As a power user what problems can I face with testing?

Here you can discuss every aspect of Debian. Note: not for support requests!

As a power user what problems can I face with testing?

Postby prahladyeri » 2016-02-18 01:48

I always get a mixed combination of replies when I ask people whether I should use a Debian stable or testing? I've been using Ubuntu and/or Debian since years and I want to switch to a rolling release instead of downloading a new distro and formatting my machine whenever a new "stable" release comes. I don't even want to use GNOME, all I want to use are these software packages:

1. xorg, xdm, openbox, lxpanel, network-manager.
2. firefox, thunderbird, libreoffice.
3. geany, glade, virtualbox, emacs, vim.
4. ssh client, mysql, postgresql, php, python.
5. evince, calibre.

If I switch to stretch and then be on a rolling release, say upgrading every 15 days or so, then will I generally have a stable experience with this setup?

Also, what are the security implications of using testing? I've heard that testing receives security-updates too late. For example, only yesterday [this](https://news.ycombinator.com/item?id=11122377) vulnerability in glibc was declared on hacker news. How long for the patch to be developed and reach testing from now? And if I were a testing user, what should I do until that time?
prahladyeri
 
Posts: 42
Joined: 2012-07-11 20:38
Location: India

Re: As a power user what problems can I face with testing?

Postby sgosnell » 2016-02-18 03:38

Testing occasionally breaks, with the effects being anything from inability to boot to a package not running, or almost anything else. Fixes can take days or weeks. Borkage is expected and tolerated. It's Testing, after all. If you absolutely have to have your computer working and available at all times, don't install Testing. If you can afford to have it not working for several days, then jump in and help provide support. You will not have a "stable experience" with Testing in the long run. It may be stable for day, weeks, or even months, but eventually it will break. If that's ok with you, fine. But do not complain when it happens, because it will happen. I run Sid, aka Unstable, because I find breakage in it is fixed much more quickly than in Testing. But I only run it on machines I can afford to have unbootable for awhile. On machines on which I truly rely, I run Stable, because it works every time. Again, do not run Testing on a machine you can't afford to have not working. Do not run Unstable on a machine you can't afford to have not working. But if you can accept the occasional breakage, and live without the computer for some days, Testing is fine.
sgosnell
 
Posts: 715
Joined: 2011-03-14 01:49

Re: As a power user what problems can I face with testing?

Postby kedaha » 2016-02-18 05:48

prahladyeri wrote:I don't even want to use GNOME, all I want to use are these software packages:
[...]
2. firefox, thunderbird[...]

Firefox or Thunderbird are not available from the official Debian repository which only provides the free fork versions, namely, Iceweasel and Icedove so if you prefer the Mozilla versions, then why not stick with Ubuntu for these and other more recent software packages?
Nothing to add to sgonell's reply except to quote Osamu Aoki's "Debian Reference" at life_with_eternal_upgrades:
Even though the unstable suite of the Debian system looks very stable for most of the times, there have been some package problems on the testing and unstable suite of the Debian system and a few of them were not so trivial to resolve. It may be quite painful for you. Sometimes, you may have a broken package or missing functionality for a few weeks.
Mate DE & OSSv4.
FreedomBox in Debian
ispmail
Debian Stable

Words, as is well known, are the great foes of reality. Joseph Conrad.
Kedaha's Conjecture
User avatar
kedaha
 
Posts: 3036
Joined: 2008-05-24 12:26

Re: As a power user what problems can I face with testing?

Postby Hallvor » 2016-02-18 06:59

[...]I want to switch to a rolling release instead of downloading a new distro and formatting my machine whenever a new "stable" release comes


The whole "I don't ever have to reinstall if I use a rolling release"-argument is flawed - at least when it comes to Debian: If you stick with stable, don't install packages from sid or experimental and read the release notes before upgrading, you should be able to upgrade flawlessly from one stable release to the next.

The oldest Debian install in this house is from 2011, I think - dist-upgraded several times.

Most rolling release distros out there would be hosed up and reinstalled several times in five years.
Last edited by Hallvor on 2016-02-18 07:00, edited 1 time in total.
Lenovo Thinkpad T440S, Intel Core i7-4600U CPU @ 2.10GHz, 8 GB RAM, 256 GB SSD, Debian Buster (KDE)
Lenovo Thinkpad X240, Intel Core i5-4300U CPU @ 2.90GHz, 8 GB RAM, 120 GB SSD, Debian Buster (KDE)
User avatar
Hallvor
 
Posts: 997
Joined: 2009-04-16 18:35
Location: Norway

Re: As a power user what problems can I face with testing?

Postby thanatos_incarnate » 2016-02-18 07:00

kedaha wrote:Firefox or Thunderbird are not available from the official Debian repository which only provides the free fork versions, namely, Iceweasel and Icedove so if you prefer the Mozilla versions, then why not stick with Ubuntu for these and other more recent software packages?

You make it sound as though Firefox and Thunderbird were drastically different from Iceweasel and Icedove, which is bogus. They have different names and slightly different packaging. While Icedove doesn't actually track the latest versions, it gets the latest security patches. Iceweasel is available in the ESR version, but the Debian Mozilla repo gives you the latest Iceweasel as well.
User avatar
thanatos_incarnate
 
Posts: 717
Joined: 2012-11-04 20:36

Re: As a power user what problems can I face with testing?

Postby prahladyeri » 2016-02-18 08:24

sgosnell wrote:Testing occasionally breaks, with the effects being anything from inability to boot to a package not running, or almost anything else. Fixes can take days or weeks. Borkage is expected and tolerated. It's Testing, after all. If you absolutely have to have your computer working and available at all times, don't install Testing. If you can afford to have it not working for several days, then jump in and help provide support. You will not have a "stable experience" with Testing in the long run. It may be stable for day, weeks, or even months, but eventually it will break. If that's ok with you, fine. But do not complain when it happens, because it will happen. I run Sid, aka Unstable, because I find breakage in it is fixed much more quickly than in Testing. But I only run it on machines I can afford to have unbootable for awhile. On machines on which I truly rely, I run Stable, because it works every time. Again, do not run Testing on a machine you can't afford to have not working. Do not run Unstable on a machine you can't afford to have not working. But if you can accept the occasional breakage, and live without the computer for some days, Testing is fine.


In general, can you specify whether the borkage is more in the kernel-land or user-land? Also, my upgrade strategy is something like this: I upgrade roughly once a month and before doing it take the entire backup of /var/cache/apt/archives/*.deb and add it to my separate offline repo, so if something goes wrong, I can always remove everything and reinstall. I'm also fine with editing the configuration files in /etc/*, changing dconf settings, etc. and compiling a specific package from source and installing using make install. And of course, I can come here and post any questions if something went wrong with a specific package. With that kind of mindset, how can I fare with testing? Will it still be tough?

My only reason for being on testing is that I like to have the latest that things like LibreOffice, GIMP, etc. provides, even if that comes at the cost of a slight annoyance or borkage. I would have gone with Fedora, but I like the Debian based distros more and I'm also familiar with the apt subsystem and the "Debian way" so I don't want to go that route.
prahladyeri
 
Posts: 42
Joined: 2012-07-11 20:38
Location: India

Re: As a power user what problems can I face with testing?

Postby kedaha » 2016-02-18 10:45

thanatos_incarnate wrote:You make it sound as though Firefox and Thunderbird were drastically different from Iceweasel and Icedove, which is bogus. They have different names and slightly different packaging. While Icedove doesn't actually track the latest versions, it gets the latest security patches. Iceweasel is available in the ESR version, but the Debian Mozilla repo gives you the latest Iceweasel as well.

It wasn't my intention; whilst it's true that they're basically the same, albeit with different names, logos & packaging, both Firefox & Thunderbird - if these versions are preferred - require a different installation procedure since they're not included in Debian repositories so, for clarity, it would be best not to use the names interchangeably.
prahladyeri wrote:My only reason for being on testing is that I like to have the latest things

This may be your reason but the purpose of testing is not to provide the latest things but to develop and test the next stable version of Debian.
Mate DE & OSSv4.
FreedomBox in Debian
ispmail
Debian Stable

Words, as is well known, are the great foes of reality. Joseph Conrad.
Kedaha's Conjecture
User avatar
kedaha
 
Posts: 3036
Joined: 2008-05-24 12:26

Re: As a power user what problems can I face with testing?

Postby fireExit » 2016-02-18 17:32

prahladyeri wrote:If I switch to stretch and then be on a rolling release

testing is not a real rolling-release; you need to consider the deep-freeze: more or less every 2 years, testing development stops for at least 6-8 months (and the wheezy deep-freeze was almost 1 year long)
https://wiki.debian.org/DebianReleases# ... statistics

prahladyeri wrote:what are the security implications of using testing?

if security is paramount, testing is a bad option
https://www.debian.org/security/faq#testing
http://secure-testing-master.debian.net/
User avatar
fireExit
 
Posts: 559
Joined: 2014-11-20 11:22

Re: As a power user what problems can I face with testing?

Postby stevepusser » 2016-02-18 18:18

If you can live with the stable base packages, such as libc6, Python, GTK+3, and so on, almost all shiny new stuff in upstream Debian currently can be backported to Jessie. Many GNOME/GTK+3 developers seem not to care much about making their code backwards-compatible, and their apps demand a newer version than Jessie's 3.14, but so far I haven't run into any Qt programs that demand a newer version than Jessie's.

That's basically what we're doing with MX 15, retaining the stable base while keeping the userspace applications current--but the user can also choose not to upgrade any applications and use the Jessie versions if they want.

I'm not sure exactly you are going for with "kernelspace", I'm running a backported 4.4 Liquorix kernel on a Jessie base, and jessie-backports has a 4.3 kernel right now. You can have as many different kernels installed as you have room for.
The MX Linux repositories: Backports galore! If we don't have something, just ask and we'll try--we like challenges. New packages: Kodi 18.5, Blender 2.8.2, 5.5 kernels, Openshot 2.5.0, Telegram-desktop 1.9.14, LibreOffice 6.4.1~rc1
User avatar
stevepusser
 
Posts: 11468
Joined: 2009-10-06 05:53

Re: As a power user what problems can I face with testing?

Postby dasein » 2016-02-18 22:02

@OP: You say in so many words that you want a rolling release with the latest software.

Here's the problem: Debian Testing is neither of those things.

What you're saying, even if you don't realize it, is that you want to run Arch. And it will feel so much faster (not to mention trendier, cooler, etc.)
User avatar
dasein
 
Posts: 7775
Joined: 2011-03-04 01:06
Location: Terra Incantationum

Re: As a power user what problems can I face with testing?

Postby Innovate » 2016-02-19 03:19

Seriously, I've less problem with Stretch Testing more than Jessie Stable.
On Jessie I can't find the way to solve wallpaper reset itself every time I restart/shut down my pc on XFCE 4.10 session
On Stretch I've already fixed that XFCE4.12 workspace too long by just turn off the dual monitor.
Overall I don't have much problem with Stretch I used to have more living hell on Jessie Stable.

I also start to wondering why Debian unstable & Arch bleeding edges are treat so differently.
Arch ppl always boast about their stable & bleeding edge releases softwares.
If Debian Testing that scary wouldn't Arch bleeding edge more horrifying?
Innovate
 
Posts: 188
Joined: 2015-12-27 01:28
Location: /dev/urandom

Re: As a power user what problems can I face with testing?

Postby mor » 2016-02-19 09:21

Innovate wrote:If Debian Testing that scary wouldn't Arch bleeding edge more horrifying?

One thing that I've been suggesting a lot lately, is to not just depict Testing/Unstable as simply scary or dangerous, because although they can be, and for most reckless and clueless users will eventually, the real main issue with them is that they are something different from Stable, and not a scarier or more dangerous version of.

Testing and Unstable are development branches and as such operate under a different paradigm. This paradigm is made of frequent updates/upgrades whereas on Stable the exact opposite is true.
The Debian project engineered the two development branches for the purpose of having a working ground to setup and organize the packages for each next version, the fact that these two branches are "runnable" doesn't necessarily mean that they are supposed to be run for production (even if and when some, like yours truly, do), an definitely doesn't mean that they inherit all the characteristics of Stable just because they are called Debian too.

Think of Testing and also Unstable not as the parents of a Stable release, who beget a child which is a lot like the parents.
Think of Unstable and Testing as the caterpillar that will eventually morph into a butterfly.
A caterpillar is not necessarily scarier than the butterfly, but it definitely isn't the same thing, not even similar.

For this reason, all those who move to Testing/Unstable for reasons such as the desire for newer version of some or all applications, are misguided under the false assumption that the system will also still retain the properties of solidness and dependability that Stable is renowned for, even if just to a lesser extent.
But this assumption is wrong because Stable is renowned for its dependability and solidness exactly because of the way it is engineered, that is "by not changing anymore after it is thoroughly tested and then released".
If you remove or alter this paradigm, which is exactly what Testing and Unstable do (read link about the Concept of Stability in my signature), then you lose the true essence of Stable, thus of Debian, and you might as well be on another system.

Which is why suggesting Arch, or Fedora and openSUSE to a lesser extent, which are distros designed to follow upstream development up close, is not at all wrong, is indeed exactly right for those who prefer the new version of everything.

Why isn't running Arch more horrifying?
A certain risk is always implied with bleeding edge distros (and, by the way, Unstable is generally a few steps behind these distros, so much for having the latest and greatest!), exactly because the software hasn't been tested long enough like it is the case for Stable, as well as other distributions that work under the same paradigm.
But unlike Debian's development branches, bleeding edge distros are designed to be "Official Releases" based on software that evolves constantly, bringing in solutions and fixes as fast as bugs and other changes.
Testing/Unstable are not "Official Releases", they have the sole objective to prepare a new Stable Release; they don't mind leaving packages and bugs hanging for a while, or remove them because maybe the license has issues, or because a transition is too big to do overnight.

Well, ultimately every user will do as they please, but Debian is exemplified by the official Stable release, that is exactly what Debian's operating system and computing paradigm is: a thoroughly tested, unchanging and low maintenance system, and if Stable doesn't suit one's need, time to look further, Testing and Unstable are not the solution.

Bye
User avatar
mor
 
Posts: 970
Joined: 2010-08-28 15:16
Location: mor@debian

Re: As a power user what problems can I face with testing?

Postby Innovate » 2016-02-19 10:12

mor wrote:
Innovate wrote:If Debian Testing that scary wouldn't Arch bleeding edge more horrifying?

One thing that I've been suggesting a lot lately, is to not just depict Testing/Unstable as simply scary or dangerous, because although they can be, and for most reckless and clueless users will eventually, the real main issue with them is that they are something different from Stable, and not a scarier or more dangerous version of.

Testing and Unstable are development branches and as such operate under a different paradigm. This paradigm is made of frequent updates/upgrades whereas on Stable the exact opposite is true.
The Debian project engineered the two development branches for the purpose of having a working ground to setup and organize the packages for each next version, the fact that these two branches are "runnable" doesn't necessarily mean that they are supposed to be run for production (even if and when some, like yours truly, do), an definitely doesn't mean that they inherit all the characteristics of Stable just because they are called Debian too.

Think of Testing and also Unstable not as the parents of a Stable release, who beget a child which is a lot like the parents.
Think of Unstable and Testing as the caterpillar that will eventually morph into a butterfly.
A caterpillar is not necessarily scarier than the butterfly, but it definitely isn't the same thing, not even similar.

For this reason, all those who move to Testing/Unstable for reasons such as the desire for newer version of some or all applications, are misguided under the false assumption that the system will also still retain the properties of solidness and dependability that Stable is renowned for, even if just to a lesser extent.
But this assumption is wrong because Stable is renowned for its dependability and solidness exactly because of the way it is engineered, that is "by not changing anymore after it is thoroughly tested and then released".
If you remove or alter this paradigm, which is exactly what Testing and Unstable do (read link about the Concept of Stability in my signature), then you lose the true essence of Stable, thus of Debian, and you might as well be on another system.

Which is why suggesting Arch, or Fedora and openSUSE to a lesser extent, which are distros designed to follow upstream development up close, is not at all wrong, is indeed exactly right for those who prefer the new version of everything.

Why isn't running Arch more horrifying?
A certain risk is always implied with bleeding edge distros (and, by the way, Unstable is generally a few steps behind these distros, so much for having the latest and greatest!), exactly because the software hasn't been tested long enough like it is the case for Stable, as well as other distributions that work under the same paradigm.
But unlike Debian's development branches, bleeding edge distros are designed to be "Official Releases" based on software that evolves constantly, bringing in solutions and fixes as fast as bugs and other changes.
Testing/Unstable are not "Official Releases", they have the sole objective to prepare a new Stable Release; they don't mind leaving packages and bugs hanging for a while, or remove them because maybe the license has issues, or because a transition is too big to do overnight.

Well, ultimately every user will do as they please, but Debian is exemplified by the official Stable release, that is exactly what Debian's operating system and computing paradigm is: a thoroughly tested, unchanging and low maintenance system, and if Stable doesn't suit one's need, time to look further, Testing and Unstable are not the solution.

Bye

Thanks for clarified details I get the picture now. :D
I see, So the heart of the point is official releases. That's sounds trickery.
Innovate
 
Posts: 188
Joined: 2015-12-27 01:28
Location: /dev/urandom

Re: As a power user what problems can I face with testing?

Postby kedaha » 2016-02-19 10:55

I wonder if the term power user - while applicable to Windows up to Vista - is appropriate for users of Debian or other Unix-like systems but one can hardly declare, "I am a root user" or "I am a superuser" to claim expert status. :mrgreen:
Mate DE & OSSv4.
FreedomBox in Debian
ispmail
Debian Stable

Words, as is well known, are the great foes of reality. Joseph Conrad.
Kedaha's Conjecture
User avatar
kedaha
 
Posts: 3036
Joined: 2008-05-24 12:26

Re: As a power user what problems can I face with testing?

Postby mor » 2016-02-19 14:30

Little premise:
I always use the uppercase "S" in "Stable" to indicate the Stable release of Debian's Operating System, and the lowercase "s" in "stable" to indicate the property of being unchanging.
Stable=Debian Stable currently Jessie
stable=unchanging, can be a system or anything.

Same goes for Testing and testing or Unstable and unstable.

Innovate wrote:Thanks for clarified details I get the picture now. :D
I see, So the heart of the point is official releases. That's sounds trickery.

No, the point is that Testing/Unstable are completely different from Stable and not, like most mistakenly end up understanding, some sort of linear prosecution from Stable.

The point was to explain that all those who come to Debian usually do so because Debian has a reputation of being solid and reliable, and then when they discover that Debian operates on super old, super tested software (which is exactly the point and the strength of Debian) they do the magical mental leap and convince themselves that shifting to development branches of Debian will give them newer software while at the same time retaining the solidness and reliability of Stable, as if it was a characteristic inherent to the Debian brand.

This is false, but not in the sense that, like many seem to only focus on,Testing/Unstable are utterly dangerous and crashy-buggy, but in the sense that one cannot expect them to behave and operate in the same manner as a non-developmental system, and a stable one on top of that (again, read the Concept of Stability in my signature).

Debian Stable's experience doesn't translate into Testing/Unstable's experience "with a few manageable bugs".
Any stable operating system, is one that is supposed to be set and configured once, and then work without hiccups for as long as possible (the release cycle, usually a couple of years in Debian) and with as little maintenance as possible.
Debian Stable is all that, Testing/Unstable are not at all that.

Now the tricky point about "official releases".
While Arch and other bleeding edge systems have in common with Debian's Testing/Unstable the property of being unstable (aka constantly changing), they differ in two fundamental aspects, namely their purpose and the fact that they are not development systems.
Arch and bleeding edge systems are in fact supposed to be production operating systems, even within the confines of an inevitably lesser (when compared to stable systems) reliability due to their unstable nature, and their peculiar purpose is that of offering their users the latest software available at all times (and whether the reasons for wanting this are futile or not doesn't concern me).

Testing/Unstable on the other hand, do constantly get stuff from upstream (not as swiftly by the way), but do it for the purpose of shaking down the bad stuff from it, over a long long period of time, in order to leave only the safe stuff for when the time of a new release will come.
This is an immensely different goal: they don't get newer software for the purpose of providing Testing/Unstable users with the latest (that, again, never is) software available.

Being development branches, Debian's Testing and Unstable do not have any "obligation" towards their users in terms of being usable at all times, they are indeed supposed to break exactly because fixing issues is the goal and finding them is the first step.
They are development systems and not Official Releases.

Bye
User avatar
mor
 
Posts: 970
Joined: 2010-08-28 15:16
Location: mor@debian

Next

Return to General Discussion

Who is online

Users browsing this forum: No registered users and 8 guests

fashionable