why more discussion of samba development?

Luke Kenneth Casson Leighton lkcl at samba-tng.org
Mon Jan 7 14:33:23 GMT 2002

On Sun, Jan 06, 2002 at 03:58:46PM +0100, Volker Lendecke wrote:
> Hi, Luke!
 hi there volker,

 you ready for this?  please remember, this is _not_ directed
 at you, but you _are_ asking some hard-ball questions, and
 you know i answer truthfully and without reserve, so i'm
 presuming that you [and andrew b.] are prepared [psychologically!]
 for the answers...

 deep breath...

> On Sun, Jan 06, 2002 at 12:15:27PM +0000, Luke Kenneth Casson Leighton wrote:
> >  take out the -L option that stupid tridgell added to rpcclient.
> Why do you emphasize what you think about Andrew in every single posting where
> you have to mention 'the other samba'? 

if you read below, and the other message i sent to andrew b.,
you can guage the answer to this one.

> This does not help anybody. 

i could not care less, i am sorry to say.

> And everyone
> now knows that you do not like him. 

i liked and respected andrew a _great deal_.

his actions managed to destroy any respect that i had for
him, which is very sad.

if andrew wishes regain my respect for him, he will take
on board and act upon what i have been recommending for
the last two, two and a half years.

if andrew does not _care_ what i think, then that's very
sad to hear, but there's nothing i can do about that.

> I can really understand that you are
> annoyed, and I always read tng-technical with great interest, but your comments
> sometimes make it hard to enjoy it.

well i am sitting here wondering where i am going to get
enough money to pay my mortgage next month.  in that
context, reading things that make it very clear to me that
people are making money from my work and research and
that i am not just _really_ gets under my skin.

i explained each piece of the architecture to andrew
and he got more and more out of his depth.

he didn't want to admit it or lose control over the code.

in particular, he didn't want to accept technically "inferior"
solutions or "inferior" code.

his response was to fob me off, requiring removal of pieces
of functionality that he could not understand, or to have
me do _anything_ that meant that the code would not meet
his standards.

if andrew had not raised the technical bar higher and
higher and higher in order to, ultimately, block TNG code from
going mainstream, then i would be in a much more stable financial
situation than i am now.

i am working extremely long hours on a building site, earning
approximately one quarter to one half of the amount of money
that i need to pay for everything, and i come home to read
my email once every three days, and find messages that remind
me that some extremely good code that i wrote OVER TWO YEARS
AGO and spent FOUR YEARS WRITING still hasn't made it into
main-stream samba.

that's insane and ridiculous.

and you are asking me to stop making comments about this?

you want me to stop sending comments that spoil _your_ day?

okay, well i'll tell you how you can do that.

you can take a look at the samba-tng source code, take
a look at the dcerpc.net web site, read the articles there,
implement a portable, root-accessible-only
inter-process-communication interface that allows for

- an NT security context to be transferred over it.

- a complete NT user profile including the 16-byte user
session key plus the 8-byte NTLMSSP information
[see fields in NET_USER_INFO_3, basically].

- plus allows some NetBIOS context information to be
transferred [to fake up a NetBIOS-equivalent of
gethostname() - it must be possible for get_client_addr(),
get_client_name(), get_remote_name() and get_global_myname()
to all work]

- plus allow arbitrary-sized message passing with functionality
equivalent to that offered by file descriptors (read, write,
open, close).

_when_ you can come back to me with such an implementation
of the above, that allows independently linking programs
to use the above API and nothing BUT the above API, and
still expect the independent program to receive DCE/RPC
and NT user-context info on port 139 whilst Samba [or any
_other_ SMB server that conforms to the server-side
of the same API] is running, then i will stop making you
"not enjoy" reading tng-technical.

if things get too bad for me, i will just unsubscribe from
tng-technical and all other samba and samba-tng mailing lists.
the open source community has already lost out on a golden
opportunity by not ensuring that i remain involved in tracking
and making up for microsoft's anti-competitive practices.

there are _very_ few people in this world who are capable of
doing what i can [catch up with microsoft man-decades development
in a tenth to one hundredth of the time], and andrew tridgell has demonstrated by
his behaviour whilst we were at linuxcare, and by management
since of samba's development, that he is _not_ such a person.

the deployment of TNG mainstream and its subsequent hardening
and the rough edges cut off in a mainstream development /
release cycle would have achieved in under four [probably
painful!]  months what andrew tridgell and jeremy allison
wanted to see _me_ do [on my own] over a further development
period of one or two years].

that would result in accelerated development the likes of
which we [painfully!] saw with nmbd.  bugs, bugs, galore
_and fixed_ galore.

we _learned_ from 1.9.17's nmbd [very bad] implementation
and deployment.

the result was another rapid rewrite and a very stable

broken code released into the wild encourages other programmers
to get involved and to see if they can fix it.

if it ain't broke, they won't fix it!

WHEN will you _get_ this, tridgell?

WHEN will you _get_ this, allison?

you can't do everything on your own.

there is about TWO TO THREE ***HUNDRED*** man-years of development
to catch up on.

you've _got_ to take advantage of every development and
man-management technique and strategy there is.

and _you_ lot out there - users - reading this.
yes, you, the ones bitching about microsoft all the

if you can't get off your arses and support samba development,
and instead are willing to pay thousands of dollars in
license fees, then don't go crying to the US government
or the European Commission about anti-competitive practices.

if you are a company that has more than four hundred employees,
the license fees that you pay in server / upgrade costs each
year could fund two to three samba developers to replace and
then maintain, at a quarter of the price, the server software
you keep paying for.


More information about the samba-ntdom mailing list