It is ridiculous because we have no need to reverse engineer anything in that spec because we already implement everything in that spec. I'm sorry I thought I was being clear above.
The 2.2.4 comment was in response to the "new features" question.
This spec is a *subset* of what Samba already implements. The SNIA CIFS spec (which we helped to create) already documents far more than is in this spec. (Not that I've read it this spec, obviously, but I've spoken to people who have read both).
This spec. is an irrelevence. Try implementing it to the letter and see how many Microsoft clients actually *work* against you. (Hint - none:-).
> The Kerberos-like authentication is apparently > much stronger, but there don't seem to be any > open implementations of it yet.
Yes there are. Andrew Tridgell has coded one up inside Samba 3.0.x (still in alpha). Available as the Samba HEAD CVS tree. It'll be the standard auth mechanism for 3.0.
Regards,
Jeremy Allison,
Samba Team.
Re:Ability to tag friend or foe
on
Slashdot Code Update
·
· Score: 3, Insightful
No, this is not a good example. Noam Chomsky doesn't have a daily syndicated radio show.
In fact I defy you to name *one* left wing commentator who does within the USA.
And the right wing still bleats about "liberal bias....".
CIFS is just a marketing name. The protocols on ports 445 and 139 are functionally identical, and you can call it CIFS or SMB or my favourite ("a hideous wart":-) or whatever you like.
I was at SGI during the Belluzzo years. I have to agree with this comment. How this man escaped criminal proceedings against him is beyond me....
The best comment I read on the SGI mailing lists when Belluzzo went to Microsoft (at about the same time that Start Wars, Episode one was released) was "so now it is revealed, there are always two, a master *and* an apprentice...".
> For intsance, user-level applications do not
> have to be rewritten to move from FAT to NTFS
> or to support zip disks. Porting between
> versions of Windows is less effort than
> porting between versions of Unix.
Errr - yes they do if they want to do any security work - which isn't available on FAT.
Or if they want to use the many API's that are "only supported in Windows NT/2000, but not on Win9x" (just look in the MSDN - there are many of them). Locking API's are completely different between Win9x and WinNT/200 for example. Why is LockFileEx unsupported in Win9x ? Why are such basic things as locking a region of a file different between Microsoft platforms ?
Porting between Win9x/WinNT/Win2000 is a *nightmare* compared to porting between UNIX varients. Not if you're doing simple "hello world" window GUI stuff, but if you're doing anything of any complexity you run into such issues (security, file locking etc.) all the time.
> The way it would work:
> You license your code such that it is Open.
> I take your code gratefully and change it a
> little.
> I build the binary and release the binary as
> my own product with attribution given to your
> product.
There's the problem. The people who write GPL code (myself included) explicitly don't want you to be able to "release only the binary" if you're using our code.
If you don't like it, don't use our code, write your own. I don't see you complaining that you can't use code from Microsoft, Sun, Oracle, etc. Why do you think GPL code is any different ?
If you don't like the costs of the license, don't use the code.
It won't work. Samba requires working seteuid() calls, and full POSIX locking functionality, which is not possible to emulate in a Win32 program (although probably possible in a native NT API program, via hidden calls - you know, the ones Microsoft claim don't exist:-).
That's why Microsoft's "Services for UNIX" product must have a kernel component - Win32 locking is unbearably primitive compared to POSIX locking. We can emulate Win32 locking semantics on top of POSIX, but it's not possible to do this the other way around.
Regards,
Jeremy Allison,
Samba Team.
Re:Yawn. Why not a GPL'd NFS driver for windows.
on
MS getting rid of SAMBA?
·
· Score: 3, Interesting
> samba still cannot serve the "user list" to
> windows 9x machines for USER level sharing.
Finally fixed in Samba 2.2.1a. I'm sure you'll now upgrade...:-):-).
> They might release some kind of "client
> upgrade", which coincidentally breaks Samba.
Sigh. What do you think Windows 2000 did ? Why do you think we had to get Samba 2.2.1 out as soon as we did ? What do you think Windows XP is planning *right now* (search the Samba lists for the new breakage in XP... it's not hard to find).
They do this *EVERY RELEASE*.
This is what it means to be on a Microsoft treadmill. I want to warn the Ximian/Mono folks not to get into the same situation that we're in.....
You don't really think Microsoft exposes all the needed API's to write a redirector without getting WNT/W2K source code access, do you ?
There's not even a"redirector writers kit" you can buy ! Microsoft doesn't *want* people to be able to write replacement redirectors. If you could do that, you might reduce dependencies on Windows Domain/ADS servers - why, you might even plug in your own authentication client, removing the need for a PDC/ADS server ! That would never do, now would it. Where would the monopoly go then ?
That's why almost no one writes decent replacement redirectors for Microsoft clients except Microsoft.
Why do you think all the PC/NFS products don't work very well ? Why do you think anyone who has to support Windows clients in a serious way (for a NAS product etc.) has to implement SMB ? It isn't because it's a beautiful or elegent protocol:-):-).
*Very* good point. I didn't say they were planning to enforce it. I didn't really want to talk about it much at all, and wish it hadn't ended up at ZDnet or Slashdot to be honest. I tried to get Charlie to remove this section from the article, but it was one of the only pieces of evidence that he had about patents, and in his judgement that made it important enough to mention.
It's not "sensationalist jounalism", though. Charlie is trying to make an important point which I will discuss below.
The reason I spoke about it at all is that my personal feelings are that implementing *NEW* Microsoft-revisioned protocols is a waste of people's time. Once they've become a de-facto standard, like SMB, then we have no choice but to try and implement them, just in the same way that Abiword, StarOffice and KWord have to load Microsoft Word file formats.
But to start implementing new Microsoft designed protocols and *help* them become ubiquitous is insane. All IMHO of course.
I don't think Microsoft is planning to wipe out Samba and it is sheer paranoia to speculate on that point.
Actually it was me that said "SMB sucks", although tridge has mentioned similar statements:-).
But I understand. People over here think we're both Australian for some reason:-). Tridge and I have speculated that we're both part of some wierd gestalt entity in people's minds:-).
Jeremy Allison,
Samba Team.
Quite simply the best...
on
Lord of Light
·
· Score: 3
Novel I've ever read. I keep having to buy new copies as I keep giving them away to friends:-).
My favourite quote : "So that's what they're playing on their fascist banjo's these days..."
after Yama lectures Sam on the "true" meaning of being a god to the populace:-).
I wish the reviewer hadn't given away the fact that none of the "gods" are real, and that" Niriti the black" was the ships chaplain:-). I discovered these wonderful suprises myself whilst reading the book.
I spent the first third of the book wondering what the hell was going on, then immediately had to re-read it once I'd finished it (after going "oh... that's what it all meant").
Well air is definately an essential resource:-). But Samba is probably more directly related to the business products from HP, SGI, Veritas, IBM and Sun than air (both are essential however:-).
Just because we don't have a samba.com doesn't mean we're not a business success. There are (reasonable) estimates that 30% of Windows clients connect to some Samba server. People claim apache as a business success all the time, there isn't an official apache.com. There are however, many businesses based upon it. Or do you separate out the success of Apache (>60% market share, that's a business success in my eyes) from that of Samba because Apache uses a license you agree with ?
Your comment "programmers cannot make money off the code they actually generate" is incorrect. I have personally been living from the work I do on Samba for nearly five years now. I'm making money from what I do, and what I do is write GPL code.
Regards,
Jeremy Allison,
Samba Team.
Re:The Million MBA... er, lemming march
on
Mandrake Shakeup
·
· Score: 2
Both IBM and HP have donated significant pieces of code to Samba (more from HP to be honest). You can't discount the business decisions of such companies as being from "instant MBA's". I know you disagree with their decisions, however they're doing it to maximize shareholder value in the best way they see how. Many of them have chosen to help GPL projects. Given this fact, you can't realistically argue the GPL is anti-business.
I don't think they're trying to leverage a fad either. I think they see the GPL as their only hope to compete with an existing monopoly. Donating code to the BSD movement does then no good, as Microsoft can co-opt it (as they and NetApp have with BSD based code already).
They're using the GPL as a business weapon.
I don't think the GPL had anything to do with the failure of Eazel or Mandrake. Management, and bad luck (bubble bursting when it did) had more to do with it than any software license.
Slashdot Mirror
It is ridiculous because we have no need to reverse
engineer anything in that spec because we already
implement everything in that spec. I'm sorry I
thought I was being clear above.
The 2.2.4 comment was in response to the "new features"
question.
Sorry for being unclear.
Regards,
Jeremy Allison,
Samba Team.
This comment is plain ridiculous (not +2 insightful) !
There is nothing in this spec that Samba has not already
implemented.
This spec is irrelevent to Samba.
As to the "will have to switch to a BSD license in order
to add features to Samba".... words fail me !
If you want new features (such as a recycle bin or the
new NT-ACL code) then just keep doing a CVS update from
samba.org. Next release will be 2.2.4.
Regards,
Jeremy Allison,
Samba Team.
Because we don't all agree on the BSD license. I won't
work on any BSD licensed code any more for example.
Look at the reasons that Wine changed from BSD style
licensing to LGPL for a clue as to why that might be....
Regards,
Jeremy Allison,
Samba Team.
This always comes up. What, do you think Microsoft are
:-).
*stupid* or something ?
Do you actually think they document the internal interfaces
needed to write such a thing (NFS client for Windows
NT/2000/XP) ?
Go ahead and try. Good luck !
Jeremy Allison,
Samba Team.
Don't be silly.
:-).
This spec is a *subset* of what Samba already implements.
The SNIA CIFS spec (which we helped to create) already
documents far more than is in this spec. (Not that I've
read it this spec, obviously, but I've spoken to people
who have read both).
This spec. is an irrelevence. Try implementing it to
the letter and see how many Microsoft clients actually
*work* against you. (Hint - none
Jeremy Allison,
Samba Team.
> The Kerberos-like authentication is apparently > much stronger, but there don't seem to be any > open implementations of it yet.
Yes there are. Andrew Tridgell has coded one up inside Samba 3.0.x (still in alpha). Available as the Samba HEAD CVS tree. It'll be the standard
auth mechanism for 3.0.
Regards,
Jeremy Allison,
Samba Team.
No, this is not a good example. Noam Chomsky doesn't have a daily syndicated radio show.
In fact I defy you to name *one* left wing commentator who does within the USA.
And the right wing still bleats about "liberal bias....".
Jeremy.
You're wrong.
:-) or whatever you like.
CIFS is just a marketing name. The protocols on ports 445 and 139 are functionally identical, and you can call it CIFS or SMB or my favourite ("a hideous wart"
CIFS is not a replacement for SMB. CIFS *is* SMB.
Regards,
Jeremy Allison,
Samba Team.
I was at SGI during the Belluzzo years. I have to agree with this comment. How this man escaped criminal proceedings against him is beyond me....
:-).
The best comment I read on the SGI mailing lists when Belluzzo went to Microsoft (at about the same time that Start Wars, Episode one was released) was "so now it is revealed, there are always two, a master *and* an apprentice...".
:-)
Great place to work though - GO SGI !!!!!
Jeremy.
> For intsance, user-level applications do not
> have to be rewritten to move from FAT to NTFS
> or to support zip disks. Porting between
> versions of Windows is less effort than
> porting between versions of Unix.
Errr - yes they do if they want to do any security work - which isn't available on FAT.
Or if they want to use the many API's that are "only supported in Windows NT/2000, but not on Win9x" (just look in the MSDN - there are many of them). Locking API's are completely different between Win9x and WinNT/200 for example. Why is LockFileEx unsupported in Win9x ? Why are such basic things as locking a region of a file different between Microsoft platforms ?
Porting between Win9x/WinNT/Win2000 is a *nightmare* compared to porting between UNIX varients. Not if you're doing simple "hello world" window GUI stuff, but if you're doing anything of any complexity you run into such issues (security, file locking etc.) all the time.
Regards,
Jeremy Allison,
Samba Team.
No it isn't.
Unfortunately this article is
comparing apples and oranges.
The Win32 call you need to use is
CreatePipe(), not CreateNamedPipe().
CreatePipe is exactly equivalent to
the UNIX pipe() call. CreateNamedPipe
with the \\pipe prefix is equivalent
to mkfifo on UNIX.
No wonder Win32 is much slower, you're
going through many more layers in the
kernel.
Regards,
Jeremy Allison,
Samba Team.
Unfortunately this article is
comparing apples and oranges.
The Win32 call you need to use is
CreatePipe(), not CreateNamedPipe().
CreatePipe is exactly equivalent to
the UNIX pipe() call. CreateNamedPipe
with the \\pipe prefix is equivalent
to mkfifo on UNIX.
No wonder Win32 is much slower, you're
going through many more layers in the
kernel.
Regards,
Jeremy Allison,
Samba Team.
Unfortunately this article is
comparing apples and oranges.
The Win32 call you need to use is
CreatePipe(), not CreateNamedPipe().
CreatePipe is exactly equivalent to
the UNIX pipe() call. CreateNamedPipe
with the \\pipe prefix is equivalent
to mkfifo on UNIX.
No wonder Win32 is much slower, you're
going through many more layers in the
kernel.
Regards,
Jeremy Allison,
Samba Team.
AC writes:
> The way it would work:
> You license your code such that it is Open.
> I take your code gratefully and change it a
> little.
> I build the binary and release the binary as
> my own product with attribution given to your
> product.
There's the problem. The people who write GPL code (myself included) explicitly don't want you to be able to "release only the binary" if you're using our code.
If you don't like it, don't use our code, write your own. I don't see you complaining that you can't use code from Microsoft, Sun, Oracle, etc. Why do you think GPL code is any different ?
If you don't like the costs of the license, don't use the code.
Regards,
Jeremy Allison,
Samba Team.
It won't work. Samba requires working seteuid() calls, and full POSIX locking functionality, which is not possible to emulate in a Win32 program (although probably possible in a native NT API program, via hidden calls - you know, the ones Microsoft claim don't exist :-).
That's why Microsoft's "Services for UNIX" product must have a kernel component - Win32 locking is unbearably primitive compared to POSIX locking. We can emulate Win32 locking semantics on top of POSIX, but it's not possible to do this the other way around.
Regards,
Jeremy Allison,
Samba Team.
> samba still cannot serve the "user list" to
:-) :-).
> windows 9x machines for USER level sharing.
Finally fixed in Samba 2.2.1a. I'm sure you'll now upgrade...
Jeremy Allison,
Samba Team.
> They might release some kind of "client
> upgrade", which coincidentally breaks Samba.
Sigh. What do you think Windows 2000 did ? Why do you think we had to get Samba 2.2.1 out as soon as we did ? What do you think Windows XP is planning *right now* (search the Samba lists for the new breakage in XP... it's not hard to find).
They do this *EVERY RELEASE*.
This is what it means to be on a Microsoft treadmill. I want to warn the Ximian/Mono folks not to get into the same situation that we're in.....
Regards,
Jeremy Allison,
Samba Team.
An AC wrote :
:-) :-).
> I assume that "high Microsoft official" was
> probably yanking Allison's (and indirectly,
> slashdot's) chain.
Nope. Definately not. Also it was 2 years ago, not 2 months, but journalism isn't an exact science
Jeremy Allison,
Samba Team.
You don't really think Microsoft exposes all the needed API's to write a redirector without getting WNT/W2K source code access, do you ?
:-) :-).
There's not even a"redirector writers kit" you can buy ! Microsoft doesn't *want* people to be able to write replacement redirectors. If you could do that, you might reduce dependencies on Windows Domain/ADS servers - why, you might even plug in your own authentication client, removing the need for a PDC/ADS server ! That would never do, now would it. Where would the monopoly go then ?
That's why almost no one writes decent replacement redirectors for Microsoft clients except Microsoft.
Why do you think all the PC/NFS products don't work very well ? Why do you think anyone who has to support Windows clients in a serious way (for a NAS product etc.) has to implement SMB ? It isn't because it's a beautiful or elegent protocol
Jeremy Allison,
Samba Team.
*Very* good point. I didn't say they were planning to enforce it. I didn't really want to talk about it much at all, and wish it hadn't ended up at ZDnet or Slashdot to be honest. I tried to get Charlie to remove this section from the article, but it was one of the only pieces of evidence that he had about patents, and in his judgement that made it important enough to mention.
It's not "sensationalist jounalism", though. Charlie is trying to make an important point which I will discuss below.
The reason I spoke about it at all is that my personal feelings are that implementing *NEW* Microsoft-revisioned protocols is a waste of people's time. Once they've become a de-facto standard, like SMB, then we have no choice but to try and implement them, just in the same way that Abiword, StarOffice and KWord have to load Microsoft Word file formats.
But to start implementing new Microsoft designed protocols and *help* them become ubiquitous is insane. All IMHO of course.
I don't think Microsoft is planning to wipe out Samba and it is sheer paranoia to speculate on that point.
Jeremy Allison,
Samba Team.
Actually it was me that said "SMB sucks", although tridge has mentioned similar statements :-).
:-). Tridge and I have speculated that we're both part of some wierd gestalt entity in people's minds :-).
But I understand. People over here think we're both Australian for some reason
Jeremy Allison,
Samba Team.
Novel I've ever read. I keep having to buy new copies as I keep giving them away to friends :-).
:-).
:-). I discovered these wonderful suprises myself whilst reading the book.
My favourite quote : "So that's what they're playing on their fascist banjo's these days..."
after Yama lectures Sam on the "true" meaning of being a god to the populace
I wish the reviewer hadn't given away the fact that none of the "gods" are real, and that" Niriti the black" was the ships chaplain
I spent the first third of the book wondering what the hell was going on, then immediately had to re-read it once I'd finished it (after going "oh... that's what it all meant").
A *perfect* book !
Jeremy Allison,
Samba Team.
No, layout won't change. What'll change is IE specific plugins and x86 Win32-only binaries downloaded to your browser will be expected to run......
Layout is irrelevent. Why bother controlling that standard when you can "extend" the Web experience to make it Microsoft-only.
That's what scares me.
Jeremy Allison,
Samba Team.
Well air is definately an essential resource :-). But Samba is probably more directly related to the business products from HP, SGI, Veritas, IBM and Sun than air (both are essential however :-).
Just because we don't have a samba.com doesn't mean we're not a business success. There are (reasonable) estimates that 30% of Windows clients connect to some Samba server. People claim apache as a business success all the time, there isn't an official apache.com. There are however, many businesses based upon it. Or do you separate out the success of Apache (>60% market share, that's a business success in my eyes) from that of Samba because Apache uses a license you agree with ?
Your comment "programmers cannot make money off the code they actually generate" is incorrect. I have personally been living from the work I do on Samba for nearly five years now. I'm making money from what I do, and what I do is write GPL code.
Regards,
Jeremy Allison,
Samba Team.
Both IBM and HP have donated significant pieces of code to Samba (more from HP to be honest). You can't discount the business decisions of such companies as being from "instant MBA's". I know you disagree with their decisions, however they're doing it to maximize shareholder value in the best way they see how. Many of them have chosen to help GPL projects. Given this fact, you can't realistically argue the GPL is anti-business.
I don't think they're trying to leverage a fad either. I think they see the GPL as their only hope to compete with an existing monopoly. Donating code to the BSD movement does then no good, as Microsoft can co-opt it (as they and NetApp have with BSD based code already).
They're using the GPL as a business weapon.
I don't think the GPL had anything to do with the failure of Eazel or Mandrake. Management, and bad luck (bubble bursting when it did) had more to do with it than any software license.
Regards,
Jeremy Allison,
Samba Team.