More Vista Shitiness (NTFS)
yttrx
Huh, you'd think that NTFS v. 3.1 (whoops, thats exactly the same
version in XP!), which is what Vista ships with, would, as a "journaled
filesystem", be able to delete 68GB in the form of 422,000 files in
about the same time as FreeBSD with softupdates turned on (about
ten seconds) or Linux running ReiserFS (about 14 seconds), or Linux
running ext3 (about 20 seconds).
Nope.
Vista, running NTFS: 21 minutes and 49 seconds.
Hey eric, where does NTFS store its journal again?
-----yttrx
Erik Funkenbusch
> Huh, you'd think that NTFS v. 3.1 (whoops, thats exactly the same
> version in XP!), which is what Vista ships with, would, as a "journaled
> filesystem", be able to delete 68GB in the form of 422,000 files in
> about the same time as FreeBSD with softupdates turned on (about
> ten seconds) or Linux running ReiserFS (about 14 seconds), or Linux
> running ext3 (about 20 seconds).
Why would you think that? NTFS does a clean erase, zeroing out the sectors
as it deletes files. FreeBSD and Reiser do not, and in fact just unlink
the file. It has nothing to do with journaling.
Erik Funkenbusch
Also, Vista, using the GUI, will first count all the files and display the
status while deleting, I am guessing you're comparing a command line rm -rf
with a GUI delete as well.
yttrx
Hi, you don't understand journaling then, do you sweetie.
Thats alright. So what exactly is it you do for a living again?
-----yttrx
yttrx
Actually yup. But just for analism, it's rm -Rf.
Seriously though eric, why did you delete the time windows took to do it
from your response? Are you humiliated by how long it takes your favorite
operating system to do a very, very simple task that is more and more
commonplace on home systems these days?
And by the way, the glob of files that was deleted was the backup I took
from this laptop before the Vista upgrade. I stuck it on three separate
machines, plus an external USB drive just to be sure.
-----yttrx
Linonut
Then why do these guys sell a disk-cleaner:
--
"Developers! Developers! Developers!" -- Steve Ballmer, CEO Microsoft
The Ghost In The Machine
<yt...@yttrx.net>
wrote
on Thu, 08 Feb 2007 17:55:41 GMT
<xuJyh.4630$Yy1....@textfe.usenetserver.com>:
Confusion of the COLA regs? :-)
>
>
>
> -----yttrx
>
>
--
#191, ewi...@earthlink.net
Useless C++ Programming Idea #11823822:
signal(SIGKILL, catchkill);
--
Posted via a free Usenet account from http://www.teranews.com
7
> Erik Funkenbusch <er...@despam-funkenbusch.com> wrote:
>> On Thu, 8 Feb 2007 11:37:28 -0600, Erik Funkenbusch wrote:
>>
>>> On Thu, 08 Feb 2007 16:59:48 GMT, yttrx wrote:
>>>
>>>> Huh, you'd think that NTFS v. 3.1 (whoops, thats exactly the same
>>>> version in XP!), which is what Vista ships with, would, as a "journaled
>>>> filesystem", be able to delete 68GB in the form of 422,000 files in
>>>> about the same time as FreeBSD with softupdates turned on (about
>>>> ten seconds) or Linux running ReiserFS (about 14 seconds), or Linux
>>>> running ext3 (about 20 seconds).
'Nope.
Vista, running NTFS: 21 minutes and 49 seconds.
Hey eric, where does NTFS store its journal again?'
>>> Why would you think that? NTFS does a clean erase, zeroing out the
>>> sectors
>>> as it deletes files. FreeBSD and Reiser do not, and in fact just unlink
>>> the file. It has nothing to do with journaling.
>>
>> Also, Vista, using the GUI, will first count all the files and display
>> the status while deleting, I am guessing you're comparing a command line
>> rm -rf with a GUI delete as well.
>
> Actually yup. But just for analism, it's rm -Rf.
>
> Seriously though eric, why did you delete the time windows took to do it
> from your response? Are you humiliated by how long it takes your favorite
> operating system to do a very, very simple task that is more and more
> commonplace on home systems these days?
Of course he is humiliated.
He's just admitted that it takes 21 minutes 19 seconds
for the display to update. Gosh thats Aieeero for u.
tha...@tux.glaci.remove-this.com
>
> Why would you think that? NTFS does a clean erase, zeroing out the sectors
> as it deletes files. FreeBSD and Reiser do not, and in fact just unlink
> the file. It has nothing to do with journaling.
Actually, NTFS does NOT do that, at least it certainly does not do it by
default. Furthermore, writing zeroes is not enough to fully protect against
recovery of the deleted data (though it does take special equipment to perform
that magic). If you really want to be sure, you should write random bytes to
the sectors. If you are truly paranoid, you will perform multiple passes of
that.
If Microsoft really did perform a zero on delete as you suggest, my opinion
of their operating system and there software design skills would plummet even
further. I cannot imagine a better way to kill your overall disk performance
with unnecessary seek latency.
Later,
Thad
yttrx
Eric, we are ALL waiting for you to answer the question which you also
snipped from your response, curiously enough...
WHERE DOES NTFS KEEP ITS JOURNAL, ERIC??????? ANSWER THE QUESTION. IF YOU
CANNOT ANSWER THIS QUESTION, YOU ONCE AGAIN WILL BE A CONFIRMED LIAR AND
IDIOT.
So answer it already. Lets hear it.
-----yttrx
graeme
(snips)
> Erik Funkenbusch <er...@despam-funkenbusch.com> wrote:
>> Also, Vista, using the GUI, will first count all the files and display the
>> status while deleting, I am guessing you're comparing a command line rm -rf
>> with a GUI delete as well.
>
> Actually yup. But just for analism, it's rm -Rf.
What's the time for the same done with deltree?
Erik Funkenbusch
wrote:
> Erik Funkenbusch <er...@despam-funkenbusch.com> wrote:
>>
>> Why would you think that? NTFS does a clean erase, zeroing out the sectors
>> as it deletes files. FreeBSD and Reiser do not, and in fact just unlink
>> the file. It has nothing to do with journaling.
>
> Actually, NTFS does NOT do that, at least it certainly does not do it by
> default. Furthermore, writing zeroes is not enough to fully protect against
> recovery of the deleted data (though it does take special equipment to perform
> that magic). If you really want to be sure, you should write random bytes to
> the sectors. If you are truly paranoid, you will perform multiple passes of
> that.
Actually, after some research, you're right. I had mistaken NTFS's ability
to wipe sectors before handing them to the next application with zeroing
them when deleted. NTFS will zero a sector before it's given to a new
application so that old data cannot be exposed.
However, the piont still stands that using the GUI to delete is different
from the command prompt. The GUI must count all the files and their sizes
to display status information, command prompts don't. With a a large
directory tree with a lot of small files, this can take some time.
Erik Funkenbusch
> Erik Funkenbusch <er...@despam-funkenbusch.com> wrote:
>> On Thu, 8 Feb 2007 11:37:28 -0600, Erik Funkenbusch wrote:
>>
>>> On Thu, 08 Feb 2007 16:59:48 GMT, yttrx wrote:
>>>
>>>> Huh, you'd think that NTFS v. 3.1 (whoops, thats exactly the same
>>>> version in XP!), which is what Vista ships with, would, as a "journaled
>>>> filesystem", be able to delete 68GB in the form of 422,000 files in
>>>> about the same time as FreeBSD with softupdates turned on (about
>>>> ten seconds) or Linux running ReiserFS (about 14 seconds), or Linux
>>>> running ext3 (about 20 seconds).
>>>
>>> Why would you think that? NTFS does a clean erase, zeroing out the sectors
>>> as it deletes files. FreeBSD and Reiser do not, and in fact just unlink
>>> the file. It has nothing to do with journaling.
>>
>> Also, Vista, using the GUI, will first count all the files and display the
>> status while deleting, I am guessing you're comparing a command line rm -rf
>> with a GUI delete as well.
>
> Actually yup. But just for analism, it's rm -Rf.
No, -r and -R are the same.
from the man file:
-r, -R, --recursive
remove the contents of directories recursively
> Seriously though eric, why did you delete the time windows took to do it
> from your response? Are you humiliated by how long it takes your favorite
> operating system to do a very, very simple task that is more and more
> commonplace on home systems these days?
You're comparint apples and oranges. Try deleting the folder from a
command prompt in both situations.
Erik Funkenbusch
> So answer it already. Lets hear it.
No.
yttrx
I see.
Just admit that NTFS is NOT A JOURNALED FILESYSTEM. That you've
BEEN LYING.
Theres not one truly journaled filesystem in EXISTENCE that would
take so long to delete such a collection of files. Your little
NTFS there eric is off the mark by about 10,000%.
You should be fucking humiliated, and I know you are.
-----yttrx
yttrx
> On Thu, 08 Feb 2007 17:59:08 GMT, yttrx wrote:
>
>> Erik Funkenbusch <er...@despam-funkenbusch.com> wrote:
>>> On Thu, 8 Feb 2007 11:37:28 -0600, Erik Funkenbusch wrote:
>>>
>>>> On Thu, 08 Feb 2007 16:59:48 GMT, yttrx wrote:
>>>>
>>>>> Huh, you'd think that NTFS v. 3.1 (whoops, thats exactly the same
>>>>> version in XP!), which is what Vista ships with, would, as a "journaled
>>>>> filesystem", be able to delete 68GB in the form of 422,000 files in
>>>>> about the same time as FreeBSD with softupdates turned on (about
>>>>> ten seconds) or Linux running ReiserFS (about 14 seconds), or Linux
>>>>> running ext3 (about 20 seconds).
>>>>
>>>> Why would you think that? NTFS does a clean erase, zeroing out the sectors
>>>> as it deletes files. FreeBSD and Reiser do not, and in fact just unlink
>>>> the file. It has nothing to do with journaling.
>>>
>>> Also, Vista, using the GUI, will first count all the files and display the
>>> status while deleting, I am guessing you're comparing a command line rm -rf
>>> with a GUI delete as well.
>>
>> Actually yup. But just for analism, it's rm -Rf.
>
> No, -r and -R are the same.
>
> from the man file:
>
> -r, -R, --recursive
> remove the contents of directories recursively
>
Child, do not presume to lecture me about traditional Unix commands.
If you weren't such a fucking moron, you'd know that there are more "rm"s
than the GNU version, and that some of them do indeed differentiate case
with the -R/-r flag. It is good practice to always use the big -R when
you want to really blow away everything, because some unices need it to
blow away things like symbolic links and/or actually free up *every*
inode.
>> Seriously though eric, why did you delete the time windows took to do it
>> from your response? Are you humiliated by how long it takes your favorite
>> operating system to do a very, very simple task that is more and more
>> commonplace on home systems these days?
>
> You're comparint apples and oranges. Try deleting the folder from a
> command prompt in both situations.
Tell ya what, I just tried deleting it from Gnome on the FreeBSD machine.
About 12 seconds.
-----yttrx
Erik Funkenbusch
What the fuck are you talking about? Journaling has nothing to do with it.
Hint: Journaling does not make a file system faster than a non-journaled
one.
Compare apples to apples. Delete the folder from the command line on both,
otherwise you're just being a moron.
Erik Funkenbusch
>>> Actually yup. But just for analism, it's rm -Rf.
>>
>> No, -r and -R are the same.
>>
>> from the man file:
>>
>> -r, -R, --recursive
>> remove the contents of directories recursively
>>
>
> Child, do not presume to lecture me about traditional Unix commands.
> If you weren't such a fucking moron, you'd know that there are more "rm"s
> than the GNU version, and that some of them do indeed differentiate case
> with the -R/-r flag. It is good practice to always use the big -R when
> you want to really blow away everything, because some unices need it to
> blow away things like symbolic links and/or actually free up *every*
> inode.
We're talking about Linux here, which is gnu.
>>> Seriously though eric, why did you delete the time windows took to do it
>>> from your response? Are you humiliated by how long it takes your favorite
>>> operating system to do a very, very simple task that is more and more
>>> commonplace on home systems these days?
>>
>> You're comparint apples and oranges. Try deleting the folder from a
>> command prompt in both situations.
>
> Tell ya what, I just tried deleting it from Gnome on the FreeBSD machine.
>
> About 12 seconds.
Still not the same thing. Gnome doesn't count the files first. Vista
does. Delete them both from the command prompt.
yttrx
> On Thu, 08 Feb 2007 21:53:05 GMT, yttrx wrote:
>
>> Erik Funkenbusch <er...@despam-funkenbusch.com> wrote:
>>> On Thu, 08 Feb 2007 20:38:04 GMT, yttrx wrote:
>>>
>>>> So answer it already. Lets hear it.
>>>
>>> No.
>>
>> I see.
>>
>> Just admit that NTFS is NOT A JOURNALED FILESYSTEM. That you've
>> BEEN LYING.
>>
>> Theres not one truly journaled filesystem in EXISTENCE that would
>> take so long to delete such a collection of files. Your little
>> NTFS there eric is off the mark by about 10,000%.
>>
>> You should be fucking humiliated, and I know you are.
>
> What the fuck are you talking about? Journaling has nothing to do with it.
> Hint: Journaling does not make a file system faster than a non-journaled
> one.
>
Yes I know, but can you name ONE other journaled filesystem that needs
to be defragged?
Just do that. Just one.
> Compare apples to apples. Delete the folder from the command line on both,
> otherwise you're just being a moron.
I did, I hit the deltree command almost ten minutes ago. It's still
going.
You were saying, moron?
-----yttrx
Linonut
> On Thu, 8 Feb 2007 19:16:33 +0000 (UTC), tha...@tux.glaci.remove-this.com
That's still a design flaw, if it were true. I seriously doubt that
Explorer drills down into unopened directories to count them and add up
their sizes before allowing you to delete them. But if it really does,
and is accessing 100,000 file entries, shouldn't that take only a second
or two. And if you do wait for the GUI to update, then the counting is
already done, and the rest is just estimated progress calculations.
--
Microsoft -- Whose throat do you want to choke today?
yttrx
> On Thu, 08 Feb 2007 21:55:59 GMT, yttrx wrote:
>
>>>> Actually yup. But just for analism, it's rm -Rf.
>>>
>>> No, -r and -R are the same.
>>>
>>> from the man file:
>>>
>>> -r, -R, --recursive
>>> remove the contents of directories recursively
>>>
>>
>> Child, do not presume to lecture me about traditional Unix commands.
>> If you weren't such a fucking moron, you'd know that there are more "rm"s
>> than the GNU version, and that some of them do indeed differentiate case
>> with the -R/-r flag. It is good practice to always use the big -R when
>> you want to really blow away everything, because some unices need it to
>> blow away things like symbolic links and/or actually free up *every*
>> inode.
>
> We're talking about Linux here, which is gnu.
>
You're a fucking idiot if you didnt see where I also mentioned FREEBSD,
and you know goddamn well ive been working with at least a dozen
interpolations of Unix since the late 80s, bonehead. I have a damn
good reason to use -R, and I don't need some windows loving, microsoft
sucking toner monkey to tell ME how to make anything remotely close to
Unix work, thanks.
That'd be like me telling you how to clean a toilet, you know?
>>>> Seriously though eric, why did you delete the time windows took to do it
>>>> from your response? Are you humiliated by how long it takes your favorite
>>>> operating system to do a very, very simple task that is more and more
>>>> commonplace on home systems these days?
>>>
>>> You're comparint apples and oranges. Try deleting the folder from a
>>> command prompt in both situations.
>>
>> Tell ya what, I just tried deleting it from Gnome on the FreeBSD machine.
>>
>> About 12 seconds.
>
> Still not the same thing. Gnome doesn't count the files first. Vista
> does. Delete them both from the command prompt.
Did. Twelve minutes and counting.
By the way, putz, it only took Vista four minutes to count all the files
THE ENTIRE REST OF THE TIME WAS SPENT DELETING THEM, ASSHOLE.
You're just an idiot, really. Too bad you use your real name on Usenet,
now prospective employers can see that you're an idiot too!
-----yttrx
yttrx
Alright eric, a deltree took 15 minutes and 2 seconds, more or less. Look
at all the time saved because Vista didnt have to count all those files!
-----yttrx
yttrx
It really does, and it takes for fucking ever. And yes, its a design
flaw. And yes, eric is a moron.
-----yttrx
Erik Funkenbusch
>> What the fuck are you talking about? Journaling has nothing to do with it.
>> Hint: Journaling does not make a file system faster than a non-journaled
>> one.
>
> Yes I know, but can you name ONE other journaled filesystem that needs
> to be defragged?
>
> Just do that. Just one.
I would say that NTFS doesn't need to be defragged at all. You might want
to, but it isn't needed. I defrag about once a year. There are defrag
tools for Linux as well. If linux filesystems don't need defragging, why
do they exist?
Remember, Windows and NTFS was designed to be both a multi-user and
single-user OS. In a single-user OS fragmentation slows performance
because you're usually only doing one thing at a time, but in a multi-user
OS fragmentation is less of a problem, and can actually improve
performance.
Read this:
http://www.salmar.com/pipermail/wftl-lug/2002-March/000603.html
>> Compare apples to apples. Delete the folder from the command line on both,
>> otherwise you're just being a moron.
>
> I did, I hit the deltree command almost ten minutes ago. It's still
> going.
>
> You were saying, moron?
Remember that Linux has a "delayed" delete. It just marks files deleted in
memory quickly, then later runs a cleanup process to physically delete the
files.
Also, I get the feeling you're deleting across a network... You're
dishonest enough to pull that crap, just like when you tried to delete with
a gui versus command line and pass it off as the same thing.
Peter Köhlmann
> On Thu, 08 Feb 2007 22:15:58 GMT, yttrx wrote:
>
>>> What the fuck are you talking about? Journaling has nothing to do with
>>> it. Hint: Journaling does not make a file system faster than a
>>> non-journaled one.
>>
>> Yes I know, but can you name ONE other journaled filesystem that needs
>> to be defragged?
>>
>> Just do that. Just one.
>
> I would say that NTFS doesn't need to be defragged at all. You might want
> to, but it isn't needed. I defrag about once a year. There are defrag
> tools for Linux as well. If linux filesystems don't need defragging, why
> do they exist?
>
Really? And they are younger than several *years* Erik?
When where the project pages updated last, Erik? 1998? Earlier?
Why aren't those "defrag tools" included with linux distros, Erik?
> Remember, Windows and NTFS was designed to be both a multi-user and
> single-user OS.
Well, windows has never been multi-user. The hooks are there, but that's it
> In a single-user OS fragmentation slows performance
> because you're usually only doing one thing at a time, but in a multi-user
> OS fragmentation is less of a problem, and can actually improve
> performance.
Lets say it does not hinder performance
< snip more Funkenbusch droppings >
--
Avoid reality at all costs.
Linonut
> By the way, putz, it only took Vista four minutes to count all the files
Wow. It's difficult to believe that a programmer would want to code in
that kind of time-killer, even in a GUI.
> THE ENTIRE REST OF THE TIME WAS SPENT DELETING THEM, ASSHOLE.
And it was an excessively long time. How many files was it? How many
bytes?
--
There's no place like ~
tha...@tux.glaci.remove-this.com
>
> That's still a design flaw, if it were true. I seriously doubt that
> Explorer drills down into unopened directories to count them and add up
> their sizes before allowing you to delete them. But if it really does,
> and is accessing 100,000 file entries, shouldn't that take only a second
> or two. And if you do wait for the GUI to update, then the counting is
> already done, and the rest is just estimated progress calculations.
Just for grins, I decided to do a test on my new ubuntu system. I did
a find command, recursively outputing the name of every file in the /usr
partition, counting the number of file and directory names and totaling
up the time it took. Here is what I ended up with:
root@zim:/usr# time find . -print | wc -l
177546
real 0m0.995s
user 0m0.412s
sys 0m0.620s
That 177546 directory entries traversed in less than one second. Then
I remembered that I had done a find on that partition earlier in the
day, so I suspected I was benefiting from a lot of cache hits. I
tried another partition:
root@zim:/u# time find . -print | wc -l
266807
real 2m15.130s
user 0m0.652s
sys 0m2.008s
More than a quarter million directory entries in 2 and quarter minutes.
Finally, to test my theory about cache hits, I tried the same partition
again:
root@zim:/u# time find . -print | wc -l
266807
real 0m1.528s
user 0m0.568s
sys 0m0.804s
Now it was down to a second and a half. This is on an untuned ext3
partition, in case anyone is curious. I would love to test the same
sort of thing on a Windows box for comparison purposes, but I don't
have one handy at the moment.
Thad
The Ghost In The Machine
<lin...@bone.com>
wrote
on Thu, 08 Feb 2007 17:55:33 -0600
<ccydnYNVN7PoJVbY...@comcast.com>:
Also, on what hardware? It wouldn't be too difficult
for a UML pseudo-system to mount an NFS server through a
pseudo-jammer of some sort, and have packets coming in at
such a slow rate that things take forever.
More likely, yttrx is using an existing system performing
a local delete. Whether that's a realistic test against
another system with a similar distribution of files is
unclear, unless one can guarantee hardware is identical
and installation almost identical, since obviously, there
is the minor difference of OS.
(Mindcraft had some interesting ideas in that area.
Inaccurate, but interesting.)
I've seen ext2 take some time to delete files -- but that's
only if there's a *lot* of files.
--
#191, ewi...@earthlink.net
fortune: not found
Erik Funkenbusch
wrote:
> root@zim:/u# time find . -print | wc -l
> 266807
>
> real 2m15.130s
> user 0m0.652s
> sys 0m2.008s
>
> More than a quarter million directory entries in 2 and quarter minutes.
> Finally, to test my theory about cache hits, I tried the same partition
> again:
What kind of system is this?
here's what I got, and I killed it because it was taking too long:
[root@cartrux /]# time find . -print | wc -l
real 8m33.791s
user 0m0.185s
sys 0m1.267s
[root@cartrux /]# uname -a
Linux zzz.com 2.6.9-023stab033.7-smp #1 SMP Wed Nov 15 14:43:01 MSK 2006
i686 i686 i386 GNU/Linux
i'll let it run while i go to dinner and report back...
bugbuster
It looks like you're starting from /. You wouldn't happen to have any
remote file systems mounted? Anyway, here's mine (no remote file systems):
First time:
[root@amd64a /]# time find . -print | wc -l
1.07user 4.61system 2:51.98elapsed 3%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (0major+5338minor)pagefaults 0swaps
1848690
Secomd time:
[root@amd64a /]# time find . -print | wc -l
0.98user 0.95system 0:02.08elapsed 92%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (0major+5338minor)pagefaults 0swaps
1848690
[root@amd64a /]# uname -a
Linux amd64a.localdomain 2.6.12-27mdksmp #1 SMP Tue Sep 26 11:20:54 MDT 2006 x86_64 AMD Athlon(tm) 64 X2 Dual Core Processor 4600+ unknown GNU/Linux
All file systems are ext3 - much in LVM.
Bug
tha...@tux.glaci.remove-this.com
Two 120GB ATA/UDMA harddrives (ST3120026A) with 8192KiB Cache
Running Ubuntu 6.10
Linux zim 2.6.17-10-SMP
(Note SMP kernel to take advantage of dual core CPU)
> here's what I got, and I killed it because it was taking too long:
>
> [root@cartrux /]# time find . -print | wc -l
>
> real 8m33.791s
> user 0m0.185s
> sys 0m1.267s
It looks like you started your find from /, which means it would also
traverse the virtual /proc and /sys filesystems. That could skew your
results if your just looking for filesystem performance info.
> [root@cartrux /]# uname -a
> Linux zzz.com 2.6.9-023stab033.7-smp #1 SMP Wed Nov 15 14:43:01 MSK 2006
> i686 i686 i386 GNU/Linux
>
> i'll let it run while i go to dinner and report back...
Cool. Enjoy your meal. Tip your server. ;)
Thad
Erik Funkenbusch
Here's the final results:
[root@cartrux usr]# time find . -print | wc -l
200604
real 18m34.776s
user 0m0.986s
sys 0m6.424s
That's 8x as long, and this is the machine:
processor : 0
vendor_id : GenuineIntel
cpu family : 15
model : 4
model name : Intel(R) Xeon(TM) CPU 2.80GHz
stepping : 1
cpu MHz : 2793.649
cache size : 1024 KB
...
processor : 1
vendor_id : GenuineIntel
cpu family : 15
model : 4
model name : Intel(R) Xeon(TM) CPU 2.80GHz
stepping : 1
cpu MHz : 2793.649
cache size : 1024 KB
...
MemTotal: 3110300 kB
MemFree: 7252 kB
Buffers: 220316 kB
Cached: 663916 kB
NoNamer
> Huh, you'd think that NTFS v. 3.1 (whoops, thats exactly the same
> version in XP!), which is what Vista ships with
This is partially incorrect. Vista (and Longhorn server) come with an
improved NTFS. Symbolic links (yes I know UNIX has had them for
years), and atomic transactions are now supported in NTFS.
Please see this link:
http://msdn2.microsoft.com/en-us/library/aa363764.aspx
Also here (and a lot of cool new kernel improvements in Vista):
http://www.microsoft.com/technet/technetmag/issues/2007/02/VistaKernel/default.aspx
>, would, as a "journaled
> filesystem", be able to delete 68GB in the form of 422,000 files in
> about the same time as FreeBSD with softupdates turned on (about
> ten seconds) or Linux running ReiserFS (about 14 seconds), or Linux
> running ext3 (about 20 seconds).
Journaling has nothing to do with performance of a file system, it has
a lot to do with data integrity.
> Nope.
>
> Vista, running NTFS: 21 minutes and 49 seconds.
>
> Hey eric, where does NTFS store its journal again?
Technical details regarding the exact location of the journal are hard
to come by, but Windows provides an API to access the NTFS journal.
See this document and sample open source code from Microsoft:
http://msdn2.microsoft.com/en-us/library/aa365736.aspx
Have a nice day!
>
> -----yttrx
>
> --http://www.yttrx.net
Erik Funkenbusch
wrote:
>> here's what I got, and I killed it because it was taking too long:
>>
>> [root@cartrux /]# time find . -print | wc -l
>>
>> real 8m33.791s
>> user 0m0.185s
>> sys 0m1.267s
>
> It looks like you started your find from /, which means it would also
> traverse the virtual /proc and /sys filesystems. That could skew your
> results if your just looking for filesystem performance info.
I noticed that when i restarted it, and redid it from /usr, as my other
results show.
Jim Richardson
Hash: SHA1
On Thu, 8 Feb 2007 19:10:03 -0600,
jim@dragon:/$ time sudo find . -print |wc -l
find: ./proc/18629/task: No such file or directory
find: ./proc/18629/fd: No such file or directory
629719
real 2m0.688s
user 0m1.276s
sys 0m3.736s
jim@dragon:/$ uname -a
Linux dragon 2.6.17-10-generic #2 SMP Tue Dec 5 22:28:26 UTC 2006 i686
GNU/Linu
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)
iD8DBQFFy/XKd90bcYOAWPYRAiGhAKCQ2v9+KmA7bJaLRUbJydzA/+179gCfTaHR
Do2xq9iCtLWfBYX8uX5S4JA=
=azBG
-----END PGP SIGNATURE-----
--
Jim Richardson http://www.eskimo.com/~warlock
My theory is that the great success of Windoze is based entirely on
providing someone to blame....
Jon Tillman in ASP
Johan Lindquist
Erik Funkenbusch was all like, "Dude,
Seriously, what /is/ that machine of yours doing?
-----8<-----
trotskij etc # time find / -print | wc -l
364019
real 0m7.189s
user 0m0.400s
sys 0m0.870s
processor : 0
vendor_id : AuthenticAMD
cpu family : 15
model : 47
model name : AMD Athlon(tm) 64 Processor 3500+
trotskij etc # df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/md/1 988152 64168 873788 7% /
udev 514776 124 514652 1% /dev
/dev/mapper/vg00-lv_usr
8256952 2350724 5486800 30% /usr
/dev/mapper/vg00-lv_var
8388348 3883420 4504928 47% /var
/dev/mapper/vg00-lv_mail
41284928 1178132 38009644 4% /home
----->8-----
This is a noname single-cpu with two software-mirrored sata drives and
lvm2, btw. No network mounts. Only handles mail and http stuff for a
small company. Guess it's a good thing we didn't cheap out and get one
of those slow-moving dual xeons, eh.
--
Time flies like an arrow, fruit flies like a banana. Perth ---> *
09:14:37 up 92 days, 6:56, 5 users, load average: 1.12, 0.57, 0.26
Linux 2.6.18.1 x86_64 GNU/Linux Registered Linux user #261729
AB
Proof that the GUI method is plain stupid (if you're right, which I
actually don't believe).
Back to the questions that you made me come up with while you were
TIPPETY-TAPping around, apologizing for MS:
1. *WHY* the fuck does NTFS have to count the files before deleting
them?? I knew they were complete imbeciles at that place, but I never
realized they might be that bald-assed stupid! Admit it, you made that
up.
2. Wouldn't you say one would have to be completely retarded, even more
retarded than retarded, to have a window of flying-whatever to show the
user that it's doing what it's told to do? I mean, it's going to count
the files (probably a load of crap) for 10 minutes, then it's going to
have to waste about 75% of it's time making shit move across the screen
in a dance that says "Look! I'm really doing something!" 'Fess up.
The guys making all of this shit up are a big group of Jerry's Kids,
aren't they?
--
"In short: just say NO TO DRUGS, and maybe you won't end up like
the Windows people."
AB
Here's what I got out of /usr with my super-high-speed 500MHz P3, 384M
ram:
root@3[usr]# time find . -print | wc -l
159835
real 0m44.457s
user 0m1.012s
sys 0m2.044s
Here it is from the / directory, /usr already cached:
root@3[/]# time find . -print | wc -l
309008
real 1m4.178s
user 0m1.924s
sys 0m2.496s
And:
root@3[usr]# uname -a
Linux home 2.6.15-27-desktop #1 PREEMPT Thu Dec 14 14:55:05 EST 2006
i686 GNU/Linux
Nice try, Mr Apology. But you still haven't explained why the ultra-
modern NTFS/Windows combination is slow as molasses. You've simply
tried to take advantage of a diversion to avoid answering altogether.
--
"One world, one web, one program" -- Microsoft promotional ad
"Ein Volk, ein Reich, ein Füehrer" -- Adolf Hitler
Peter Köhlmann
> On Thu, 8 Feb 2007 19:16:33 +0000 (UTC), tha...@tux.glaci.remove-this.com
> wrote:
>
>> Erik Funkenbusch <er...@despam-funkenbusch.com> wrote:
>>>
>>> Why would you think that? NTFS does a clean erase, zeroing out the
>>> sectors
>>> as it deletes files. FreeBSD and Reiser do not, and in fact just unlink
>>> the file. It has nothing to do with journaling.
>>
>> Actually, NTFS does NOT do that, at least it certainly does not do it by
>> default. Furthermore, writing zeroes is not enough to fully protect
>> against recovery of the deleted data (though it does take special
>> equipment to perform
>> that magic). If you really want to be sure, you should write random
>> bytes to
>> the sectors. If you are truly paranoid, you will perform multiple passes
>> of that.
>
> Actually, after some research, you're right. I had mistaken NTFS's
> ability to wipe sectors before handing them to the next application with
> zeroing
> them when deleted. NTFS will zero a sector before it's given to a new
> application so that old data cannot be exposed.
>
> However, the piont still stands that using the GUI to delete is different
> from the command prompt. The GUI must count all the files and their sizes
> to display status information, command prompts don't.
Do you like being so ridiculous wrong all the time, Erik?
GUIs don't display status info all the time. If I want to have file size etc
of any given file, I right click on its icon. *Then* I get the status info.
And selecting several files in the GUI gets me the count of the files
selected displayed plus the combined sizes. That operation is *fast*
And it takes place *before* chosing the action (like delete)
> With a a large
> directory tree with a lot of small files, this can take some time.
What *is* taking the time in GUIs is not status info and file sizes, it is
displaying the appropriate icons for the file.
--
Microsoft's Guide To System Design:
If it starts working, we'll fix it. Pronto.
yttrx
> On Feb 9, 12:59 am, y...@yttrx.net (yttrx) wrote:
>> Huh, you'd think that NTFS v. 3.1 (whoops, thats exactly the same
>> version in XP!), which is what Vista ships with
>
> This is partially incorrect. Vista (and Longhorn server) come with an
> improved NTFS. Symbolic links (yes I know UNIX has had them for
> years), and atomic transactions are now supported in NTFS.
>
I see. So they dont change revision numbers when they revise. Nice.
> Please see this link:
>
> http://msdn2.microsoft.com/en-us/library/aa363764.aspx
>
> Also here (and a lot of cool new kernel improvements in Vista):
>
> http://www.microsoft.com/technet/technetmag/issues/2007/02/VistaKernel/default.aspx
>
>>, would, as a "journaled
>> filesystem", be able to delete 68GB in the form of 422,000 files in
>> about the same time as FreeBSD with softupdates turned on (about
>> ten seconds) or Linux running ReiserFS (about 14 seconds), or Linux
>> running ext3 (about 20 seconds).
>
> Journaling has nothing to do with performance of a file system, it has
> a lot to do with data integrity.
>
Whoa, you're way off. It depends on how the journal is implemented. XFS
under AIX for example uses journal indexing features to greatly improve
FS performance.
>> Nope.
>>
>> Vista, running NTFS: 21 minutes and 49 seconds.
>>
>> Hey eric, where does NTFS store its journal again?
>
> Technical details regarding the exact location of the journal are hard
> to come by, but Windows provides an API to access the NTFS journal.
> See this document and sample open source code from Microsoft:
>
Thats because it doesn't actually exist. I understand that there's
an API to "see inside" the journal of NTFS, but that does not mean
that a journal actually exists---it means that microsoft wants to
be competative, so they SAY that it actually exists.
How do I know? Two reasons:
1. NTFS is not a reliable filesystem--if it were journaled, it would be
2. writing to an NTFS filesystem does NOT require access to a journal
of any kind.
-----yttrx
yttrx
You're an idiot, eric.
All you've proven is that any operating system can grind a drive.
I've shown, on the other hand, that windows is horrific at deleting
a buncha shit, and that it always takes forever to do it.
-----yttrx
Linonut
> On Thu, 8 Feb 2007 19:10:03 -0600, Erik Funkenbusch wrote:
>
>>> root@zim:/u# time find . -print | wc -l >>> 266807
>>>
>>> real 2m15.130s >>> user 0m0.652s >>> sys 0m2.008s
>
> Here's the final results:
>
> [root@cartrux usr]# time find . -print | wc -l > 200604
>
> real 18m34.776s > user 0m0.986s > sys 0m6.424s
>
> That's 8x as long, and this is the machine:
>
> model name : Intel(R) Xeon(TM) CPU 2.80GHz
> MemTotal: 3110300 kB
Here's my main (but crappy) ASrock-based desktop:
$ time find /usr/ -print | wc -l
485397
real 8m11.531s
$ uname -a
Linux 2.6.17.4-rock-1 #1 PREEMPT
i686 AMD Athlon(tm) XP 2000+ AuthenticAMD GNU/Linux
So my crap box is faster at this task than your Xeon.
Maybe it pays to configure and build your own kernel? <grin>
Oh, here's the result from my DELL Centrino-based laptop (same RAM, 1
Gb, as the desktop).
$ time find /usr/ -print | wc -l
221428
real 0m0.426s
$ uname -a
Linux 2.6.18-1 #1
i686 GNU/Linux
I'll have to try that one again after a reboot. Later, mater.
OK
Too much hard drugs?
Peter Köhlmann
> Here's the final results:
>
> [root@cartrux usr]# time find . -print | wc -l
> 200604
>
> real 18m34.776s
> user 0m0.986s
> sys 0m6.424s
>
> That's 8x as long, and this is the machine:
>
< snip >
peter@Laotse:/usr> time find . -print | wc -l
198213
real 0m29.861s
user 0m0.248s
sys 0m0.820s
Linux Laotse 2.6.13-15-default #1 Tue Sep 13 14:56:15 UTC 2005 x86_64 x86_64
x86_64 GNU/Linux
You have successfully borked your system, Erik
Did flatttestbrain assist in screwing it up that badly, or had linux-sux and
DFS their helping hand in it as well?
Once again you show that you have no idea about linux. None at all
Just give it up pretending to have a working linux setup. You don't
--
If you had any brains, you'd be dangerous.
tha...@tux.glaci.remove-this.com
>
> root@3[usr]# time find . -print | wc -l
> 159835
>
> real 0m44.457s
> user 0m1.012s
> sys 0m2.044s
fresh reboot.
root@zim:/usr# time find . -print | wc -l
177783
real 0m29.702s
user 0m0.280s
sys 0m0.952s
> Here it is from the / directory, /usr already cached:
>
> root@3[/]# time find . -print | wc -l
> 309008
>
> real 1m4.178s
> user 0m1.924s
> sys 0m2.496s
And similarly, here is a find from /.
root@zim:/# time find . -print | wc -l
627506
real 2m52.746s
user 0m0.956s
sys 0m2.704s
Adjusting for number of files, we are seeing similar results.
It looks like this is largely an I/O bound task and not heavily
effected by differences in CPU. The type of disk, amount of
drive and filesystem cache, or competition with other disk
activity could effect the numbers quite a bit though.
Later,
Thad
tha...@tux.glaci.remove-this.com
>
> Technical details regarding the exact location of the journal are hard
> to come by, but Windows provides an API to access the NTFS journal.
> See this document and sample open source code from Microsoft:
>
> http://msdn2.microsoft.com/en-us/library/aa365736.aspx
>
> Have a nice day!
I've actually poked around the Windows journaling API before but it
is still mostly a mystery to me. What I would really like to know
(but have been unable to determine) is if there is a way to move the
journal to a different drive from the one it is journaling. I can
do this in various Linux file systems, and it has huge performance
benefits under certain high I/O loads (think streaming video).
Later,
Thad
AB
> Erik Funkenbusch wrote:
>
>> Here's the final results:
>>
>> [root@cartrux usr]# time find . -print | wc -l
>> 200604
>>
>> real 18m34.776s
>> user 0m0.986s
>> sys 0m6.424s
>>
>> That's 8x as long, and this is the machine:
>>
>< snip >
>
> peter@Laotse:/usr> time find . -print | wc -l
> 198213
>
> real 0m29.861s
> user 0m0.248s
> sys 0m0.820s
>
>
> Linux Laotse 2.6.13-15-default #1 Tue Sep 13 14:56:15 UTC 2005 x86_64 x86_64
> x86_64 GNU/Linux
>
>
> You have successfully borked your system, Erik
> Did flatttestbrain assist in screwing it up that badly, or had linux-sux and
> DFS their helping hand in it as well?
I can't imagine either of those people screwing things up that badly.
Not even on purpose. He's probably using the MS method of running stuff
totally off of flash sticks.
> Once again you show that you have no idea about linux. None at all
> Just give it up pretending to have a working linux setup. You don't
No wonder he has problems liking linux. He can't even get it to do
simple tasks without messing them up royally.
Most anybody else would be embarassed to admit they destroyed their
machine that badly.
--
Chimpanzees can point and click on a Windows UI. That's mostly who's
doing it.
NoNamer
> NoNamer <grug2...@hotmail.com> wrote:
> > On Feb 9, 12:59 am, y...@yttrx.net (yttrx) wrote:
> >> Huh, you'd think that NTFS v. 3.1 (whoops, thats exactly the same
> >> version in XP!), which is what Vista ships with
>
> > This is partially incorrect. Vista (and Longhorn server) come with an
> > improved NTFS. Symbolic links (yes I know UNIX has had them for
> > years), and atomic transactions are now supported in NTFS.
>
> I see. So they dont change revision numbers when they revise. Nice.
> > Please see this link:
>
> >http://msdn2.microsoft.com/en-us/library/aa363764.aspx
>
> > Also here (and a lot of cool new kernel improvements in Vista):
>
> >http://www.microsoft.com/technet/technetmag/issues/2007/02/VistaKerne...
>
> >>, would, as a "journaled
> >> filesystem", be able to delete 68GB in the form of 422,000 files in
> >> about the same time as FreeBSD with softupdates turned on (about
> >> ten seconds) or Linux running ReiserFS (about 14 seconds), or Linux
> >> running ext3 (about 20 seconds).
>
> > Journaling has nothing to do with performance of a file system, it has
> > a lot to do with data integrity.
>
> Whoa, you're way off. It depends on how the journal is implemented. XFS
> under AIX for example uses journal indexing features to greatly improve
> FS performance.
The whole purpose of journaling is to be able to recover from an
operation that, for some reason (power failure, etc.), failed.
> >> Nope.
>
> >> Vista, running NTFS: 21 minutes and 49 seconds.
>
> >> Hey eric, where does NTFS store its journal again?
>
> > Technical details regarding the exact location of the journal are hard
> > to come by, but Windows provides an API to access the NTFS journal.
> > See this document and sample open source code from Microsoft:
>
> Thats because it doesn't actually exist. I understand that there's
> an API to "see inside" the journal of NTFS, but that does not mean
> that a journal actually exists---it means that microsoft wants to
> be competative, so they SAY that it actually exists.
Sorry, but this is not correct.
Please see these links:
http://www.microsoft.com/whdc/system/winpreinst/ntfs-preinstall.mspx
http://faq.arstechnica.com/link.php?i=1227
http://arstechnica.com/paedia/n/ntfs/ntfs4-3.html
http://www.extremetech.com/article2/0,1697,1168822,00.asp
http://articles.techrepublic.com.com/5102-10880-6091177.html
http://www.answers.com/topic/ntfs
http://www.mcmillan.cx/ntfs.html
NTFS is indeed a metadata journaled file system.
Not sure why you are harping on this great file system... most
technical sites agree that NTFS is *one* of the best file systems
around.
It's something that MS did right.
> How do I know? Two reasons:
>
> 1. NTFS is not a reliable filesystem--if it were journaled, it would be
NTFS *is* a very reliable filesystem. Do you have links to prove your
statement?
> 2. writing to an NTFS filesystem does NOT require access to a journal
> of any kind.
I have no idea what you are trying to say here...
You need to get your facts straight about NTFS, that is for sure.
A quick search on google will bring up at least 20 good articles about
NTFS... I linked to some of the more straight forward ones.
> -----yttrx
>
> --http://www.yttrx.net- Hide quoted text -
>
> - Show quoted text -
yttrx
That's ONE reason for a journal. I know you're all into windows and
stuff, friend, but microsoft did not invent the concept (or really, any
concept), and everyone else has been doing it for a much, much longer
time. See HFS+, UFS + Softupdates, JFS, XFS, ext3 and reiserfs for
details.
>> >> Nope.
>>
>> >> Vista, running NTFS: 21 minutes and 49 seconds.
>>
>> >> Hey eric, where does NTFS store its journal again?
>>
>> > Technical details regarding the exact location of the journal are hard
>> > to come by, but Windows provides an API to access the NTFS journal.
>> > See this document and sample open source code from Microsoft:
>>
>> Thats because it doesn't actually exist. I understand that there's
>> an API to "see inside" the journal of NTFS, but that does not mean
>> that a journal actually exists---it means that microsoft wants to
>> be competative, so they SAY that it actually exists.
>
> Sorry, but this is not correct.
>
> Please see these links:
>
> http://www.microsoft.com/whdc/system/winpreinst/ntfs-preinstall.mspx
> http://faq.arstechnica.com/link.php?i=1227
> http://arstechnica.com/paedia/n/ntfs/ntfs4-3.html
> http://www.extremetech.com/article2/0,1697,1168822,00.asp
> http://articles.techrepublic.com.com/5102-10880-6091177.html
> http://www.answers.com/topic/ntfs
> http://www.mcmillan.cx/ntfs.html
>
> NTFS is indeed a metadata journaled file system.
>
Where exactly is this metadata kept?
> Not sure why you are harping on this great file system... most
> technical sites agree that NTFS is *one* of the best file systems
> around.
>
What you mean to say is most windows-centric, microsoft sucking
technical sites. Those of us who have real world experience with
a very wide variety of technologies know that NTFS is shit in
comparison to everything current.
> It's something that MS did right.
>
>> How do I know? Two reasons:
>>
>> 1. NTFS is not a reliable filesystem--if it were journaled, it would be
>
> NTFS *is* a very reliable filesystem. Do you have links to prove your
> statement?
>
Nope, I have realworld experience. In fact, I've rarely been involved
in the deployment of a windows-based system that did not include the
deployment of Veritas Volume Manager (and therefore VxFS) along with
it, because NTFS sucks so horribly.
>> 2. writing to an NTFS filesystem does NOT require access to a journal
>> of any kind.
>
> I have no idea what you are trying to say here...
>
What I'm saying is that if you do raw writes to an NTFS filesystem
with any operating system you like, you need not update a journal
of any kind in order to keep filesystem integrity.
> You need to get your facts straight about NTFS, that is for sure.
>
> A quick search on google will bring up at least 20 good articles about
> NTFS... I linked to some of the more straight forward ones.
>
Thanks, I dont really read too well though.
-----yttrx
cc
$LogFile, nice job reading the links provided.
> > Not sure why you are harping on this great file system... most
> > technical sites agree that NTFS is *one* of the best file systems
> > around.
>
> What you mean to say is most windows-centric, microsoft sucking
> technical sites. Those of us who have real world experience with
> a very wide variety of technologies know that NTFS is shit in
> comparison to everything current.
>
> > It's something that MS did right.
>
> >> How do I know? Two reasons:
>
> >> 1. NTFS is not a reliable filesystem--if it were journaled, it would be
>
> > NTFS *is* a very reliable filesystem. Do you have links to prove your
> > statement?
>
> Nope, I have realworld experience. In fact, I've rarely been involved
> in the deployment of a windows-based system that did not include the
> deployment of Veritas Volume Manager (and therefore VxFS) along with
> it, because NTFS sucks so horribly.
>
> >> 2. writing to an NTFS filesystem does NOT require access to a journal
> >> of any kind.
>
> > I have no idea what you are trying to say here...
>
> What I'm saying is that if you do raw writes to an NTFS filesystem
> with any operating system you like, you need not update a journal
> of any kind in order to keep filesystem integrity.
In order to keep filesystem integrity you do. Where do you get this
information?
> > You need to get your facts straight about NTFS, that is for sure.
>
> > A quick search on google will bring up at least 20 good articles about
> > NTFS... I linked to some of the more straight forward ones.
>
> Thanks, I dont really read too well though.
>
Obviously.
yttrx
Yeah, maybe you should just run along and let the grownups talk,
k nubbs?
Thanks.
-----yttrx
cc
Oh so it's grownup talk to go "Hey, the journal doesn't
exists" [poster puts links showing you where it is] "Ok, but where is
it stored" [someone points out the links] [you use the same played out
comeback *yawn*]
Maybe if you read the information provided to you, didn't go around
making grand claims about your "real world" experience with no
evidence to back it up, and didn't cry like a bitch after the previous
two points were pointed out to you, you'd actually get treated like an
adult. Who are you, Peter Kohlmann?
NoNamer
Would like to be able to help you, but I'm not that technically
knowledgable enough regarding NTFS to be able to answer that question.
yttrx
NTFS is NOT a journaled filesystem, PERIOD. It's no more journaled
than NT/2000/2003/XP/Vista are POSIX compliant (which yes, is yet another
fucking microsoft lie), AND...you're an idiot for thinking it is,
case closed.
I know you're probably some kind of excited computer science major
somewhere who dreams of one day pressing shiny buttons on an Intel
based HP SUPERCOMPUTER running VISTA EXTREME, but really you'll just
end up doing tech support for a pharmaceutical spammer.
-----yttrx
yttrx
NoNamer
I have the integrity to say I don't know something when I don't know
something.
You, on the other hand, keep saying the same thing even after I give
you link after link contradicting what you have said.
yttrx
Linonut
Here it is, more like what you'd think, but still faster than the ASrock
box or Eric's Xeon:
$ time find /usr -print | wc -l
221428
real 2m27.158s
--
Windows XP. The operating system with a load in its pants.
cc
Why isn't it journaled? If it has a journal, then doesn't that make it
journaled filesystem? Or is it because you can't do all the things
with the NTFS log that you would expect out of a journaled filesystem?
> I know you're probably some kind of excited computer science major
> somewhere who dreams of one day pressing shiny buttons on an Intel
> based HP SUPERCOMPUTER running VISTA EXTREME, but really you'll just
> end up doing tech support for a pharmaceutical spammer.
>
That's rich. Sorry, no tech support for me. I do mostly embedded linux
development, some windows development(.Net lately) as well. Is it my
dream? Well, not working at all is my dream, but as I am not rich I
have to do something. You don't do shit but bitch on Usenet all day
and try to pretend like you know what you're talking about. Peter
Kohlmann, or Roy Schestowitz, I can't tell which one you resemble
most.
yttrx
Maverick
I got: root# time find . -print | wc -l
970
real 0m0.389s
user 0m0.019s
sys 0m0.166s
on my ppc running OS X.
Looks like there are a few things that Vista just couldn't copy.
yttrx
What, like being able to list 970 files in root's directory? That's
a bit different than the ROOT directory, you know. /root as opposed
to /. Get it?
-----yttrx
Oliver Wong
news:5ZMyh.4637$Yy1...@textfe.usenetserver.com...
> Erik Funkenbusch <er...@despam-funkenbusch.com> wrote:
>> On Thu, 08 Feb 2007 20:38:04 GMT, yttrx wrote:
>>
>>> So answer it already. Lets hear it.
>>
>> No.
>
> I see.
>
> Just admit that NTFS is NOT A JOURNALED FILESYSTEM. That you've
> BEEN LYING.
http://en.wikipedia.org/wiki/Ntfs
<quote>
NTFS has several improvements over FAT such as improved support for metadata
and the use of advanced data structures to improve performance, reliability,
and disk space utilization plus additional extensions such as security
access control lists and file system journaling. The exact specification is
a trade secret of Microsoft.
</quote>
http://technet2.microsoft.com/WindowsServer/en/library/8cc5891d-bf8e-4164-862d-dac5418c59481033.mspx?mfr=true
<quote>
As files, folders, and other NTFS objects are added, deleted, and modified,
NTFS enters change journal records in streams, one for each volume on the
computer.
</quote>
- Oliver
Handover Phist
Holy crap that's fast, then again there's not much in the way of files
there.
jason@jason:~$ time find . -print | wc -l
63784
real 0m11.035s
user 0m0.092s
sys 0m0.584s
jason@jason:~$
--
"I wake up every morning and I wish I were dead, and so does Jim."
[Tammy Fae Bakker]
Peter Köhlmann
> On Thu, 08 Feb 2007 20:38:04 GMT, yttrx wrote:
>
>> So answer it already. Lets hear it.
>
> No.
Still running from that question, Erik?
Well, you better should. It was one of the first where you blundered in a
major way. Still claiming that EXT3 is much worse than NTFS in the handling
of the journal data and its location?
--
Security is one of those funny things. You can talk about being "more"
secure, but there's no such thing. A vulnerability is a vulnerability, and
even one makes you just as insecure as anyone else. Security is a binary
condition, either you are or you aren't. - Funkenbusch 1 Oct 2006
Oliver Wong
> NoNamer <grug...@hotmail.com> wrote:
>>
>> I have the integrity to say I don't know something when I don't know
>> something.
>>
>> You, on the other hand, keep saying the same thing even after I give
>> you link after link contradicting what you have said.
>>
>
> You're a link monkey. What exactly do you do for a living?
yttrx, if someone has evidence that you are wrong about something, how
would you prefer they behave?
(a) Present the evidence to you.
(b) Claim to have some evidence somewhere, but not present it, and hope
you take his/her word for it.
(c) Silently let you continue believing falsehoods.
- Oliver
Oliver Wong
"AB" <fardb...@gmail.com> wrote in message
news:9p2u94-...@clark.harry.net...
> On 2007-02-08, Erik Funkenbusch <er...@despam-funkenbusch.com> claimed:
>>
>> However, the piont still stands that using the GUI to delete is different
>> from the command prompt. The GUI must count all the files and their
>> sizes
>> to display status information, command prompts don't. With a a large
>> directory tree with a lot of small files, this can take some time.
>
> Proof that the GUI method is plain stupid (if you're right, which I
> actually don't believe).
>
> Back to the questions that you made me come up with while you were
> TIPPETY-TAPping around, apologizing for MS:
>
> 1. *WHY* the fuck does NTFS have to count the files before deleting
> them?? I knew they were complete imbeciles at that place, but I never
> realized they might be that bald-assed stupid! Admit it, you made that
> up.
When I delete a group of files using the GUI in XP, it'll show a
confirmation box "Are you sure you want to send these N items to the
recycling bin?" where Nis replaced by the appropriate number. I wouldn't be
surprised if Vista did the same thing. This isn't a requirement of NTFS, but
rather a design choice (arguably a poor one) for the Explorer team.
I had assumed Ubuntu presented a similar dialog when you try to delete
from the GUI (i.e. Gnome). Maybe I'll test that tonight.
- Oliver
Erik Funkenbusch
> 1. *WHY* the fuck does NTFS have to count the files before deleting
> them?? I knew they were complete imbeciles at that place, but I never
> realized they might be that bald-assed stupid! Admit it, you made that
> up.
So that a) It can pop up a dialog saying "You're about to delete 2,000,000
files, are you sure?" and then b) can show an accurate progress bar during
the delete and estimate time remaining.
> 2. Wouldn't you say one would have to be completely retarded, even more
> retarded than retarded, to have a window of flying-whatever to show the
> user that it's doing what it's told to do? I mean, it's going to count
> the files (probably a load of crap) for 10 minutes, then it's going to
> have to waste about 75% of it's time making shit move across the screen
> in a dance that says "Look! I'm really doing something!" 'Fess up.
> The guys making all of this shit up are a big group of Jerry's Kids,
> aren't they?
Most people like feedback from their computer about what it's doing. This
is one of the reasons Windows wins with non-technical users, and one of the
reasons Linux people can't understand why people like Windows.
AB
And one thing the troolies are going to have trouble claiming anyone
copied from Vister.
--
I use linux because I've used Windows.
Peter Köhlmann
> On Fri, 9 Feb 2007 05:33:29 -0600, AB wrote:
>
>> 1. *WHY* the fuck does NTFS have to count the files before deleting
>> them?? I knew they were complete imbeciles at that place, but I never
>> realized they might be that bald-assed stupid! Admit it, you made that
>> up.
>
> So that a) It can pop up a dialog saying "You're about to delete 2,000,000
> files, are you sure?" and then b) can show an accurate progress bar during
> the delete and estimate time remaining.
>
You mean they have not counted those files while selecting them?
And added their file sizes while at it?
Well, Erik, those coding monkeys at MS are even dumber than bread, it seems
Programmers with half a working brain would do these things while at the
selection stage, when there is *lots* of processing power to display that
so all-important statistics
< snip more Erik F idiocy >
--
Microsoft's Guide To System Design:
Form follows malfunction.
yttrx
None of the above.
(d) SHUT THE FUCK UP ABOUT IT UNTIL I ASK FOR YOUR INPUT, WHICH I WILL
VERY LIKELY NEVER DO, DICKHEAD.
Unless I think you're pretty cool, then (a), and I'll of course back
off immediately, do the research, and come back and correct my mistake.
See GreyCloud, GITM, etc. for details.
-----yttrx
Linonut
> http://en.wikipedia.org/wiki/Ntfs
> <quote>
> NTFS has several improvements over FAT such as improved support for metadata
> and the use of advanced data structures to improve performance, reliability,
> and disk space utilization plus additional extensions such as security
> access control lists and file system journaling. The exact specification is
> a trade secret of Microsoft.
> </quote>
>
> http://technet2.microsoft.com/WindowsServer/en/library/8cc5891d-bf8e-4164-862d-dac5418c59481033.mspx?mfr=true
> <quote>
> As files, folders, and other NTFS objects are added, deleted, and modified,
> NTFS enters change journal records in streams, one for each volume on the
> computer.
> </quote>
One little problem with Microsoft is their use of words:
Normal IT people Microsoft
failover clustering
serialization marshalling
directory folder
function method
C++/CLI C++
copying innovation
stable crazy uptimes
??? journalling
--
"I'm going to f'in *kill* Google!"
-- Steve Ballmer, CEO Microsoft
Maverick
Opps. I forgot to cd /.
/ root# time find . -print | wc -l
591284
real 3m13.779s
user 0m5.566s
sys 0m44.087s
And from /usr
/usr root# time find . -print | wc -l
33431
real 0m0.600s
user 0m0.099s
sys 0m0.396s
Not too bad.
Maverick
No, I forgot to cd /.
cc
Yeah, that's what I thought, you're a stay at home mom.
yttrx
Wow, you're good. Actually, the fact is that as soon as you mentioned
that you're a .NET developer I stopped being interested in anything you
have to say.
-----yttrx
cc
> Oliver Wong <o...@castortech.com> wrote:
> > "yttrx" <y...@yttrx.net> wrote in message
> >news:763zh.4661$Yy1....@textfe.usenetserver.com...
> >> NoNamer <grug2...@hotmail.com> wrote:
>
> >>> I have the integrity to say I don't know something when I don't know
> >>> something.
>
> >>> You, on the other hand, keep saying the same thing even after I give
> >>> you link after link contradicting what you have said.
>
> >> You're a link monkey. What exactly do you do for a living?
>
> > yttrx, if someone has evidence that you are wrong about something, how
> > would you prefer they behave?
>
> > (a) Present the evidence to you.
> > (b) Claim to have some evidence somewhere, but not present it, and hope
> > you take his/her word for it.
> > (c) Silently let you continue believing falsehoods.
>
> > - Oliver
>
> None of the above.
>
> (d) SHUT THE FUCK UP ABOUT IT UNTIL I ASK FOR YOUR INPUT, WHICH I WILL
> VERY LIKELY NEVER DO, DICKHEAD.
Feeling underappreciated? Don't feel bad, I'm sure your man really
does appreciate all the hard work you do all day around the house.
Just because he comes home yelling about why dinner isn't made doesn't
mean you need to take it out on others.
Maverick
Well, the times I got were from ~root rather than from /. But the times
aren't too much slower than linux.
I've still got to resolve another issue with the Apple version of
tcl/tk. A certain program that uses puts "Hi\nto all" only puts the hi
on screen but not the to all part. Not sure whether it is a bug or a
misconfiguration. Works fine on my other Suse PC.
I think I'll move eventually to a dual core x64 box and just install
some version of linux on it. OS X isn't all that bad, and is way much
better than XP or Vista. I'll get the family pack for Leopard when it
comes out and after I've looked for any major complaints.
yttrx
cc
I know you have trouble reading, but since you got to the second part
of the sentence, I figured you'd read the first part. You know where I
said I do mostly embedded linux work. Or do you have something against
that too. Go do some more dishes.
yttrx
That you are actually willing to sit down and write .NET code under
ANY circumstance negates everything else you've ever said.
And luckily I have a dishwasher!
-----yttrx
yttrx
cc
Sitting down and writing .NET code is a hell of lot better than being
jobless. We can't all have a sugar daddy like you do.
cc
Which links? The link to your shitty ass blog? Or the links to the
websites which contradict everything you say? I've clicked on both.
yttrx
As ive made the offer a million times, I suppose I'll make it a million
and one. You're more than welcome to stop by my office in manhattan any
time you like. We'll do a quick lunch, and I'll even get it myself just
to lift some of the strain of being a dipshit for you. I know that must
be hard.
-----yttrx
yttrx
Yeah, not until after you read my reply, mister 71.166.26.108. You can't
hide from me.
But please do not look at my blog anymore. You're not welcome there.
I don't want to have to ban the entire 71.* class A, but I sure as hell
will if I have to.
-----yttrx
AB
>
>> 1. *WHY* the fuck does NTFS have to count the files before deleting
>> them?? I knew they were complete imbeciles at that place, but I never
>> realized they might be that bald-assed stupid! Admit it, you made that
>> up.
>
> So that a) It can pop up a dialog saying "You're about to delete 2,000,000
> files, are you sure?" and then b) can show an accurate progress bar during
> the delete and estimate time remaining.
OK, so it can nag me. Even if I'd rather it just do the effin' job and
get on with it.
But "accurate progress bar"? You've never done this, have you?
>> 2. Wouldn't you say one would have to be completely retarded, even more
>> retarded than retarded, to have a window of flying-whatever to show the
>> user that it's doing what it's told to do? I mean, it's going to count
>> the files (probably a load of crap) for 10 minutes, then it's going to
>> have to waste about 75% of it's time making shit move across the screen
>> in a dance that says "Look! I'm really doing something!" 'Fess up.
>> The guys making all of this shit up are a big group of Jerry's Kids,
>> aren't they?
>
> Most people like feedback from their computer about what it's doing. This
> is one of the reasons Windows wins with non-technical users, and one of the
> reasons Linux people can't understand why people like Windows.
Most people don't *like* that kind of feedback. Not after a time or
two. Most people have learned to /live with/ that kind of feedback. On
a delete. On one that's already nagged me about getting rid of what I
told it to get rid of.
--
When all else fails, you must be using Windows.
cc
What the fuck does your "office" have to do with anything? Nice of you
to make fun of someone for doing a little .NET coding, when you
completely sold out to get a job. Friends don't let other friends cut
their hair. You're a poser here just like you are in real life, now
move along.
cc
That will affect so many people though! Oh wait, just me. And I saw
the made up story of you getting your panties in a bunch and yelling
at a biker. There's nothing there I can't live without.
yttrx
Uh...how exactly did I sell out? And you do realize that I have quite
public tattoos, right? I still know how to rock and roll, punk.
Actually getting my current position was some pretty hard damn work, and
it remains very challenging almost every minute of every day, and never
once have I had to compromise my values to do it.
Be gone, .NET monkey.
-----yttrx
yttrx
You're free of course to come visit and see exactly what I'm like and
how I live any time you like. I'm not demanding it of course, it is
entirely your perogative, but it would require you being pretty ok with
being proven quite wrong about a lot of things.
-----yttrx
cc
No thanks, watching a poser do laundry for his sugar daddy is not what
I consider a good time. And anyone who has cut his hair, talks about
how nice his fucking watch is, and talks about how much money he
makes, but insists that his public tattoos make it all kosher is a
fucking poser sell-out, there's no way around it. Isn't it dinner time?
The Ghost In The Machine
<lin...@bone.com>
wrote
on Fri, 09 Feb 2007 17:38:58 -0600
<QvSdnbEt8ZePm1DY...@comcast.com>:
> After takin' a swig o' grog, Oliver Wong belched out this bit o' wisdom:
>
>> http://en.wikipedia.org/wiki/Ntfs
>> <quote>
>> NTFS has several improvements over FAT such as improved support for metadata
>> and the use of advanced data structures to improve performance, reliability,
>> and disk space utilization plus additional extensions such as security
>> access control lists and file system journaling. The exact specification is
>> a trade secret of Microsoft.
>> </quote>
>>
>> http://technet2.microsoft.com/WindowsServer/en/library/8cc5891d-bf8e-4164-862d-dac5418c59481033.mspx?mfr=true
>> <quote>
>> As files, folders, and other NTFS objects are added, deleted, and modified,
>> NTFS enters change journal records in streams, one for each volume on the
>> computer.
>> </quote>
>
> One little problem with Microsoft is their use of words:
>
> Normal IT people Microsoft
>
> failover clustering
> serialization marshalling
Pedant Point: "Marshalling" is also used by Java
and by projects such as Castor and SOAP.
Just to confuse both of us. :-)
> directory folder
> function method
Pedant Point #2: a function requires no class context;
a method requires either class or object context. In
C++ one has the following cases:
function: void something() {}
method, static: class C { public: static void something(); }
method, object: class C { public: void something(); }
method, virtual: class C { public: virtual void something(); }
There is a method pointer as well, but the syntax is
frightful in C++. :-)
> C++/CLI C++
Wow. Not sure about that one. :-)
> copying innovation
> stable crazy uptimes
> ??? journalling
>
spindle or drive ???
partition drive
file document
clustering ???
GUIfied easy to use
??? delegate (C++/C#)
hides gets rid of (e.g., DOS)
kills cuts off air supply
billions of $ free (e.g. IE development)
free high TCO
It's like another planet. :-)
--
#191, ewi...@earthlink.net
"Woman? What woman?"
--
Posted via a free Usenet account from http://www.teranews.com
Tim Fairchild
I was able to see cc say:
Oh fucking plonk to this crap...
--
Tim Fairchild - Queensland Australia
Mandriva Linux release 2007.0 (Official) for x86_64
Linux 2.6.20 x86_64 AMD Athlon(tm) 64 Processor 2800+ GNU/Linux
NVRM version: NVIDIA UNIX x86_64 Kernel Module 1.0-9746
cc
Like it's such a strain on your being to read it. COLA is no place for
the weak. That's why Roy Culley left.
Erik Funkenbusch
> One little problem with Microsoft is their use of words:
>
> Normal IT people Microsoft
>
> failover clustering
Failover is mirroring, clustering is load balancing.
> serialization marshalling
Marshalling and serialization are two different things. Serialization is,
for example, writing an object to a file. Marshalling is transfering an
object across the wire.
> directory folder
This is one of those euphemisms to help novices understand the concept.
> function method
Functions and methods are different. A method is a function that is a
member of anobject. A function is a standalone procedure.
> C++/CLI C++
No, they call it C++/CLI.
Maverick
>>
>>Opps. I forgot to cd /.
>>
>>/ root# time find . -print | wc -l
>> 591284
>>
>>real 3m13.779s
>>user 0m5.566s
>>sys 0m44.087s
>>
>>And from /usr
>>
>>/usr root# time find . -print | wc -l
>> 33431
>>
>>real 0m0.600s
>>user 0m0.099s
>>sys 0m0.396s
>>
>>Not too bad.
>>
>
>
> Yeah, those numbers make a bit more sense...:)
>
Well, at least you pay attention to the small details.
One of the qualifications to an IT admin type. :-)
I'm just getting long in the tooth is all.
yttrx
yttrx
yttrx
I can't help it. It's training I wish I could forget.
But back in The Day (TM) it did make it pretty fast to sift
through router logs manually.
-----yttrx