sub package is VERY USEFUL and remove the need to go
and check each subpackages and dozens of web pages to see what is what
and which version evolved.

Just to conclude: 2 things are usually keeping people out of python
for scientific professional use:
1- documentation (this is an open issue)
2- the fact that there is NO bundle ready to be installed.

I recently convince someone to switch to python EXACTLY because
Python(x,y) was available finally (fantastic work!) and fortunately
that guy was working with Windows... But with Linux, this argument
fails miserably.

So yes please!!! if you could distribute this with a Linux version, I
(and many other people) would be grateful

Eric

m.s.

unread,
Jun 12, 2008, 1:13:48 PM6/12/08
to python(x,y)
> I have myself installed a "Python(xy)-Linux-Equivalent" on an UBUNTU
> distribution without any difficulty and, Johann, the apt packager take
> care itself of the good interplay between all individual packages.
> It's not that evident on an winXP machine ! Here is the challenge!

Exactly. Unix package managers solve exactly the problems that
Python(x,y) solves on Windows (as far as I understand).

That's why a Python(x,y) could be nicely crafted as simple
metapackage(s) that pulls down dependencies from the standard package
manager. It would be relatively little effort (mostly writing text
files, AFAIK) and helpful. Creating new packages from zero is much
more effort, more risky and not significantly more helpful, IMHO.

> > But if you care about us (and I hope so!), you could help testing of
> > the packages in Python(x,y) on Linux, and provide advice to the major
> > distributions (Ubuntu, Fedora, Gentoo) on what versions bundle together.
>
> Of course, we care about everybody here ... but :
> I have myself some experience on Ubuntu where python is specially well
> managed, on Fedora limited to C5, because of migration towards Ubuntu,
> and RedHat, at work.
> I can give some opinion about these, but for sure, I won't install
> Gentoo to give an opinion. I would prefer someone here using Gentoo,
> to give is own opinion about it.

I am a Gentoo user, and I never had serious problems with versions of
scipy/numpy/matplotlib there maintained (at worst, I had to unmask
some unstable version). People managing the scipy/numpy/matplotlib
packages do maintain a tree of dependencies. What you can do is just
looking at the package repositories of major distributions and saying
things like "hey!in my experience, version a.b of A does not match
very well version c.d of B due to bug Z!" to package maintainers -and
providing the metapackages.

> Now, about Linux in general the real difficulty is to pick out a
> minimum set of package to do something. This minimum depends on the
> "something", and "minimum" is user-dependent also.
> I can give some experience about driving matlab-users towards python,
> as an example to fix the "something", the "minimum" would be ipython
> at least . If your minimum is
> an Eclipse workbench, I have no special experience but there is no
> difficulty to install it ===> apt-cache search eclipse .

Exactly. It seems the Python(x,y) package collection is nice (for me,
especially if there is Wx too along with Qt). I don't know if I'd
personally bundle Eclipse (tried to use it, didn't like it, came back
to simple Kate editor) but it's not going to hurt either.


Answering to rese:

>I am using Suse Linux (opensuse in fact), and it is a NIGHTMARE to
>keep up with updating all these packages or even installing some of
>them, EVEN IF some of these (or many of these) are available as Suse
>compatible rpm versions. I am reinstalling all these things on
>different computers (since I have to manage 3 or 3 PCs) and this is
>really a nightmare to keep the same versions etc.

Why is it a nightmare? It really seems it's a problem with you not
using the package management properly / SuSE packages being bad.
I manage three computers, two Gentoo Linux and a Kubuntu one, plus an
OS X partition, and I've no problems in managing my scipy/numpy/mpl/
wxpython combo on the Linux systems. Could you please tell what
specific troubles are you meeting?

Problem for me is OS X, where it is PURE HELL. Guys, seriously: if you
can do something to streamline OS X installs, *do it* and I'll bow to
you forever :)

>So a BUNDLE including all these packages with an update of the
>versions of each sub package is VERY USEFUL and remove the need to go
>and check each subpackages and dozens of web pages to see what is what
>and which version evolved.

You are asking for a metapackage, not a bundle. Add +1 to the
metapackage request

>But with Linux, this argument
>fails miserably.

It fails miserably, indeed, if you use source packages.
Use your package manager (or tell your package repository maintainers
to include packages you need).

I started using Linux, almost five years ago, also for the *ease* of
installing programs on it using package managers. No more web
searches, cds, anti-piracy measures and so on: just fire up Synaptic,
tick'em, install...

m.

Mike Sarahan

unread,
Jun 12, 2008, 2:20:09 PM6/12/08
to python(x,y)
I support the metapackage idea. I'm not sure if Fedora has an
equivalent, but it makes tying things together in Ubuntu/Debian a real
breeze. Didn't Fedora 9 have some newfangled package system that did
both RPM and APT?

I also agree with the no easy-install bit - having more than one tool
updating things gets you into hot water fast.

-Mike

rese

unread,
Jun 13, 2008, 4:13:46 AM6/13/08
to python(x,y)
Hi again,

> Why is it a nightmare? It really seems it's a problem with you not
> using the package management properly / SuSE packages being bad.
> I manage three computers, two Gentoo Linux and a Kubuntu one, plus an
> OS X partition, and I've no problems in managing my scipy/numpy/mpl/
> wxpython combo on the Linux systems. Could you please tell what
> specific troubles are you meeting?

The problems I have come from different sources:
- packages not being upated regurlarly enough. For ex. OpenSuse
science repository has mpl 0.91... I found a bug in this package and I
had then to rely on the source distribution which itself did not work
with the rpm based distrib of numpy/scipy for which bugs were
corrected later on... This is why it is very nice/critical to have a
maintained metapackage which we can easily install. Of course another
way is to have OpenSuse doing a better job. But then this is in some
way what a metapackage could do for me, and here is the next pb:

- I am often looking for extra packages which are NOT on the science
repository for openSuse. This means sometimes relying on sources
distrib (because for some reason the x86_64 distrib is not working or
requires libraries which are not compatible with my own installation).
And this is where it becomes a nightmare. First to look for the right
package. Then realising you need to change the version of another lib
(sometimes downgrading..) and then everything breaks down. I also
stopped looking for some useful modules (3D rendering/plotting, ide,
etc) because it was too complicated to install (you often have to
install many other subpackages and then it is difficult to keep track
of these to maintain 3 different machines)

- finally the system is slightly different on each of my PC because I
have one 32b and 2 64b

So a metapackage which is a self-consistent robust distrib (and easy
to install, e.g. rpm or equivalent) would very probably 1- minimise
any need to tune my own distrib 2- help providing packages that I
often have pb to install/upgrade 3- minimise compatibility pb 4- allow
me to advertise python for science (and therefore help me to then be
able to share modules, etc)

> It fails miserably, indeed, if you use source packages.
> Use your package manager (or tell your package repository maintainers
> to include packages you need).
could do that indeed, but havign a consistent metapackage would solve
the pb too...:-)

Eric

Nicolas

unread,
Jun 13, 2008, 10:00:38 AM6/13/08
to python(x,y)
Hi everybody,
Well, I find all this very interesting, because the main reason why I
switched from Suse to Ubuntu (two years ago) was that installing Scipy
and friends was a nightmare on Suse, and a breeze on Ubuntu!
So that may explain the different perceptions between Suse users and
Gentoo/Ubuntu/Debian/etc users on this issue.
Suse is great otherwise, but the availability of many science packages
in the repos is very important to me, and Debian or Ubuntu are very
good in that respect, IMHO.
NC

m.s.

unread,
Jun 13, 2008, 11:58:50 AM6/13/08
to python(x,y)
> Well, I find all this very interesting, because the main reason why I
> switched from Suse to Ubuntu (two years ago) was that installing Scipy
> and friends was a nightmare on Suse, and a breeze on Ubuntu!
> So that may explain the different perceptions between Suse users and
> Gentoo/Ubuntu/Debian/etc users  on this issue.

Probably. Now I understand rese troubles, and they all seem to boil
down to bad maintainance of scipy-related packages to SuSE. Perhaps
someone should tell that in Suse bugzilla or what else it is. In
Debian-based distros and Gentoo (the two I use the most) I've never
met problems. I have an undergraduate that uses my software on Fedora,
and also for him installing the required scipy/numpy/mpl dependencies
was just a matter of pulling down a few packages from his package
manager, but that's as much testing as I've got on that distribution.

m.

Vinzent Steinberg

unread,
Jun 14, 2008, 7:23:01 AM6/14/08
to python(x,y)


On Jun 13, 4:00 pm, Nicolas <nicolas.chopin...@gmail.com> wrote:
> Hi everybody,
> Well, I find all this very interesting, because the main reason why I
> switched from Suse to Ubuntu (two years ago) was that installing Scipy
> and friends was a nightmare on Suse, and a breeze on Ubuntu!
> So that may explain the different perceptions between Suse users and
> Gentoo/Ubuntu/Debian/etc users  on this issue.
> Suse is great otherwise, but the availability of many science packages
> in the repos is very important to me, and Debian or Ubuntu are very
> good in that respect, IMHO.
> NC
>

SUSE has a build service [1], so it should be possible to set it up
properly to work nicely for all needed packages. It creates
automatically packages for many different distributions [2] (and it
can access VCS afaik to compile bleeding edge stuff). The main problem
is to get everything working. There is for example already a science
repository [3] for SUSE only.

[1] https://build.opensuse.org/
[2] http://software.opensuse.org/search
[3] http://download.opensuse.org/repositories/science/openSUSE_10.3/

Vinzent

Pierre Raybaut

unread,
Jun 14, 2008, 12:51:54 PM6/14/08
to python(x,y)
Hi all,

Are you familiar with "Conduit.Deploy"? [*]

Apparently that is an installation client available for Linux (Ubuntu
compatible), Windows XP and Mac OS X which handles deployment tasks
(installing/uninstalling) as well as package dependencies.

However it's still in development.

[*] http://conduit.simula.no/

Pierre

Ondrej Certik

unread,
Jun 15, 2008, 5:34:39 PM6/15/08
to pyth...@googlegroups.com

Well, easy answer is to use Debian (or Ubuntu). :) One nice thing
about it is that there
are hundreds of Debian developers and maintainers over the world, if I
am not mistaken,
in the last month 500 different people had their packages uploaded and
you can easily be one too,

I myself got fedup with numpy+scipy+matplotlib uninstallable in Debian
last year, so that was my own motivation to become
a Debian Maintainer, so now if something doesn't work in these
packages, I simply fix it in Debian itself
and then all my computers are fixed, no matter if 32 or 64 bits.

So if you like opensuse, I suggest you do the same, i.e. become an
opensuse developer and that's it.
I myself chose Debian becase I think it has the highest number of
developers and there are a lot of people helping
to maintain numpy, scipy, matplotlib, and I also like to work with
other people on the packages, not just myself.

Ondrej

rese

unread,
Jun 16, 2008, 3:37:10 AM6/16/08
to python(x,y)
Hi,
thanks for all the feedback.

I think OpenSuse is an excellent distrib (although it took some time
to reach the level they have now with the 10.3 - still need to have a
look at the 11.0), and indeed I mentioned the science repository in my
email. I guess a better list of packages as well as a more regularly
updated set would solve most of the problems I have.
I am not sure how to proceed there in fact, i.e. to put the right and
"positive" pressure on people to deliver such a service.
(I'll have a look at the "build" Suse site)

Anyway, no question for me to become a developer. This would be in
contradiction with the fact that I wish to spend LESS time to maintain
all this. I know someone has to do the work, but I cannot afford to do
it myself these days.
So either it shows up (and my mail was to motivate the need for a nice
metapackage under linux), or I just will continue wasting some time
and more importantly only using the minimal set of packages (numpy/
scipy/mpl/Ipython), which I think, would be a shame (but well, cannot
blame anyone there).

thanks and cheers

Eric
P.S.: I did not switch to Debian because I feel it is still too much
oriented towards "more expert" linux users, not that it is not user-
friendly, etc, but OpenSuse is an excellent compromise where you have
a relatively easy way of maintaining the system but where you can also
"twiggle" things if needed (as root). (I guess this argument is only
valid because I am an experienced Suse user, and doing the same with
Debian or Ubuntu would only take me some months of getting used to the
configs).

Ondrej Certik

unread,
Jun 16, 2008, 7:27:41 AM6/16/08
to pyth...@googlegroups.com

Just a note: if you are an experienced user, then it will just take
you a couple of minutes to fix things,
so if you were a developer, you could not only fix it for yourself,
but also for all the other people
using opensuse. I do this for Debian and in return, the other Debian
developers help me when I need
it, so in the end, it is an efficient way of doing things.

Ondrej

Reply all
Reply to author
Forward
0 new messages