Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
How do I build a minimal Linux distro?
53 views
Skip to first unread message
petru...@gmail.com
Oct 8, 2012, 8:45:00 PM10/8/12
to
Hi everyone, I've been looking into Linux from Scratch for the past few months. I was wondering whether or not that was the most simple way to build a bare-bones distribution of Linux. My goal right now is to just get it to boot into a bash shell. Any help would be appreciated. Thanks.
Aragorn
Oct 8, 2012, 11:38:07 PM10/8/12
to
On Tuesday 09 October 2012 02:45, petru...@gmail.com conveyed the
following to alt.comp.os.linux...
LFS is certainly the hands-on-and-get-yourself-dirty approach, but
perhaps you would have been better off looking into either...
° Gentoo;
° Slackware;
° Arch; or
° SourceMage.
Gentoo and SourceMage also start off from a sources-built distribution,
but they allow you to take it anywhere you want from there without
forcing a certain layout on you.
Slackware and Arch are binary-based distributions, but do still allow
you lots of customization.
--
= Aragorn =
(registered GNU/Linux user #223157)
following to alt.comp.os.linux...
perhaps you would have been better off looking into either...
° Gentoo;
° Slackware;
° Arch; or
° SourceMage.
Gentoo and SourceMage also start off from a sources-built distribution,
but they allow you to take it anywhere you want from there without
forcing a certain layout on you.
Slackware and Arch are binary-based distributions, but do still allow
you lots of customization.
--
= Aragorn =
(registered GNU/Linux user #223157)
RayLopez99
Oct 14, 2012, 5:49:32 AM10/14/12
to
On Tuesday, October 9, 2012 3:45:01 AM UTC+3, petru...@gmail.com wrote:
> Hi everyone, I've been looking into Linux from Scratch
I don't know "Scratch" I assume you mean scratch but do yourself a favor and go to Piratebay.se, get Windows 7 warez and go to town. You'll be way ahead in the long and short run from where you'd be in Linux. And nobody will arrest you or sue you, trust me.
> Hi everyone, I've been looking into Linux from Scratch
RL
adiabic
Oct 15, 2012, 10:56:16 AM10/15/12
to
Eyetee
Dec 24, 2012, 10:19:25 PM12/24/12
to
The piratebay fan doesn't state what "social diseases" your OS can
acquire from piratebay. I'm sure having a black hat freely access your
entire home network through a backdoor built into a customized Win 7
.iso is loads of fun.
Eyetee
--
website: www.eyetee.org
Aragorn
Dec 25, 2012, 2:18:41 AM12/25/12
to
On Tuesday 25 December 2012 04:19, Eyetee conveyed the following to
alt.comp.os.linux...
Not to mention that all versions of Windows NT since NT 4.0 have at
least two _identified_ backdoors built into them by default: one for use
by the NSA and one for use by Microsoft.
All versions of Microsoft Windows since NT 5.1 (i.e. Windows XP and 2003
Server) are also known to phone home every week with a detailed report
on what hardware is in the machine, what software is installed on the
machine, and how often that software is used. Microsoft has
acknowledged this but claims that this information is only used for
statistics, plus that, as of Windows 7 (NT 6.1), it is now possible to
disable this "feature" - it is called the "Microsoft User Experience
Program" or something of that sorts.
Anyway, Microsoft is irrelevant. They only owe the success of Windows
on the desktop to their extreme monopolist tactics and unholy alliances
with hardware vendors. In the server rooms of the world, where
operating system quality really matters, Windows is insignificant, and
for good reason too, since it is the worst possible operating system
design on the planet.
alt.comp.os.linux...
least two _identified_ backdoors built into them by default: one for use
by the NSA and one for use by Microsoft.
All versions of Microsoft Windows since NT 5.1 (i.e. Windows XP and 2003
Server) are also known to phone home every week with a detailed report
on what hardware is in the machine, what software is installed on the
machine, and how often that software is used. Microsoft has
acknowledged this but claims that this information is only used for
statistics, plus that, as of Windows 7 (NT 6.1), it is now possible to
disable this "feature" - it is called the "Microsoft User Experience
Program" or something of that sorts.
Anyway, Microsoft is irrelevant. They only owe the success of Windows
on the desktop to their extreme monopolist tactics and unholy alliances
with hardware vendors. In the server rooms of the world, where
operating system quality really matters, Windows is insignificant, and
for good reason too, since it is the worst possible operating system
design on the planet.
Eyetee
Dec 25, 2012, 2:36:51 AM12/25/12
to
On 12-12-25 02:18 AM, Aragorn wrote:
<snippage of good stuff>
> Anyway, Microsoft is irrelevant. They only owe the success of Windows
> on the desktop to their extreme monopolist tactics and unholy alliances
> with hardware vendors. In the server rooms of the world, where
> operating system quality really matters, Windows is insignificant, and
> for good reason too, since it is the worst possible operating system
> design on the planet.
We are very like-minded when it comes to certain things. The problem
with Windows is that the commercial developers who developed and
continue to maintain it work under some severe constraints. First, when
it comes to people knowledgeable about IT and able to give intelligent
feedback, they have a relatively tiny testing and debugging community
(unlike GNU, where the number of qualified developers constantly
testingt and debugging is huge). That is automatically going to lead to
a worse OS. Second, and more importantly, commercial outfits such as
Microsoft must pay heavy mind to marketing considerations. It's no
mystery that technical people working for Microsoft are constantly
grabbing their ankles for the marketing people, and even the planning
stages of development are subject to marketing considerations. Marketing
and tech are involved in a zero-sum game, and the more marketing you
have, the worse your tech is.
I can't tell whether the developers working for Microsoft are bitter
about what they are subjected to in their jobs, especially the
horrendous non-compete agreements they were required to sign in order to
get hired, which make it permanently impossible for them to quit and go
work elsewhere. Maybe they're so brainwashed by the cult-like Microsoft
corporate culture that they think doing development work purely as a
selling tool somehow enhances the technical quality of what's being
developed--which even a sane layman will tell you is ridiculous. But
money talks, so here you and I are exchanging posts on a newsgroup while
nearly a billion lamers who have trouble finding athe power switch are
convinced they know more about IT than we do because they use "what
everybody uses."
One thing you can help me with: my testing of OpenSUSE suggests that
Novell has turned it into a piece of dirt. It looks very much like that
distribution doesn't let you have access to X11 functionality without
also installing a GUI because Novell made X11 dependent on GUI in order
to run. Which gives rise to the ridiculous situation of needing GUI
functionality in order to launch a graphical application from CLI. Let
me know whether I've got that right.
Eyetee
--
website: www.eyetee.org
<snippage of good stuff>
> Anyway, Microsoft is irrelevant. They only owe the success of Windows
> on the desktop to their extreme monopolist tactics and unholy alliances
> with hardware vendors. In the server rooms of the world, where
> operating system quality really matters, Windows is insignificant, and
> for good reason too, since it is the worst possible operating system
> design on the planet.
with Windows is that the commercial developers who developed and
continue to maintain it work under some severe constraints. First, when
it comes to people knowledgeable about IT and able to give intelligent
feedback, they have a relatively tiny testing and debugging community
(unlike GNU, where the number of qualified developers constantly
testingt and debugging is huge). That is automatically going to lead to
a worse OS. Second, and more importantly, commercial outfits such as
Microsoft must pay heavy mind to marketing considerations. It's no
mystery that technical people working for Microsoft are constantly
grabbing their ankles for the marketing people, and even the planning
stages of development are subject to marketing considerations. Marketing
and tech are involved in a zero-sum game, and the more marketing you
have, the worse your tech is.
I can't tell whether the developers working for Microsoft are bitter
about what they are subjected to in their jobs, especially the
horrendous non-compete agreements they were required to sign in order to
get hired, which make it permanently impossible for them to quit and go
work elsewhere. Maybe they're so brainwashed by the cult-like Microsoft
corporate culture that they think doing development work purely as a
selling tool somehow enhances the technical quality of what's being
developed--which even a sane layman will tell you is ridiculous. But
money talks, so here you and I are exchanging posts on a newsgroup while
nearly a billion lamers who have trouble finding athe power switch are
convinced they know more about IT than we do because they use "what
everybody uses."
One thing you can help me with: my testing of OpenSUSE suggests that
Novell has turned it into a piece of dirt. It looks very much like that
distribution doesn't let you have access to X11 functionality without
also installing a GUI because Novell made X11 dependent on GUI in order
to run. Which gives rise to the ridiculous situation of needing GUI
functionality in order to launch a graphical application from CLI. Let
me know whether I've got that right.
Eyetee
--
website: www.eyetee.org
Aragorn
Dec 25, 2012, 3:20:31 AM12/25/12
to
On Tuesday 25 December 2012 08:36, Eyetee conveyed the following to
alt.comp.os.linux...
> On 12-12-25 02:18 AM, Aragorn wrote:
>
> <snippage of good stuff>
>
>> Anyway, Microsoft is irrelevant. They only owe the success of
>> Windows on the desktop to their extreme monopolist tactics and unholy
>> alliances with hardware vendors. In the server rooms of the world,
>> where operating system quality really matters, Windows is
>> insignificant, and for good reason too, since it is the worst
>> possible operating system design on the planet.
>
> We are very like-minded when it comes to certain things. The problem
> with Windows is that the commercial developers who developed and
> continue to maintain it work under some severe constraints. First,
> when it comes to people knowledgeable about IT and able to give
> intelligent feedback, they have a relatively tiny testing and
> debugging community (unlike GNU, where the number of qualified
> developers constantly testingt and debugging is huge). That is
> automatically going to lead to a worse OS.
The above is one of the big selling points - or at least, in a manner of
speaking - of the whole Free & Open Source Software development model,
or as Linus Torvalds puts it: "More eyes to spot the bugs, and more eyes
to fix them."
Ever since Coverity started with their annual comparisons of the code
quality of proprietary software versus free/open source software, the
latter has always stood out head and shoulders above proprietary
software. As an example, a few years ago, it was found that for an
equal amount of code, proprietary software contained over 600 times as
many bugs as open source code.
> Second, and more importantly, commercial outfits such as Microsoft
> must pay heavy mind to marketing considerations.
Before going on about this particular aspect, I do feel that it is
important to point out that Microsoft is not after the money, as
everyone thinks it is. What Microsoft is after is the absolute power,
due to the proprietary nature of the licensing.
That is why they are now also buying as much so-called "intellectual
property" as they can, and whereas they used to do their patent trolling
by way of sock puppet companies - like the law firm of Bill Gates's
father - they are now also openly coming out of the closet as patent
trolls.
> It's no mystery that technical people working for Microsoft are
> constantly grabbing their ankles for the marketing people, and even
> the planning stages of development are subject to marketing
> considerations. Marketing and tech are involved in a zero-sum game,
> and the more marketing you have, the worse your tech is.
It all comes with the atmosphere. Free/Open Source Software is
developed to work, and to work as best as possible. Proprietary
software is developed so as to sell licenses and provide for legal
leverage to a corporation.
In other words, the proprietary software in itself is only the means to
an end, not the goal itself. That's why the code quality can never be
too high.
> I can't tell whether the developers working for Microsoft are bitter
> about what they are subjected to in their jobs, especially the
> horrendous non-compete agreements they were required to sign in order
> to get hired, which make it permanently impossible for them to quit
> and go work elsewhere. Maybe they're so brainwashed by the cult-like
> Microsoft corporate culture that they think doing development work
> purely as a selling tool somehow enhances the technical quality of
> what's being developed--which even a sane layman will tell you is
> ridiculous.
Microsoft /does/ use brainwashing techniques. All Microsoft upper staff
are required to follow a weekend-long seminar at a company called
Landmark Education, which is an offshoot of an organization - I forgot
its name - which was known for its brainwashing techniques in the mid-
to-late 1970s, and which at the time was very successful among the
yuppies.
> But money talks, so here you and I are exchanging posts on
> a newsgroup while nearly a billion lamers who have trouble finding
> athe power switch are convinced they know more about IT than we do
> because they use "what everybody uses."
I find myself constantly annoyed by how narrow the mind of the average
Win-droid is. They see "the computer" as a household kitchen sink
appliance and nothing more, and they are totally oblivious of any other
type of computer technology.
In addition to that, the poster going by the pseudonym RayLopez99 is a
known Windows troll from comp.os.linux.advocacy, a self-confessed
Microsoft shareholder /and/ a software thief. He also constantly brags
about his IQ, but all of his posts show that he is in fact not quite as
bright as he thinks he is.
> One thing you can help me with: my testing of OpenSUSE suggests that
> Novell has turned it into a piece of dirt. It looks very much like
> that distribution doesn't let you have access to X11 functionality
> without also installing a GUI because Novell made X11 dependent on GUI
> in order to run. Which gives rise to the ridiculous situation of
> needing GUI functionality in order to launch a graphical application
> from CLI. Let me know whether I've got that right.
Hmm... No, openSUSE does allow you to install a minimal X11 environment
with any window manager you like, but you do have to specify it at
installation time, and then you also have to specify that you don't want
to start the GUI by default.
However, things may have changed a little bit in the latest release of
openSUSE due to the push from upstream - read: RedHat - towards a merge
udev and systemd, and systemd does things quite differently from a
traditional System V init. In fact, it does a lot more than a true init
system should be doing. All RedHat derivatives - and openSUSE /is/ a
RedHat derivative - now use systemd and its awkward ways.
That is why, given the requirements you posted earlier, you should stay
away from RedHat derivatives. You want something that sits closer to
the traditional UNIX recipe than to the appliance model, and that means
that you're pretty much limited to Gentoo [*], Arch or Slackware, or
perhaps - if you're adventurous - LFS.
[*] Gentoo uses OpenRC as the default System V init replacement, but it
does also support systemd. However, due to the planned integration
by Sievers and Poettering at RedHat between udev and systemd, a
number of Gentoo developers have now decided to fork udev. The fork
is still in heavy testing phase and is called eudev. It attempts to
provide udev functionality without relying on systemd and to undo
the breakages introduced by udev and systemd, such as the need to
boot with an initramfs if you have /usr split off from the root
filesystem. This breakage was arrogantly introduced by Sievers and
Poettering by making udev and systemd rely on stuff which is located
under /usr, and was then further pushed downstream by RedHat (via
Fedora 17) by making /bin, /lib and /sbin symlinks to their
equivalent directories under the /usr tree.
alt.comp.os.linux...
> On 12-12-25 02:18 AM, Aragorn wrote:
>
> <snippage of good stuff>
>
>> Anyway, Microsoft is irrelevant. They only owe the success of
>> Windows on the desktop to their extreme monopolist tactics and unholy
>> alliances with hardware vendors. In the server rooms of the world,
>> where operating system quality really matters, Windows is
>> insignificant, and for good reason too, since it is the worst
>> possible operating system design on the planet.
>
> We are very like-minded when it comes to certain things. The problem
> with Windows is that the commercial developers who developed and
> continue to maintain it work under some severe constraints. First,
> when it comes to people knowledgeable about IT and able to give
> intelligent feedback, they have a relatively tiny testing and
> debugging community (unlike GNU, where the number of qualified
> developers constantly testingt and debugging is huge). That is
> automatically going to lead to a worse OS.
speaking - of the whole Free & Open Source Software development model,
or as Linus Torvalds puts it: "More eyes to spot the bugs, and more eyes
to fix them."
Ever since Coverity started with their annual comparisons of the code
quality of proprietary software versus free/open source software, the
latter has always stood out head and shoulders above proprietary
software. As an example, a few years ago, it was found that for an
equal amount of code, proprietary software contained over 600 times as
many bugs as open source code.
> Second, and more importantly, commercial outfits such as Microsoft
> must pay heavy mind to marketing considerations.
important to point out that Microsoft is not after the money, as
everyone thinks it is. What Microsoft is after is the absolute power,
due to the proprietary nature of the licensing.
That is why they are now also buying as much so-called "intellectual
property" as they can, and whereas they used to do their patent trolling
by way of sock puppet companies - like the law firm of Bill Gates's
father - they are now also openly coming out of the closet as patent
trolls.
> It's no mystery that technical people working for Microsoft are
> constantly grabbing their ankles for the marketing people, and even
> the planning stages of development are subject to marketing
> considerations. Marketing and tech are involved in a zero-sum game,
> and the more marketing you have, the worse your tech is.
developed to work, and to work as best as possible. Proprietary
software is developed so as to sell licenses and provide for legal
leverage to a corporation.
In other words, the proprietary software in itself is only the means to
an end, not the goal itself. That's why the code quality can never be
too high.
> I can't tell whether the developers working for Microsoft are bitter
> about what they are subjected to in their jobs, especially the
> horrendous non-compete agreements they were required to sign in order
> to get hired, which make it permanently impossible for them to quit
> and go work elsewhere. Maybe they're so brainwashed by the cult-like
> Microsoft corporate culture that they think doing development work
> purely as a selling tool somehow enhances the technical quality of
> what's being developed--which even a sane layman will tell you is
> ridiculous.
are required to follow a weekend-long seminar at a company called
Landmark Education, which is an offshoot of an organization - I forgot
its name - which was known for its brainwashing techniques in the mid-
to-late 1970s, and which at the time was very successful among the
yuppies.
> But money talks, so here you and I are exchanging posts on
> a newsgroup while nearly a billion lamers who have trouble finding
> athe power switch are convinced they know more about IT than we do
> because they use "what everybody uses."
Win-droid is. They see "the computer" as a household kitchen sink
appliance and nothing more, and they are totally oblivious of any other
type of computer technology.
In addition to that, the poster going by the pseudonym RayLopez99 is a
known Windows troll from comp.os.linux.advocacy, a self-confessed
Microsoft shareholder /and/ a software thief. He also constantly brags
about his IQ, but all of his posts show that he is in fact not quite as
bright as he thinks he is.
> One thing you can help me with: my testing of OpenSUSE suggests that
> Novell has turned it into a piece of dirt. It looks very much like
> that distribution doesn't let you have access to X11 functionality
> without also installing a GUI because Novell made X11 dependent on GUI
> in order to run. Which gives rise to the ridiculous situation of
> needing GUI functionality in order to launch a graphical application
> from CLI. Let me know whether I've got that right.
with any window manager you like, but you do have to specify it at
installation time, and then you also have to specify that you don't want
to start the GUI by default.
However, things may have changed a little bit in the latest release of
openSUSE due to the push from upstream - read: RedHat - towards a merge
udev and systemd, and systemd does things quite differently from a
traditional System V init. In fact, it does a lot more than a true init
system should be doing. All RedHat derivatives - and openSUSE /is/ a
RedHat derivative - now use systemd and its awkward ways.
That is why, given the requirements you posted earlier, you should stay
away from RedHat derivatives. You want something that sits closer to
the traditional UNIX recipe than to the appliance model, and that means
that you're pretty much limited to Gentoo [*], Arch or Slackware, or
perhaps - if you're adventurous - LFS.
[*] Gentoo uses OpenRC as the default System V init replacement, but it
does also support systemd. However, due to the planned integration
by Sievers and Poettering at RedHat between udev and systemd, a
number of Gentoo developers have now decided to fork udev. The fork
is still in heavy testing phase and is called eudev. It attempts to
provide udev functionality without relying on systemd and to undo
the breakages introduced by udev and systemd, such as the need to
boot with an initramfs if you have /usr split off from the root
filesystem. This breakage was arrogantly introduced by Sievers and
Poettering by making udev and systemd rely on stuff which is located
under /usr, and was then further pushed downstream by RedHat (via
Fedora 17) by making /bin, /lib and /sbin symlinks to their
equivalent directories under the /usr tree.
Eyetee
Dec 25, 2012, 10:17:17 AM12/25/12
to
On 12-12-25 03:20 AM, Aragorn wrote:
<snippage>
> [*] Gentoo uses OpenRC as the default System V init replacement, but it
> does also support systemd. However, due to the planned integration
> by Sievers and Poettering at RedHat between udev and systemd, a
> number of Gentoo developers have now decided to fork udev. The fork
> is still in heavy testing phase and is called eudev. It attempts to
> provide udev functionality without relying on systemd and to undo
> the breakages introduced by udev and systemd, such as the need to
> boot with an initramfs if you have /usr split off from the root
> filesystem. This breakage was arrogantly introduced by Sievers and
> Poettering by making udev and systemd rely on stuff which is located
> under /usr, and was then further pushed downstream by RedHat (via
> Fedora 17) by making /bin, /lib and /sbin symlinks to their
> equivalent directories under the /usr tree.
Is there some reason for doing things this way? I can't for the life of
me see the logic of relying on symlinks under /usr for /bin, /lib and
/sbin file operations. It's like going around the block to get to the
house next door. And it must play hell with PATH, among other things. If
you have a link to some document where RedHat explains its rationale,
please post it.
I never considered RedHat itself because I don't have local server
requirements and don't want to spend money when I can get the same thing
free of charge, but I now see it isn't the same thing, it's something
that appears to be seriously worse.
Eyetee
--
website: www.eyetee.org
<snippage>
> [*] Gentoo uses OpenRC as the default System V init replacement, but it
> does also support systemd. However, due to the planned integration
> by Sievers and Poettering at RedHat between udev and systemd, a
> number of Gentoo developers have now decided to fork udev. The fork
> is still in heavy testing phase and is called eudev. It attempts to
> provide udev functionality without relying on systemd and to undo
> the breakages introduced by udev and systemd, such as the need to
> boot with an initramfs if you have /usr split off from the root
> filesystem. This breakage was arrogantly introduced by Sievers and
> Poettering by making udev and systemd rely on stuff which is located
> under /usr, and was then further pushed downstream by RedHat (via
> Fedora 17) by making /bin, /lib and /sbin symlinks to their
> equivalent directories under the /usr tree.
me see the logic of relying on symlinks under /usr for /bin, /lib and
/sbin file operations. It's like going around the block to get to the
house next door. And it must play hell with PATH, among other things. If
you have a link to some document where RedHat explains its rationale,
please post it.
I never considered RedHat itself because I don't have local server
requirements and don't want to spend money when I can get the same thing
free of charge, but I now see it isn't the same thing, it's something
that appears to be seriously worse.
Eyetee
--
website: www.eyetee.org
Aragorn
Dec 25, 2012, 12:06:20 PM12/25/12
to
On Tuesday 25 December 2012 16:17, Eyetee conveyed the following to
alt.comp.os.linux...
> On 12-12-25 03:20 AM, Aragorn wrote:
>
>> Gentoo uses OpenRC as the default System V init replacement, but
>> it does also support systemd. However, due to the planned
>> integration by Sievers and Poettering at RedHat between udev and
>> systemd, a number of Gentoo developers have now decided to fork udev.
>>
>> The fork is still in heavy testing phase and is called eudev. It
>> attempts to provide udev functionality without relying on systemd and
>> to undo the breakages introduced by udev and systemd, such as the
>> need to boot with an initramfs if you have /usr split off from the
>> root filesystem. This breakage was arrogantly introduced by Sievers
>> and Poettering by making udev and systemd rely on stuff which is
>> located under /usr, and was then further pushed downstream by
>> RedHat (via Fedora 17) by making /bin, /lib and /sbin symlinks to
>> their equivalent directories under the /usr tree.
>
> Is there some reason for doing things this way? I can't for the life
> of me see the logic of relying on symlinks under /usr for /bin, /lib
> and /sbin file operations.
/bin, /lib and /sbin are the symlinks. They point to /usr/bin, /usr/lib
and /usr/sbin respectively.
> It's like going around the block to get to the house next door. And it
> must play hell with PATH, among other things. If you have a link to
> some document where RedHat explains its rationale, please post it.
It does not affect the $PATH - that's why the symlinks are there, for
backward compatibility. There are many rationales leading to this
decision, but most of them are actually retroactive justification.
One of the rationales is that proprietary UNIX - and most notably
Solaris - also does it this way, and that /usr was already a hack to
begin with in the original AT&T Unix. That part is true. As history
had it, Unix became so big that it wouldn't fit on a single hard disk
anymore at the time, so a second disk was added and its volume was
mounted to /usr.
One of the other arguments is that with everything living under /usr, it
is easier to make backups and to mount /usr read-only and/or over NFS,
but this in itself contradicts the "separate /usr is broken" message
that Poettering and friends sent out into the world.
The whole thing started when udev and systemd began initializing things
which live under /usr - things which are not strictly needed at boot
time except in a few corner cases - and then it got even worse by the
default installation of both udev and systemd under /usr. So the
solution is to use an initramfs with busybox, which can mount /usr
before udev and systemd are started.
There are several articles on the web about this, but most of them are
just retroactive justification of the breakage, i.e. a separate /usr on
a system without an initramfs is broken because Poettering and Sievers
broke it and have no intention of fixing it. And they don't care about
other distributions than RedHat either - they've made that abundantly
clear - so they are working towards further integrating udev with
systemd, to such an extent that you won't be able to use udev without
systemd anymore in the near future.
http://fedoraproject.org/wiki/Features/UsrMove
Note that the above link is to Fedora, which a RedHat-sponsored
community distribution and which serves as the testbed for the next
official RedHat release. You should therefore take their justifications
with a grain of salt.
> I never considered RedHat itself because I don't have local server
> requirements and don't want to spend money when I can get the same
> thing free of charge, [...
CentOS is the free version of RedHat Enterprise Linux.
> ...] but I now see it isn't the same thing, it's something that
their installers refused to let me install the system on anything else
than - at the time - ext3, in spite of my having created XFS partitions,
because XFS is vastly superior to ext3.
I've tried the same thing with reiserfs partitions, but I got the same
result. And yet both (the Linux port of [*]) XFS and reiserfs are
GPL'ed and have already been included in the upstream vanilla kernel
sources for many years.
[*] XFS is the native filesystem of SGI IRIX, a proprietary UNIX based
on the patents of AT&T Unix System V. It comes with a complete
tool chain and is a highly advanced balanced tree filesystem.
alt.comp.os.linux...
> On 12-12-25 03:20 AM, Aragorn wrote:
>
>> Gentoo uses OpenRC as the default System V init replacement, but
>> it does also support systemd. However, due to the planned
>> integration by Sievers and Poettering at RedHat between udev and
>> systemd, a number of Gentoo developers have now decided to fork udev.
>>
>> The fork is still in heavy testing phase and is called eudev. It
>> attempts to provide udev functionality without relying on systemd and
>> to undo the breakages introduced by udev and systemd, such as the
>> need to boot with an initramfs if you have /usr split off from the
>> root filesystem. This breakage was arrogantly introduced by Sievers
>> and Poettering by making udev and systemd rely on stuff which is
>> located under /usr, and was then further pushed downstream by
>> RedHat (via Fedora 17) by making /bin, /lib and /sbin symlinks to
>> their equivalent directories under the /usr tree.
>
> Is there some reason for doing things this way? I can't for the life
> of me see the logic of relying on symlinks under /usr for /bin, /lib
> and /sbin file operations.
and /usr/sbin respectively.
> It's like going around the block to get to the house next door. And it
> must play hell with PATH, among other things. If you have a link to
> some document where RedHat explains its rationale, please post it.
backward compatibility. There are many rationales leading to this
decision, but most of them are actually retroactive justification.
One of the rationales is that proprietary UNIX - and most notably
Solaris - also does it this way, and that /usr was already a hack to
begin with in the original AT&T Unix. That part is true. As history
had it, Unix became so big that it wouldn't fit on a single hard disk
anymore at the time, so a second disk was added and its volume was
mounted to /usr.
One of the other arguments is that with everything living under /usr, it
is easier to make backups and to mount /usr read-only and/or over NFS,
but this in itself contradicts the "separate /usr is broken" message
that Poettering and friends sent out into the world.
The whole thing started when udev and systemd began initializing things
which live under /usr - things which are not strictly needed at boot
time except in a few corner cases - and then it got even worse by the
default installation of both udev and systemd under /usr. So the
solution is to use an initramfs with busybox, which can mount /usr
before udev and systemd are started.
There are several articles on the web about this, but most of them are
just retroactive justification of the breakage, i.e. a separate /usr on
a system without an initramfs is broken because Poettering and Sievers
broke it and have no intention of fixing it. And they don't care about
other distributions than RedHat either - they've made that abundantly
clear - so they are working towards further integrating udev with
systemd, to such an extent that you won't be able to use udev without
systemd anymore in the near future.
http://fedoraproject.org/wiki/Features/UsrMove
Note that the above link is to Fedora, which a RedHat-sponsored
community distribution and which serves as the testbed for the next
official RedHat release. You should therefore take their justifications
with a grain of salt.
> I never considered RedHat itself because I don't have local server
> requirements and don't want to spend money when I can get the same
CentOS is the free version of RedHat Enterprise Linux.
> ...] but I now see it isn't the same thing, it's something that
> appears to be seriously worse.
I've never really liked RedHat, CentOS or Fedora, even if only because
their installers refused to let me install the system on anything else
than - at the time - ext3, in spite of my having created XFS partitions,
because XFS is vastly superior to ext3.
I've tried the same thing with reiserfs partitions, but I got the same
result. And yet both (the Linux port of [*]) XFS and reiserfs are
GPL'ed and have already been included in the upstream vanilla kernel
sources for many years.
[*] XFS is the native filesystem of SGI IRIX, a proprietary UNIX based
on the patents of AT&T Unix System V. It comes with a complete
tool chain and is a highly advanced balanced tree filesystem.
Eyetee
Dec 25, 2012, 4:37:53 PM12/25/12
to
On 12-12-25 12:06 PM, Aragorn wrote:
<snipping a ton of good stuff>
Okay, this move you describe toward component interdependency sounds
like a pure marketing strategy. In that regard, the trial transcripts of
the old Netscape lawsuit are very educational. RedHat _must_ be creating
marketing synergies with someone else who is out to make a buck. I can't
think of an alternative explanation, other than that perhaps they're
simply out of their minds and need to see a psychiatrist. Component
interdependency is like copulation with a meat grinder from a purely
technical point of view.
I tried Fedora 17 x64 Xfce and loved the functionality of the installer,
which gave me zillions of choices in terms of package selection and
walked me through a number of important steps--especially the option to
define different passwords for the root user and my limited user
account, which is something I haven't seen any other Linux distribution
installer do. (In fact, GUI-oriented distros such as Linux Mint have a
functionality where exceeding your permissions as a limited user
automatically causes you to be prompted for your limited user password,
which is the same as your root password; and that's something I find to
be plumb stupid.) But when I actually ran Fedora 17, it seemed extremely
buggy as things broke constantly all over the place, so I dropped it.
What you tell me about it gives me further insight into their
screamingly polemical homepage and non-technical documentation, because
I sensed the presence of a hidden agenda and just didn't have enough
information to puzzle out what that agenda was. (My only clues were from
the equally polemical Debian non-technical documentation, which is
something that turned me off about Debian.)
Thanks for the chance to talk on what would otherwise have been a boring
Christmas Day. Like four billion other people on earth, I don't
celebrate Christmas, but the fact that everything within a 3,000-mile
radius of me is closed for business today makes me lazy and uninterested
in doing much other than socializing. It's fun to talk to someone who
knows his s***. Pax!
Eyetee
--
website: www.eyetee.org
<snipping a ton of good stuff>
Okay, this move you describe toward component interdependency sounds
like a pure marketing strategy. In that regard, the trial transcripts of
the old Netscape lawsuit are very educational. RedHat _must_ be creating
marketing synergies with someone else who is out to make a buck. I can't
think of an alternative explanation, other than that perhaps they're
simply out of their minds and need to see a psychiatrist. Component
interdependency is like copulation with a meat grinder from a purely
technical point of view.
I tried Fedora 17 x64 Xfce and loved the functionality of the installer,
which gave me zillions of choices in terms of package selection and
walked me through a number of important steps--especially the option to
define different passwords for the root user and my limited user
account, which is something I haven't seen any other Linux distribution
installer do. (In fact, GUI-oriented distros such as Linux Mint have a
functionality where exceeding your permissions as a limited user
automatically causes you to be prompted for your limited user password,
which is the same as your root password; and that's something I find to
be plumb stupid.) But when I actually ran Fedora 17, it seemed extremely
buggy as things broke constantly all over the place, so I dropped it.
What you tell me about it gives me further insight into their
screamingly polemical homepage and non-technical documentation, because
I sensed the presence of a hidden agenda and just didn't have enough
information to puzzle out what that agenda was. (My only clues were from
the equally polemical Debian non-technical documentation, which is
something that turned me off about Debian.)
Thanks for the chance to talk on what would otherwise have been a boring
Christmas Day. Like four billion other people on earth, I don't
celebrate Christmas, but the fact that everything within a 3,000-mile
radius of me is closed for business today makes me lazy and uninterested
in doing much other than socializing. It's fun to talk to someone who
knows his s***. Pax!
Eyetee
--
website: www.eyetee.org
RayLopez99
Dec 30, 2012, 10:43:39 AM12/30/12
to
On Tuesday, December 25, 2012 9:18:41 AM UTC+2, Aragorn wrote:
> > The piratebay fan doesn't state what "social diseases" your OS can
>
> > acquire from piratebay. I'm sure having a black hat freely access your
>
> > entire home network through a backdoor built into a customized Win 7
>
> > .iso is loads of fun.
>
>
>
> Not to mention that all versions of Windows NT since NT 4.0 have at
>
> least two _identified_ backdoors built into them by default: one for use
>
> by the NSA and one for use by Microsoft.
Tin foil hat fail. Say who? Cite please?
> > The piratebay fan doesn't state what "social diseases" your OS can
>
> > acquire from piratebay. I'm sure having a black hat freely access your
>
> > entire home network through a backdoor built into a customized Win 7
>
> > .iso is loads of fun.
>
>
>
> Not to mention that all versions of Windows NT since NT 4.0 have at
>
> least two _identified_ backdoors built into them by default: one for use
>
> by the NSA and one for use by Microsoft.
>
>
>
> All versions of Microsoft Windows since NT 5.1 (i.e. Windows XP and 2003
>
> Server) are also known to phone home every week with a detailed report
>
> on what hardware is in the machine, what software is installed on the
>
> machine, and how often that software is used. Microsoft has
>
> acknowledged this but claims that this information is only used for
>
> statistics, plus that, as of Windows 7 (NT 6.1), it is now possible to
>
> disable this "feature" - it is called the "Microsoft User Experience
>
> Program" or something of that sorts.
>
>
>
> Anyway, Microsoft is irrelevant.
LOL!
>
> = Aragorn =
>
> (registered GNU/Linux user #223157)
RL
RayLopez99
Dec 30, 2012, 10:46:54 AM12/30/12
to
On Tuesday, December 25, 2012 9:36:51 AM UTC+2, Eyetee wrote:
> We are very like-minded when it comes to certain things.
Be careful, you realize this person claims to have Asperger syndrome don't you?
> We are very like-minded when it comes to certain things.
As for your other points I carefully screen what I download from Piratebay and it's rare to find a virus. Most users will comment if a virus exists, and the software will not be downloaded. If you believe these warez have an 'undetectable backdoor virus' of some sort, you're free to believe such paranoid nonsense.
RL
Aragorn
Dec 31, 2012, 11:22:27 AM12/31/12
to
On Tuesday 25 December 2012 22:37, Eyetee conveyed the following to
alt.comp.os.linux...
> On 12-12-25 12:06 PM, Aragorn wrote:
>
> <snipping a ton of good stuff>
>
> Okay, this move you describe toward component interdependency sounds
> like a pure marketing strategy. In that regard, the trial transcripts
> of the old Netscape lawsuit are very educational. RedHat _must_ be
> creating marketing synergies with someone else who is out to make a
> buck. I can't think of an alternative explanation, other than that
> perhaps they're simply out of their minds and need to see a
> psychiatrist. Component interdependency is like copulation with a meat
> grinder from a purely technical point of view.
More like "out of their mind", I'd say. ;-)
> I tried Fedora 17 x64 Xfce and loved the functionality of the
> installer, which gave me zillions of choices in terms of package
> selection and walked me through a number of important
> steps--especially the option to define different passwords for the
> root user and my limited user account, which is something I haven't
> seen any other Linux distribution installer do.
Mandriva, Mageia and PCLinuxOS still allow for that, and I think Debian
does too.
> (In fact, GUI-oriented distros such as Linux Mint have a functionality
> where exceeding your permissions as a limited user automatically
> causes you to be prompted for your limited user password, which is the
> same as your root password; and that's something I find to be plumb
> stupid.)
Actually, it's not the same as the root password, because in Ubuntu and
Mint, there *is* no root password. The root account is set up without a
password - which is not the same thing as a blank password - so as to
disable it, and the first user account created during install time is
added to the "admin" group, who has sudo rights. Ergo, every root
access functionality must be done by way of sudo and the graphical
front-ends to that, such as gksudo.
> But when I actually ran Fedora 17, it seemed extremely buggy as things
> broke constantly all over the place, so I dropped it.
Yes, Fedora is the "bleeding edge" testbed for the next official RedHat
and CentOS releases. It is not suitable for production environments.
> What you tell me about it gives me further insight into their
> screamingly polemical homepage and non-technical documentation,
> because I sensed the presence of a hidden agenda and just didn't have
> enough information to puzzle out what that agenda was. (My only clues
> were from the equally polemical Debian non-technical documentation,
> which is something that turned me off about Debian.)
>
> Thanks for the chance to talk on what would otherwise have been a
> boring Christmas Day. Like four billion other people on earth, I don't
> celebrate Christmas, [...
Same here. It's a day like any other day for me. ;-)
> ...] but the fact that everything within a 3,000-mile radius of me is
alt.comp.os.linux...
> On 12-12-25 12:06 PM, Aragorn wrote:
>
> <snipping a ton of good stuff>
>
> Okay, this move you describe toward component interdependency sounds
> like a pure marketing strategy. In that regard, the trial transcripts
> of the old Netscape lawsuit are very educational. RedHat _must_ be
> creating marketing synergies with someone else who is out to make a
> buck. I can't think of an alternative explanation, other than that
> perhaps they're simply out of their minds and need to see a
> psychiatrist. Component interdependency is like copulation with a meat
> grinder from a purely technical point of view.
> I tried Fedora 17 x64 Xfce and loved the functionality of the
> installer, which gave me zillions of choices in terms of package
> selection and walked me through a number of important
> steps--especially the option to define different passwords for the
> root user and my limited user account, which is something I haven't
> seen any other Linux distribution installer do.
does too.
> (In fact, GUI-oriented distros such as Linux Mint have a functionality
> where exceeding your permissions as a limited user automatically
> causes you to be prompted for your limited user password, which is the
> same as your root password; and that's something I find to be plumb
> stupid.)
Mint, there *is* no root password. The root account is set up without a
password - which is not the same thing as a blank password - so as to
disable it, and the first user account created during install time is
added to the "admin" group, who has sudo rights. Ergo, every root
access functionality must be done by way of sudo and the graphical
front-ends to that, such as gksudo.
> But when I actually ran Fedora 17, it seemed extremely buggy as things
> broke constantly all over the place, so I dropped it.
and CentOS releases. It is not suitable for production environments.
> What you tell me about it gives me further insight into their
> screamingly polemical homepage and non-technical documentation,
> because I sensed the presence of a hidden agenda and just didn't have
> enough information to puzzle out what that agenda was. (My only clues
> were from the equally polemical Debian non-technical documentation,
> which is something that turned me off about Debian.)
>
> Thanks for the chance to talk on what would otherwise have been a
> boring Christmas Day. Like four billion other people on earth, I don't
Same here. It's a day like any other day for me. ;-)
> ...] but the fact that everything within a 3,000-mile radius of me is
> closed for business today makes me lazy and uninterested in doing much
> other than socializing. It's fun to talk to someone who knows his
> s***. Pax!
And it wasn't even off-topic! :-D
> other than socializing. It's fun to talk to someone who knows his
> s***. Pax!
0 new messages