Linux-Advocacy Digest #227, Volume #26 Sun, 23 Apr 00 04:13:09 EDT
Contents:
Re: DCOM versus CORBA, some history (R.E.Ballard ( Rex Ballard ))
Re: KDE is better than Gnome (pete@hipe)
Re: Grasping perspective... (was Re: Forget buying drestin UNIX...) (Full Name)
----------------------------------------------------------------------------
From: R.E.Ballard ( Rex Ballard ) <[EMAIL PROTECTED]>
Subject: Re: DCOM versus CORBA, some history
Date: Sun, 23 Apr 2000 07:18:18 GMT
In article <PCSK4.2636$[EMAIL PROTECTED]>,
"Erik Funkenbusch" <[EMAIL PROTECTED]> wrote:
> Robert Morelli <[EMAIL PROTECTED]> wrote in message
> news:[EMAIL PROTECTED]...
> > This post is partly a response to a current
> > thread about Microsoft's putative
> > advantage in the enterprise. The issue centers
> > around Microsoft's CORBA counterpart, DCOM,
> > and their proposed supporting XML technology as a
> > stand in for lack of Java support.
Since both the initial and the rebuttal contain some
issues, I'm responding to both, and also some subsequent
follow-ups not included here.
> > The first misconception is that Microsoft's
> > DCOM has no counterpart outside Windows.
> > That is obviously false.
No problem here. DCOM and CORBA are both implementations
of object oriented messaging between client object stubs and
server objects. Essentially this is a series of remote procedure
calls structured to support the object oriented paradigm.
> > The second misconception is that DCOM is
> > an extension of CORBA. This is also false.
> > DCOM is not an example of "embrace and extend";
> > it's simply an example of incompatibility with open standards.
Actually, both technologies have very similar heritage.
UNIX/OPEN Microsoft
Sunview Windows C
C++ C++
Widgets MFC
X Toolkits OLE
SOM, DOM COM
CORBA DCOM
Java, JavaScript VBScript
GNOME,KDE ActiveX
Also
SGML
HTML
Lynx
Viola Cello (ported from Viola by Cornel)
Mosaic Mosaic (ported from UNIX version by Marc Andreeson)
Netscape Netscape (designed for both by Marc Andreeson
Arena Mosaic->Internet Explorer
HTML
XML XML+ (voice,inline DCOM objects).
> First, DCOM and CORBA are similar,
> but they are targeted at different
> audiences. CORBA is a remoting
> architecture which, until recently, had no
> component model.
Actually, CORBA was based on the component model
that had evolved through open standards. This
included Widgets, Toolkits, and Components used
in X11, Open Desktop, and CDE. It didn't get
much press. Probably because Microsoft was
spending $4 billion to promote Microsoft products
and anyone who said anything nice about UNIX
lost their share.
> COM was a component model, that grew a remoting
> architecture. COM was originally designed for
> network transparancy (though
> not implemented at first) back in 1987,
> before CORBA existed.
Actually, the Component Object Model was first
announced as a concept in 1990. Of course this
was a counter to the release of the first X11
products, specifically Athena Widgets, which
had was now in it's third release (X11/R3).
It's quite possible that Microsoft was aware
of X11, and was trying to keep pace, but
even as late as 1992, Microsoft windows
3.0 was still programmed using a procedural
model.
> DCOM is simply the network remoted
> version of COM, the technology that MS
> has been very successful with for the
> last decade. If you know COM, it
> doesn't take a lot to learn DCOM.
If you know C++, Java, or Python,
it doesn't take a lot to learn CORBA.
It isn't that hard to learn CORBA if
you knew C or any of the other supported
languages. The key to understanding a
component model is understanding object
oriented paradigms.
> Additionally, you can replace COM
> components with DCOM components in
> the vast majority of cases and the client
> won't even have to know or be recompiled.
Yes, but if you try and use DCOM on the same
processor, you pay a huge performance penalty.
This creates a problem when you need to combine
the same type of object with both local and remote
instances.
CORBA was originally designed for UNIX, which was
optimised for process-to-process (out of process)
communication including connection between processes
on the same processor.
> That is why DCOM exists, not because it's competing with CORBA.
Actually, Microsoft realized that many companies were using UNIX
because it was easier to combine components that didn't need
to be recompiled. Even with DCOM, you still have to decide which
components need to be bundled into which DLLs.
Using DCOM objects is pretty easy, but creating new objects
is nearly impossible.
> > There's some interesting history behind this.
> > When MS was first promoting DCOM back in 1995,
> > there was an obvious question why they were
> > flouting the open CORBA standard,
> > and deliberately introducing incompatibilities and hassles
> > for their customers.
CORBA had actually been in the cooker since 1989, and
was an effort to merge RPC, DCE, and SOM under a common
standard. Also, there was a desire to create an alternative
to the ICCCM X11 interface.
> Microsoft had been promising DCOM since
> COM was officially launched in a
> product in 1992. CORBA had very
> little to do with it, since it was an
> extension of the existing COM architecture,
> not a perversion of CORBA.
Like so many of Microsoft's promises, this one
was very late and didn't even come close
to delivering what was expected.
Microsoft also promised Windows NT in 1991,
Microsoft introduced MFCs in early 1993
(possibly late 1992) as part of the effort
to prepare for Windows NT. The OLE model
came out a bit later - as a way to minimize
the calling overhead. The COM interface
made it possible to hide the server and
the nature of the server, but true interprocess
communication was still difficult and not terribly
practical. Essentially, Windows 3.1 applications
still relied on shared memory, and NT had a problem
with that. Eventually, Winsock and OLE developed
methods to access kernel services that allowed them
to share memory.
DCOM made interprocess communication much more
practical. With DCOM, developers could create
simple components and connect them together.
Unfortunately, creating a full client and full DCOM
server from scratch was still non-trivial. It
was very easy to use a VBGraph object to create
a pretty chart, but not so easy to create a new
object from scratch. The developer would have
to support In Place Activation and Visual Editing,
Embedding, Linking, Drag and Drop, Compound files,
monikers, and automation. Even a simple structure
passing had to support all of these functions.
One of the key features of DCOM is that you could
connect a trivial server implementation that
pretty much only supported "Get the data" and
connect it to a separate graphical user interface.
Since the most important custom components for enterprise
applications were the business rules, DCOM became essential
to server implementatinos on NT.
> It didn't introduce incompatibilities
> or hassles for most customers, since it
> was designed to help them leverege their
> existing COM code, making it easier
> to do distributed processing if they
> were already heavily invested in COM.
The biggest problem with COM and DCOM was
that it was still designed for the Client-Server
paradigm, which was already becoming obsolete.
Eventually, Microsoft adopted many concepts
from CORBA to provide simple middle-tier
functions.
The other problem with DCOM was that you
had to redesign servers, often very complex
servers to support multiple connections from
numerous computers. This turned out to be
a substantial effort, and often blew the
schedules and budgets horribly.
What made matters worse was that DCOM
wasn't supported under Windows 95, and
fully distributed DCOM objects required
the use of NT Workstations.
> > At a convention, an MS representative
> > started his presentation by reading the
> > definition of "standard" from a dictionary.
> > He then slammed the dictionary shut and
> > announced that Webster's definition of
> > standard "was dead."
This was actually a very well publicised event,
and had a very chilling effect on Microsoft's
relationship with IBM, Hewlett Packard, DEC,
and Dell (who was selling UNIX on many of their
best machines), not to mention Sun. Up until that
moment, the OSF had been bending over backward
to adopt standards that were compatible with
Microsoft. What was worse is that the Applications
vendors were also alienated since this meant
that they could not depend on the open standards
to drive Microsoft.
> > The "new definition," he explained, was
> > that "de facto is de standard," and
> > "de facto is Windows."
> > I can't think of any single incident
> > that better sums up Microsoft's
> > combination of arrogance, contempt,
> > and brutishness. If you accept this
> > kind of disrespect and go on supporting
> > MS and advocating for them, I think you
> > need to go to a doctor and get your balls examined.
> Presentations are marketing pieces.
> They're designed to catch your
> attention, and it looks like this one
> succeeded. Taking any such marketing
> propoganda at face value, no matter who
> the presenter is, is a rather
> foolish thing to do.
The only problem was that Microsoft reiterated it
numerous times. Bill Gates reiterated it, Steve
Ballmer reiterated it, and only Paul Allen was
concerned that Microsoft may have gone too far.
It was about this time that Paul Allen began
selling large amounts of Microsoft stock to
buy interests in open standards companies such
as Internet companies. Paul Allen was a key
investor in AOL when it was in it's infancy.
> > The term 'hubris' (excessive arrogance) is appropriate here.
Actually, this wasn't really the "hubris". What pushed it from
simple arrogance to "hubris" was when Microsoft demanded that
OEMs wishing to sell PCs equipped with Microsoft Windows 3.1
would have to ALSO include Microsoft Word, Microsoft Excel,
and Microsoft PowerPoint - effectively driving market leaders
WordPerfect, Lotus 1,2,3, Harvard Graphics, and Corel Draw
completely out of the Windows market. It literally decimated
the price of the stock. WordPerfect was eventually purchased
by Novell who soon purchased UNIX from AT&T. Lotus was eventually
purchased by IBM who was hoping to establish OS/2 as the new
PC standard.
> Indeed it is. Since you are arrogantly assuming that one persons
> presentation at a convention is the companies official policy.
Again, nearly every key press release from 1991 to 1992 echoed
nearly the same sentiment. Microsoft had just been investigated
by the FTC and had been able to lobby the review board such that
they won by one vote. Microsoft was then remarkably confident.
It was almost as if they felt that they had the FTC in their
back pockets. In fact, 3 subsequent actions and attempts to
limit Microsoft's ability to bully OEMs and Software developers
were rebuffed by the Clinton administration. One attempt to
settle was actually rejected by the judge. This latest case
was actually an attempt to prevent the attourney's general of
20 states from filing independent antitrust and criminal charges
against Microsoft. Microsoft's attempt to manipulate the
negotiations and the arbitration of an appellate court judge
eventually resulted in the first criminal ruling against Microsoft.
At this point, we have Microsoft legally defined as a monopoly,
and we have numerous executives potentially facing not only
civil charges, but possibly criminal charges.
> > The ancient Greeks believed that hubris invokes nemesis,
> > a form of divine retribution. In fact, it was about the
> > time of the "de facto is de standard"
> > statement.
Actually within about 2 weeks. I believe it was late 1991
or early 1992 that Microsoft "went too far". Lotus 1-2-3
developed a version for UNIX that had multiple windows, and
could be dynamically updated - including by update feeds.
Novell purchased the rights no AT&T UNIX and threatened to
sell it as a retail product for $200 - for desktop use.
SCO published Open Desktop - one of the first fully integrated
office suites.
Borland made their C and C++ compilers source code compatible with
ANSII and UNIX.
The UNIX administrators who were administering the Internet began
looking to privatise the Internet.
MCI agreed to carry the NSF traffic for a fraction of the actual
cost, on the condition that they could put business traffic on
the same TCP/IP links to subsidize the traffic.
Linus Torvalds released Linux to the Free Software Foundation
and then accepted hundreds of contributions from former GNU and
BSD developers. Even employees of SUN were contributing.
Viola was released to make accessing UNIX systems simple and safe.
Cello (a Windows port of Viola) was developed to give Windows users
access to UNIX systems without the issues and cost of X11.
All of this happened from November of 1991 to July of 1992. These
were the seeds of the effort to break Microsoft's control of
information standards.
By January of 1993, a number of San Francisco newspapers were
publishing their content on the Internet. Byte Magazine was
singing the praises of Linux and Novell UNIX. And Dow Jones
accepted the challenge of being the first nationally branded
publisher on the internet.
IBM was still trying to clean up OS/2 2.0, and was racing to
create what would later be known as WARP.
By June of 1993, BBS operaters were converting from Windows
and MS-DOS to Linux, and the NSF was publishing web content
in HTML.
By November of 1993, Dow Jones was offering a subset of it's
news service (DowVision) over the internet. By January 1994,
40 publishers were trying to figure out the best way to
put their content on the internet.
By December of 1993 Trumpet Winsock was being site licensed by
several hundred corporations. In addition, Mosaic and was
being downloaded at a rate of over 200,000 copies per week.
In January of 1992, there were about 1 million Internet users.
By January of 1994, there were over 32 million Internet users.
> > that the internet started to explode,
By March of 1994, Steve Outing's mailing lists had grown to
over 2000 publishing companies, portal sites, and webmasters.
There were over 1000 web sites - most of them based on Linux
running on an old PC about to be discarded or sold to junk dealers.
Over 4000 BBS operators had switched from Wildcat and Fido to
Linux - offering PINE and PPP access including POP3 and HTTP
access. In many cases, the POP was running 3-5 dial-up 9600 baud
lines connected to an X.25 or dedicated 56kb line. Several hundred
of these POPs federated into national ISP services or were
simply managed as local independent providers.
> > Java appeared, and the Linux threat
> > started to really rear its head.
Actually, Java was a relative late-comer. By the time Java
had hit the scene, many companies had already chosen to use
Web Browsers to connect to UNIX servers. Windows suffered from
memory management problems. The main driving force behind Java
was that Netscape plug-ins had reached the point where they
were making web browsers as unreliable as dedicated "thick client"
applications.
Microsoft had tricked Novell into withdrawing it's "Desktop UNIX"
offering. They had promised not to use NT as a server if Novell
promised not to use UNIX on the desktop. This was done without
Ray Noorda's knowledge, and when he discovered what his directors
had done - in secret, he left the company and took the desktop
team to form Caldera.
Sun wanted to break Microsoft's exclusive lock on the Desktop.
They created a package that could run on both UNIX and Windows.
Unfortunately, Sun decided to ignore Linux, but many of those
interested in PC implementations of UNIX had already begun
to focus on Linux. For a period, Linus Torvalds dismissed the
need for Java and JavaSoft dismissed the role of Linux. The
Blackstone Group was typically 6-9 months behind Sun, and Sun
refused to acknowledge Kaffe and Guavac as legitimate implementations.
Kaffe and Guavac were GNU GPL implementations and Sun wanted to
make sure that they could retain control of Java.
> > In 2000, the very issue of CORBA compliance is
> > coming back to haunt MS.
Actually, not just CORBA. Microsoft is suddenly beginning to
find itself trapped between a rock and a hard place. The
DOJ trial made public some of the tactics used by Microsoft
to retain control of the market. The mere exposure and the
possibility of gaining barganing leverage by offering Linux
enabled Personal Computers caused the OEMs to push back.
Microsoft enjoyed a modest growth of the NT server market,
but Linux and UNIX were growing even faster. Eventually,
NT marketshare had dropped to 20% - off from it's high of
27% in October of 1999. Microsoft had less than 3 million
servers while UNIX and Linux had nearly 10 million.
> > In fact, Microsoft's position is
> > now considerably weakened.
Microsoft is walking a very thin line. Too much pressure on
the OEMs could result in contempt of court and harsh punitive
action, too little and the OEMS would openly revolt.
Even today Microsoft is finding it much harder to get the
exclusive contracts that give Microsoft control of the
system. Worse, the OEMs are trying harder and harder to
make their systems look less like Windows and more like
Linux.
Furthermore, you have companies like Cygnus offering source code
compatibility between Linux and Windows NT/2000 (but with slower
performance).
> > First, their proposed XML technology is fairly crude,
Even more important, the other members of the W3C are
pushing for open standards. They would rather use standards
based on GPL and sell "glitz" and support than let Microsoft
force them into accepting nondisclosure agreements.
Meanwhile, the IEEE formally released the "fire wire" specification,
which means that USB vendors are now stuck between a rock and a hard
place. If they maintain their nondisclosure agreements, Fire-wire
could result in the proliferation of fire-wire based peripherals
at 400 megabit transfer speeds. If they break the code of silence,
they risk retaliation from Microsoft. Some vendors are starting
to implement "binary only" servers that can be connected to existing
Linux USB drivers. Linux provides the equivalent of the IP layer,
but the vendors have to provide the complex scanner, printer, disk
drive, and user protocol. With fire-wire, the commands are essentially
the same as SCSI, but using 32bit frames delivered on the 400 megabit
bus.
> > it doesn't even lock people into Microsoft technology,
> > and what it accomplishes can be easily
> > duplicated by other technologies.
More importantly, you can use SAX and DOM to create a distributed
object model across either Sockets or MQSeries. Microsoft is
trying to "Embrace and Extend" their MSMQ implementation, but since
MQ Clients are free, and servers are very reasonably priced, it's
pretty easy to set up IBM's implementation on nearly any version
of Windowws.
> Uhh... you're criticizing them for
> doing something that's cross platform
> compatible? More arrogance?
Actually, Microsoft is trying to add monoplatform extensions
such as XML-Voice, XML-Video, and XML-Graphics, all of which
are effectively equivalent to using ActiveX controls, but without
the option of refusing the document.
Microsoft wanted their proprietary standards adopted and the
W3C membership almost unanimously refused. They are now
adopting OpenGL, CGM, MPEG, MP3, and MPEG-II, but without
the DVD-CSS encryption. They are instead encouraging the
use of open encryption standards bit with 128 bit RC5, PGP,
and/or DES encryption. Since this would be the same as
todays SSL, it shouldn't be a problem.
> > In fact, it is just an attempt to salvage a deteriorating
> > situation in which they are in danger of gradually
> > losing touch with an emerging, rich set
> > of e-commerce and connectivity standards.
Microsoft is still trying to bully the OEMS, and the Banks.
Unfortunately, they tried to push ActiveX as their standard
for secure transactions. Microsoft has had so many security
leaks that some companies won't even accept credit cards
from Windows based PCs.
> > Second, Microsoft is no longer in a position
> > to dictate standards. Their proposal was
> > universally rejected out of hand by the rest
> > of the industry.
It's only a matter of time before the combination of U.S. pressure,
International EU pressure, and industry pressure essentially cause
Microsoft to implode. As it is, Microsoft couldn't even give away
Windows 2000. They had presold 5 million licenses, they dropped
the upgrade price to $179, and they even included Windows 95 and
Windows 98 to the upgrade eligibility.
> Considering that Microsoft has been
> one of the heaviest hitters behind the
> XML standards process, I find this a
> rather ludicrous argument.
But Microsoft has been pushing XML to prevent the proliferation
of SGML, which Linux and other versions of UNIX can easily support.
Microsoft has also been pushing in-line binaries and executables.
This is much more obnoxious than links. With links, you can
follow the text links and skip the graphics. With in-line,
you have to know enough to be able to skip what you don't want
to process, and you must have modules to parse every component.
--
Rex Ballard - Open Source Advocate, Internet
I/T Architect, MIS Director
http://www.open4success.com
Linux - 60 million satisfied users worldwide
and growing at over 1%/week!
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
From: pete@hipe
Crossposted-To: comp.windows.x.kde,tw.bbs.comp.linux
Subject: Re: KDE is better than Gnome
Date: 22 Apr 2000 23:38:56 -0700
Compare software to construction or mechanical engineering or
electrical engineering.
In building a bridge for example, there are fixed and standard
ways to connect pieces togother. Even each bolt and hole has
standards. Yes, there are choices, but each choice is known and fixed.
Each cable type is standard, of fixed weights and thickness and other
measurments. the spacing for windows has standard. the key hole
for the door is of standard sizes. Everything
and anything has a standard. a contractor can order some piece
knowing it will fit with some other becuase the standard says so.
The same applies to electronices. There are many choices, but the choices
are fixed and standard.
But in software, each person creates their own software, and leave it
up to the others to figure how to plug into it, and many software
do not work with other software, and many environments are
incompatible with others.
When it comes to science of engineering, software is still in the dark ages.
That is only to be expected. software by its very nature requires
little science to create, it is a new field, and any one and any body
can call them a software engineer after 2 days of VB programming,
while other real engineering fields have been around for much much
longer time.
------------------------------
From: [EMAIL PROTECTED] (Full Name)
Crossposted-To: comp.os.ms-windows.nt.advocacy
Subject: Re: Grasping perspective... (was Re: Forget buying drestin UNIX...)
Date: Sun, 23 Apr 2000 07:36:17 GMT
On Sat, 22 Apr 2000 12:57:20 -0400, mlw <[EMAIL PROTECTED]> wrote:
>
>Actually I have to disgree, but not for the reasons you may want to
>hear. Windows is unreliable. Avoiding Microsoft to avoid unreliable
>software is a reasonable position.
>
We have approximately 50 NT Workstations running Office 97 on a
variety of clone hardware. All these machines are left on permanently
(for backup purposes) and the users do not in general re-boot. Many
of these machines have lengthy uptimes.
In the last 6 months I've observed a total of three "blue screens of
death" across the 50 workstations. These were fixed by pressing the
reset button.
We also have one PDC and three BDC's. Of these machines the only one
I know of that has experienced a failure of any kind was the BDC on my
desk which I use as a workstation. It unexpectedly re-booted while
downloading a runaway WWW page using Netscape. The machine is a P133
with 32 MB ram and struggles to do many things.
We also have two Sun Ultra 10's, one Sparc 10 (all running Solaris
2.7), an Intel based Linux box and an old HP Unix box. After a recent
power disruption all 50 NT systems (including the servers) came up
without any problems. One of the Ultra 10's failed to restart and
required a file system check. This resulted in down time of our WWW
server facilities.
The Sparc 10 was recently re-booted remotely using "sync;sync;reboot
-l". It failed to start requiring a file system check. This resulted
in a mad dash to work in the car and complaints from our users.
Our Linux box currently has Mandrake 7.x installed. It is only about
a month old. In that time Linux has been installed four times. The
first time it was installed with Red Hat 6.x. This was not considered
secure enough and the switch was made to Mandrake. Unfortunately I
selected high security and found it rather difficult to get the ftpd
running. So Mandrake was reinstalled a second time with low security.
Our Unix expert then took on the job of getting NFS running on the
Linux box to allow backup. Unfortunately the Linux flavour of NFS
refused to talk to our Solaris box with the tape drive. After almost
two weeks of full time tinkering (and another install of Mandrake) we
now have NFS mounts between Linux and Solaris 2.7.
At the end of last year we upgraded the Sparc 10 from Solaris 2.5 to
2.7. The three of us started at around 10 a.m. and finished most of
the work by 7pm. This was mainly due to the repartitioning of the two
hard disks and problems getting the system to boot from the correct
device. The entire down time was two days. We run Oracle 7 on the
system and have since discovered the orainst program core dumps if you
try to run it on a Solaris 2.7 system (known bug). We hope we will
never have to do a re-install of Oracle.
I recently wiped NT 4.0 from our BDC and installed NT 4.0 Terminal
Sever edition. The entire operation (including setting up shared
printers and directories) took less than two hours. This was done
solely by myself on a lazy Saturday afternoon. When the users
returned on Monday no one noticed. The advantage over NT of being
able to remotely manage our Unix boxes has vanished.
We have had a shortage of disk space on one of our Ultra's and on our
NT box. Purchasing a 28GB IDE HDD for a tad under $300 solved the
disk space problems on the NT box. The price of a 9GB SCSI disk for
the Ultra is around $1500. We would need a Sun technician to come out
and install it. I slotted the IDE in myself. We cannot afford the
Sun 9GB. We now dump tar files from our Unix boxes to our NT machines
using Samba when we require temporary backup space.
We have a 70 GB tape drive for backup purposes on our NT box. This is
used to backup our NT Workstations. We have experienced no problems
with this tape drive.
Just before Easter we discovered our 24 GB Sun tape drive was filling
with less than 12 GB transferred. A Sun technician came out and
replaced the tape drive. Fortunately we have a (rather expensive)
maintenance contract with Sun. We will test the replacement drive
after Easter.
We don't see the need for a maintenance contract for our NT boxes as
our staff has been pulling apart and assembling PC's for years.
Replacement hardware is also very inexpensive and readily available.
I do more or less all of the maintenance of our NT Server. The only
time I need touch the thing is when a new user arrives or we need some
new shared resources.
Our Unix boxes consume pretty much a full time person. A great deal
of time is spent assisting users creating scripts and giving them
access to installed software. Many of them simply refuse to learn any
Unix and expect us to construct their scripts for them. We cringe
when a user comes in and wants to start running an X application. We
push users toward NT equivalents wherever possible. Our Unix users
consume at least three times as much support time as our NT only
users. The users themselves clearly prefer to use NT equivalent
software whenever possible.
One thing I like comparing between our Unix boxes and NT is sharing a
file system that is going to hold our Access databases that are used
by many different people. Access creates a '.ldb' file that must be
read/write by all users who simultaneously access the database.
>From the NT point of view this takes about two minutes. The most time
consuming aspect is ensuring all of the users are placed into the
appropriate group using the user manager. The file system is then
shared with change permision to the new group.
>From the Unix/Samba pointing of view you must concern yourself with
the following:
The group the users are in (edit /etc/group).
The samba umask (smb.conf).
The group ownership of the directory to be shared.
The group s bit to be set on the shared directory.
The user, group and other permission on the files themselves.
All this is necessary to ensure the locking file is read/write to all
users regardless of who opened the database first and actually owns
the file locking file.
Once you have that all sorted out you will still find that the
administrators (who have permission to change the forms, queries, etc
as well as the data) will have to telnet to the Unix box and chown a
file before they can copy over it. This is due to the actual owner of
the file being the last administrator to change it. This is if course
a non-issue with an NT server.
In the first four months after I started working I found on four
mornings I was denied access to a Samba share I had access to the
previous day. This was due to one of the administrators inadvertently
changing one of the settings above. This happens quite often with
Samba resulting in irate users. One must be continually aware of file
ownership, group ownership, s bits and general file permission.
We have not had a single problem with access to out NT shares
involving established users.
------------------------------
** FOR YOUR REFERENCE **
The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:
Internet: [EMAIL PROTECTED]
You can send mail to the entire list (and comp.os.linux.advocacy) via:
Internet: [EMAIL PROTECTED]
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Advocacy Digest
******************************
