Reason *TO* pick on Windows 10

On 3/12/19 1:28 AM, Paul wrote:
MikeÂ*wasÂ*referringÂ*toÂ*"trial"Â*ofÂ*aÂ*LiveDVDÂ *withoutÂ*installation.

YouÂ*can'tÂ*trialÂ*everyÂ*possibleÂ*thingÂ*fromÂ*a Â*LiveDVD,Â*unless
youÂ*setÂ*upÂ*aÂ*persistentÂ*store.Â*ThisÂ*isÂ*typ icallyÂ*doneÂ*on
LiveDVDsÂ*transferredÂ*toÂ*USBÂ*sticks

The purpose of the Live DVD/USB is to demonstrate the
desktop and some programs. It is not intended to be a full
install of everything.

On 3/12/19 1:48 AM, T wrote:
On 3/12/19 1:28 AM, Paul wrote:
MikeÂ*wasÂ*referringÂ*toÂ*"trial"Â*ofÂ*aÂ*LiveDVDÂ *withoutÂ*installation.

YouÂ*can'tÂ*trialÂ*everyÂ*possibleÂ*thingÂ*fromÂ*a Â*LiveDVD,Â*unless
youÂ*setÂ*upÂ*aÂ*persistentÂ*store.

True and an art form in its own.

ThisÂ*isÂ*typicallyÂ*doneÂ*on
LiveDVDsÂ*transferredÂ*toÂ*USBÂ*sticks,Â*whereÂ*a *4GBÂ*persistence
fileÂ*("casper-rw")Â*isÂ*addedÂ*toÂ*theÂ*USBÂ*key,Â*soÂ*thatÂ*mor e
extensiveÂ*proceduresÂ*canÂ*beÂ*tried.Â*SuchÂ*asÂ* installingÂ*the
NVidiaÂ*graphicsÂ*driver,Â*tryingÂ*VirtualBoxÂ*or *VMPlayer.

NotÂ*everyoneÂ*hasÂ*scratchÂ*drivesÂ*forÂ*installi ngÂ*LinuxÂ*forÂ*testing.
IÂ*haveÂ*aÂ*tonÂ*ofÂ*drives,Â*andÂ*youÂ*canÂ*never Â*reallyÂ*haveÂ*enough
drivesÂ*forÂ*thisÂ*sortÂ*ofÂ*testing.Â*ForÂ*exampl e,Â*rightÂ*now,Â*when
I'mÂ*finishedÂ*withÂ*theÂ*500GBÂ*HDDÂ*inÂ*theÂ*Tes tÂ*Machine,Â*IÂ*have
toÂ*restoreÂ*fromÂ*backup,Â*toÂ*putÂ*theÂ*previous Â*experimentÂ*back
onÂ*there.

But a lot of us have Flash drives kicking around.Â* And you can
install to them and test whatever you want.

The best flash drive for this I have found are from Samsung.
No troubles with massive small file transfers.Â* Patriot
are the worst.


Install from a Live USB to a Sumsung flash drive in a USB 3.1/C
jack and it will give Windows a run for its money performance wise.
I have actually had them go faster in some customer's machines.

On 3/11/2019 9:55 PM, T wrote:
On 3/10/19 1:10 PM, Mike wrote:
you can fly before your buy.
That'sÂ*aÂ*myth.

If you install from the Live USB's installer, you get the same
exact thing as was on the Live stick.Â* Everything that
worked with the stick will work installed.

There is only one exception I have come across.Â* RHEL
does not work natively with C236 chipsets, but will
work with the Live USB because it is SLOWER and
the problem is a timing issue.Â* And RHEL won't fix
unless you pay them to.

So, you agree that there are instances where the live
differs from the installed.

Two problems I've encountered most often are when
the installed display comes up blank, and when
the LAN doesn't work and you have to search the web
for the driver that you need to search the web.

Since both worked from the live version, the required
drivers are there. It's just that nobody cared to
inform the OS upon installation. A mere mortal
can't figger out how to fix it.

The onion of linux is riddled with stuff like this.
No oversight leads to stuff not seen.

On 3/11/2019 10:15 PM, T wrote:
On 3/10/19 1:10 PM, Mike wrote:
ExperimentÂ*two:
GiveÂ*twoÂ*peopleÂ*computersÂ*newÂ*inÂ*theÂ*boxÂ*a ndÂ*internetÂ*connections.
LockÂ*'emÂ*inÂ*separateÂ*rooms.
TellÂ*'emÂ*toÂ*writeÂ*aÂ*shortÂ*memo,Â*attachÂ*aÂ* pictureÂ*andÂ*emailÂ*itÂ*toÂ*you
andÂ*you'llÂ*comeÂ*openÂ*theÂ*door.
TellÂ*personÂ*twoÂ*thatÂ*heÂ*willÂ*haveÂ*toÂ*useÂ* linuxÂ*toÂ*doÂ*it.
LeaveÂ*andÂ*waitÂ*forÂ*theÂ*emails.Â*Â*Don'tÂ*forg etÂ*toÂ*notifyÂ*next
ofÂ*kinÂ*andÂ*callÂ*hazmatÂ*forÂ*theÂ*secondÂ*guy.

I get paid all the time to set up Windows computers for customers
so they can use them.Â* Without me they are completely lost.


So lets redo your test with a good consultant setting up
the computers for them:

Install system-config-printer, Firefox, Libreoffice, Thunderbird,
Brave on both computers first with desktop icon.Â* Then repeat
the test. Tell them to set up their eMail in Thunderbird, write a
letter, set up a printer, look something up on Google.

Hmmm.Â* Linux is 40% faster, so Linux would win.

I'm on record saying that desktop linux is a great option when
a guru builds an APPLIANCE that contains all the capability
that the user will ever need...and updates don't break it.

If I could do that for myself, I'd kick Windows to the curb
in an instant.

I'm no fan of windows, but I don't see a VIABLE option.
And I've been trying forever.

On 3/11/2019 10:20 PM, T wrote:


Not to be too blunt, but technically speaking, Windows is s***.
This is what happens when the marketing department runs a company.
Fortunately, there are was to cope with Windows if you are stuck
having to run it for lack of Linux apps



We're in heated agreement on that.

The experiment is simple.
Build a dual boot system with linux and windows.

Set default boot to linux.
Run linux until you can't do something.
The metric is something YOU can't do easily via a GUI,
not something that a guru might cobble together.

At that point, set the default boot to windows and
continue on.
When you run into something that you can't do with
windows, but could do easily without a guru helping in linux.
At that point, set the default boot back to linux.

Flip back an forth as required to do all your stuff.

I've done that experiment many times.
It usually takes less than a day to flip back to windows.
I need to go back to linux...well...never.

It's not at all whether linux is technically superior.
It's about what the user can get done so he can get on
with other stuff in his life.

You can bitch about the marketing department all you want.
Fact is that's the way the world works.
Get on board, or swim upstream.
Most of us stick with the devil we know.

There is no viable alternative on the desktop.

On 3/11/2019 11:39 PM, T wrote:


So, I have no idea what you are not seeing.Â* All can say
is that Live USB's are suppose to be demonstrations so you
an hands on play with them.

Very true. I don't recall any distro that didn't do that.

But it's a hollow promise if the installed version boots
to a blank screen.

On 3/12/2019 2:21 AM, T wrote:

Install from a Live USB to a Sumsung flash drive in a USB 3.1/C
jack and it will give Windows a run for its money performance wise.
I have actually had them go faster in some customer's machines.

I've done this with external USB hard drives but have concerns
about thumb drives.
Does the installer identify a flash drive and configure itself
to be gentle with the writes to flash? Does the Samsung USB flash
have any internal help like wear leveling?

T wrote:
On 3/12/19 1:48 AM, T wrote:
On 3/12/19 1:28 AM, Paul wrote:
Mike was referring to "trial" of a LiveDVD without installation.

You can't trial every possible thing from a LiveDVD, unless
you set up a persistent store.

True and an art form in its own.

This is typically done on
LiveDVDs transferred to USB sticks, where a 4GB persistence
file ("casper-rw") is added to the USB key, so that more
extensive procedures can be tried. Such as installing the
NVidia graphics driver, trying VirtualBox or VMPlayer.

Not everyone has scratch drives for installing Linux for testing.
I have a ton of drives, and you can never really have enough
drives for this sort of testing. For example, right now, when
I'm finished with the 500GB HDD in the Test Machine, I have
to restore from backup, to put the previous experiment back
on there.

But a lot of us have Flash drives kicking around. And you can
install to them and test whatever you want.

The best flash drive for this I have found are from Samsung.
No troubles with massive small file transfers. Patriot
are the worst.


Install from a Live USB to a Sumsung flash drive in a USB 3.1/C
jack and it will give Windows a run for its money performance wise.
I have actually had them go faster in some customer's machines.


This is bad advice.

I've ruined two USB keys doing exactly that.

No more USB keys will be used here for either
persistent store or for install as slash.

When the first one failed, I pretended it was
a fluke. When the second one failed not too much
longer after that, I stopped pretending.

If it needs storage, it goes on a HDD. One
of the scratch HDDs (I probably have five or six
500GB ones for this, some are scratch because
of the danger they might fail). Even the ones
with a few reallocations showing, haven't actually
died on me yet.

The flash keys that failed, were TLC based, because
I opened up the sticks and looked up the chip numbers
in Google. If you have a stick old enough to be
based on MLC or SLC, that's probably safe for installation
purposes, just because there are more write cycles
available (even if the technology doesn't have effective
wear leveling).

If I had proof that the USB flash stick had as good
wear leveling as an SSD does, I wouldn't be nearly
as concerned. But the way these sticks fails, suggests
there is little in the way of wear leveling there. And
it could be all too easy to "burn a hole" in one as a result.
The sticks behaved slowly at first (write rate drops
to 1.5 to 3MB/sec), and it isn't that much later before
the thing disappears completely.

*******

And newer sticks are getting worse.

I bought a Sandisk Extreme maybe a year and a half ago,
and it has worked consistently.

I liked it enough, to look for such a stick at the
computer store a couple weeks ago. The model number had changed,
as you would expect.

I brought the stick home, and read speed was 50MB/sec.
That is hardly Extreme.

I tried using "dd", and did "dd if=/dev/urandom" and
used a random bit pattern to write the stick from
end to end. After that, a read test gave a more
uniform 150MB/sec on read.

So the root cause of the initial terrible performance,
is "mushy TLC" syndrome. This is where every sector
on the stick, when you get it, has errors requiring
ECC correction inside the stick. This causes the
read rate to drop, because the microcontroller inside
the chip does the correction in firmware. There isn't
a dedicated hardware block for error correction. TLC
is bad enough, they might use 50 bytes of overhead
for every 512 bytes of data. And the poor microcontroller
then has to work out which bits in the just-read block
need to be flipped.

I'm expecting if I leave data on the new stick, then
three months from now the read rate will have dropped
to 50MB/sec again. The question then is, how long
before the 512 byte block is "uncorrectable". The 50 byte
number is assumed to be powerful enough for a certain
error rate over time. And the "mushy" behavior isn't
supposed to lead to data loss. But how warm and fuzzy
do I feel about this... Hmmm.

I have other USB sticks that are slow but consistent.
I might not have a problem using those, but because
they're slow, the experience wouldn't be all that pleasant.
Nobody enjoys consistent 3MB/sec write rates (a couple
of them had that speed from day one).

If you got some good USB sticks, then great... I fully
recommend my OCZ RALLY2 8GB, which is bulletproof (at least
compared to the pile of TLC rubbish sitting on my desk
right now). I probably bought that eight or ten years ago.
That would be a good candidate for an install. But I only
own one of those (it had a mail in rebate, one per household).

Paul

"T" wrote

| Not politics. Economics. Code Weavers makes their living
| off of consulting services. They don't give anything
| away for free. My experience with them is very negative.
| You find a bug, even in their documentation (I found some
| really bad ones in their installation instructions), they
| start name calling and on and on and so forth.
|
| Wine Staging is finally fixing things.
|

This is all just red herrings. Code Weavers is a commercial
WINE wrapper. Wine Staging, according to their own
webpage, is optional beta features that can be installed
early.

None of that has anything to do with the basic problem
that WINE isn't a dependable option for Windows software
on Linux and that, in fact, the entire design of WINE is
deeply flawed. They act like teenagers trying to hack specific
Windows programs, rather than creating a cross-platform
API system that Windows programmers can target. They can
never have a finished product, by design.

| If you like way Windows does things, stick with Windows.
| The P is PC stands for Personal.

The problem is that you keep actively trying to sell
Linux to people. Your answer is always that people should
try a Linux CD. Computer problems? Dry skin? Losing crops to
drought? Wife doesn't orgasm? Your answer to everything
is Linux. Enthusiasm is fine but you're also providing
misinformation.

* You answered my complaint about firewalls without
actually knowing the answer.

* When I point out problems with WINE you go off on
unrelated tangents about Code Weavers.

* You claim Windows lacks backward compatibility. When I
ask for examples all you can name is Quicken. Yet Quicken
officially supports Win7-10 and when asked to explain the
problem you don't answer. That's 9 years of compatibility.

There are some gaps, and changes happen. Networking changed
fundamentally between Win98 and ME. So a firewall for Win98
won't work on Win10. And things that use a lot of wrappers
or are very low-level won't always work. For instance, AV
might need to be updated more often because it needs low-
level control. And DotNet software has to have its respective
runtime installed. But for documented API functions, you can
pretty much depend on any function to work, up to 25 years
old. The Win10 API is an augmented version of the Win95 API.

When asked for details to back up the claims you make,
you get evasive and finally dismissive.

In short, you're an evangelist, bending the truth to
support your enthusiasm. That does nothing to help
people who might actually want to try Linux. If you
were offering accurate information that would be
great: Firewall that does what I want? Yes or no.
If yes then which one?

But you won't answer straight. People who might try
Linux need to know they're getting themselves into a
partly finished do-it-yourself kit that probably never
will be finished. Like the teenager who's forever rebuilding
car on the front lawn rather than driving it.
And in some ways it's actually worse than that. Linux
designers, in recent years, have taken the same approach
as Microsoft -- gradually hiding how things work,
obfuscating true admin/root while offering a fake
admin/root, encouraging people to let things update
themselves... They're moving something that's eternally
beta to a kiosk design in order to attract more people.
That can make for big problems. Last time I decided to
try a multi-boot Linux install I had to quit because I
didn't trust the installer to do what I told it to. It kept
wanting to choose a partition with Windows rather than
the Linux partitions I'd already set up. In the early days
of Linux at least I didn't have to contend with ninny
functions and lying limitations.

Mayayana wrote:
"T" wrote

snip


The problem is that you keep actively trying to sell
Linux to people. Your answer is always that people should
try a Linux CD. Computer problems? Dry skin? Losing crops to
drought? Wife doesn't orgasm? Your answer to everything
is Linux. Enthusiasm is fine but you're also providing
misinformation.

So are you.


* You answered my complaint about firewalls without
actually knowing the answer.

Yes you rail about Linux doesn't have any per application based firewall
but like it is like Linux doesn't have antivirus either. The reason is
it is not really needed. Linux has a completely different approach to
security. Unlike Windows that has most things open and then adds stuff
to plug the holes, Linux starts with stuff closed until you explicitly
open open them. (Windows is moving to more things closed albeit
slowly...) So an application based firewall is not needed just don't
allow app access to networking. You can do that with services like
apparmor. Deny application access to network and it won't 'phone home'.

--
Take care,

Jonathan
-------------------
LITTLE WORKS STUDIO
http://www.LittleWorksStudio.com

On 3/12/19 7:42 AM, Jonathan N. Little wrote:
Mayayana wrote:
"T" wrote

snip


The problem is that you keep actively trying to sell
Linux to people. Your answer is always that people should
try a Linux CD. Computer problems? Dry skin? Losing crops to
drought? Wife doesn't orgasm? Your answer to everything
is Linux. Enthusiasm is fine but you're also providing
misinformation.

So are you.


* You answered my complaint about firewalls without
actually knowing the answer.

Yes you rail about Linux doesn't have any per application based firewall
but like it is like Linux doesn't have antivirus either. The reason is
it is not really needed. Linux has a completely different approach to
security. Unlike Windows that has most things open and then adds stuff
to plug the holes, Linux starts with stuff closed until you explicitly
open open them. (Windows is moving to more things closed albeit
slowly...) So an application based firewall is not needed just don't
allow app access to networking. You can do that with services like
apparmor. Deny application access to network and it won't 'phone home'.


SeLinux is a great example

On 3/12/19 3:32 AM, Mike wrote:
On 3/12/2019 2:21 AM, T wrote:

Install from a Live USB to a Sumsung flash drive in a USB 3.1/C
jack and it will give Windows a run for its money performance wise.
I have actually had them go faster in some customer's machines.

I've done this with external USB hard drives but have concerns
about thumb drives.

The only one I have found that is trouble free is the Samsung

Does the installer identify a flash drive and configure itself
to be gentle with the writes to flash?

No idea

Does the Samsung USB flash
have any internal help like wear leveling?

no idea. I only use them to troubleshoot.
Kingston and Kanguru corrupted all the time.

On 3/12/19 6:40 AM, Mayayana wrote:
This is all just red herrings. Code Weavers is a commercial
WINE wrapper. Wine Staging, according to their own
webpage, is optional beta features that can be installed
early.

Wine and Codeweavers are the same people. I have asked them
and they have verified it.

None of that has anything to do with the basic problem
that WINE isn't a dependable option for Windows software
on Linux and that, in fact, the entire design of WINE is
deeply flawed. They act like teenagers trying to hack specific
Windows programs, rather than creating a cross-platform
API system that Windows programmers can target. They can
never have a finished product, by design.

All Red Hat releases are using Staging. It is
a group better than regular Wine. And it still has a
way to go

On 3/12/19 6:40 AM, Mayayana wrote:
The problem is that you keep actively trying to sell
Linux to people

You are going off on a tangent. I work with three different
OS'es and I am just giving my evaluation. I clearly state
when I think Linux has short comings.

On 3/12/19 3:43 AM, Paul wrote:
T wrote:
On 3/12/19 1:48 AM, T wrote:
On 3/12/19 1:28 AM, Paul wrote:
Mike was referring to "trial" of a LiveDVD without installation.

You can't trial every possible thing from a LiveDVD, unless
you set up a persistent store.

True and an art form in its own.

This is typically done on
LiveDVDs transferred to USB sticks, where a 4GB persistence
file ("casper-rw") is added to the USB key, so that more
extensive procedures can be tried. Such as installing the
NVidia graphics driver, trying VirtualBox or VMPlayer.

Not everyone has scratch drives for installing Linux for testing.
I have a ton of drives, and you can never really have enough
drives for this sort of testing. For example, right now, when
I'm finished with the 500GB HDD in the Test Machine, I have
to restore from backup, to put the previous experiment back
on there.

But a lot of us have Flash drives kicking around.Â* And you can
install to them and test whatever you want.

The best flash drive for this I have found are from Samsung.
No troubles with massive small file transfers.Â* Patriot
are the worst.


Install from a Live USB to a Sumsung flash drive in a USB 3.1/C
jack and it will give Windows a run for its money performance wise.
I have actually had them go faster in some customer's machines.


This is bad advice.

I've ruined two USB keys doing exactly that.

No more USB keys will be used here for either
persistent store or for install as slash.

When the first one failed, I pretended it was
a fluke. When the second one failed not too much
longer after that, I stopped pretending.

If it needs storage, it goes on a HDD. One
of the scratch HDDs (I probably have five or six
500GB ones for this, some are scratch because
of the danger they might fail). Even the ones
with a few reallocations showing, haven't actually
died on me yet.

The flash keys that failed, were TLC based, because
I opened up the sticks and looked up the chip numbers
in Google. If you have a stick old enough to be
based on MLC or SLC, that's probably safe for installation
purposes, just because there are more write cycles
available (even if the technology doesn't have effective
wear leveling).

If I had proof that the USB flash stick had as good
wear leveling as an SSD does, I wouldn't be nearly
as concerned. But the way these sticks fails, suggests
there is little in the way of wear leveling there. And
it could be all too easy to "burn a hole" in one as a result.
The sticks behaved slowly at first (write rate drops
to 1.5 to 3MB/sec), and it isn't that much later before
the thing disappears completely.

*******

And newer sticks are getting worse.

I bought a Sandisk Extreme maybe a year and a half ago,
and it has worked consistently.

I liked it enough, to look for such a stick at the
computer store a couple weeks ago. The model number had changed,
as you would expect.

I brought the stick home, and read speed was 50MB/sec.
That is hardly Extreme.

I tried using "dd", and did "dd if=/dev/urandom" and
used a random bit pattern to write the stick from
end to end. After that, a read test gave a more
uniform 150MB/sec on read.

So the root cause of the initial terrible performance,
is "mushy TLC" syndrome. This is where every sector
on the stick, when you get it, has errors requiring
ECC correction inside the stick. This causes the
read rate to drop, because the microcontroller inside
the chip does the correction in firmware. There isn't
a dedicated hardware block for error correction. TLC
is bad enough, theyÂ* might use 50 bytes of overhead
for every 512 bytes of data. And the poor microcontroller
then has to work out which bits in the just-read block
need to be flipped.

I'm expecting if I leave data on the new stick, then
three months from now the read rate will have dropped
to 50MB/sec again. The question then is, how long
before the 512 byte block is "uncorrectable". The 50 byte
number is assumed to be powerful enough for a certain
error rate over time. And the "mushy" behavior isn't
supposed to lead to data loss. But how warm and fuzzy
do I feel about this... Hmmm.

I have other USB sticks that are slow but consistent.
I might not have a problem using those, but because
they're slow, the experience wouldn't be all that pleasant.
Nobody enjoys consistent 3MB/sec write rates (a couple
of them had that speed from day one).

If you got some good USB sticks, then great... I fully
recommend my OCZ RALLY2 8GB, which is bulletproof (at least
compared to the pile of TLC rubbish sitting on my desk
right now). I probably bought that eight or ten years ago.
That would be a good candidate for an install. But I only
own one of those (it had a mail in rebate, one per household).

Â*Â* Paul


Paul,

My experience too. I only use Samsung now for these purposes.
And I never meant to imply that your should use it for anything
other than testing.

I have even put Intel SATA SSD drives into USB 3 carriers and had
them corrupt, but Intel SSD stink anyway.

-T