Linux-Advocacy Digest #636, Volume #33 Mon, 16 Apr 01 03:13:05 EDT
Contents:
Re: To Eric FunkenBush ("Erik Funkenbusch")
Re: Linux = CHOICE! (Ed Allen)
Re: Could Linux be used in this factory environment ? (Charles E. Hill)
Re: Something cool in gcc (Mark Robinson)
Re: Blame it all on Microsoft ("Bill Todd")
Re: NT is stagnant while Linux explodes (Norman D. Megill)
Re: Linux = CHOICE! ("Les Mikesell")
Re: So much for modules in Linux! ("Les Mikesell")
Re: So much for modules in Linux! ("Les Mikesell")
Re: Could Linux be used in this factory environment ? (Franek)
----------------------------------------------------------------------------
From: "Erik Funkenbusch" <[EMAIL PROTECTED]>
Subject: Re: To Eric FunkenBush
Date: Sun, 15 Apr 2001 23:33:08 -0500
"Donn Miller" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> Erik Funkenbusch wrote:
>
> > Nope. very few people (compared to the general C++ programming
population)
> > use templates at all. I'd say less than 10% in my experience. The vast
> > majority of people out there are working on maintenance of legacy code.
> > They're not going to introduce templates into that very often.
>
> I sort of felt the same way when namespaces were introduced. I thought
> they were uncessary, because you can simulate a namespace with a
> "struct".
However, a struct has other side-effects which you might not desire. For
instance, non-member functions can be contained in a namespace without
problems, however a function in a struct is a member function, and therefore
requires an instance of the object. You can get around this with static
member functions, but that's like a doctor prescribing amphetamines to
counter the sedative nature of another drug.
Namespaces are intended to reduce the need for the static keyword, since
static is so incredibly overloaded in C as it is. For instance, file-scoped
local variables in C are declared static, in C++ static is deprecated for
this feature and unnamed namespaces are used instead.
> Also, I think it's dangerous for C++ to be deviating too far
> from C, because most view C++ as a "C with classes".
That used to be the case when everyone was a C convert. Today, I think most
new C++ programmers learn C++ first. C++ is its own language and has its
own set of issues that don't relate to C.
> I know the C99
> spec introduced the concept of using a type qualifier inside loops, and
> other scopes like C++ does. This in contrast to the C89 spec, which
> absolutely requires you to declare all variables at the start of a
> block, which is very annoying and restrictive, IMO. I don't think gcc
> 2.95 supports C99 yet, though. (I think 2.95 is still stuck with the
> C89 spec, but don't know if they've carried over any C99 features yet.)
gcc seems to support a few C99 features. But I don't think they're
advertising them.
> In theory, I think C and C++ should be merged into C++. But since C++
> is so rapidly evolving, it's best to keep them separated.
Well, C++ isn't rapidly evolving anymore. It's frozen (except for minor
changes and defect repairs) for at least another 2 years, and probably for
another 5-7.
> In FreeBSD
> terminology, C is like FreeBSD-stable, and C++ is like FreeBSD-current.
> Some features like the type qualifier inside loops is like the
> occasional feature you see MFC's to -stable.
Not really. C and C++ are on different paths, although they occasionally
intersect. This is more like OpenBSD versus FreeBSD ;)
------------------------------
Reply-To: [EMAIL PROTECTED]
Subject: Re: Linux = CHOICE!
From: [EMAIL PROTECTED] (Ed Allen)
Date: Mon, 16 Apr 2001 05:01:04 GMT
In article <0dbC6.21050$[EMAIL PROTECTED]>,
Pete Goodwin <[EMAIL PROTECTED]> wrote:
>Ed Allen wrote:
>
>> The choice of multiple alternatives at several different levels like the
>> installer, the GUI admin tools, the users desktop, the group of apps in
>> a particular distribution, the mixture of GPL vs proprietary, the focus
>> of the patches and tunings applied to the standard kernel.
>
>These multiple choices could be Linux undoing.
>
No they represent strength because each will reach into crevices
that the "one thing for everybody" Windows world cannot reach.
Not everyone wants to do the same things in the same way as everyone
else.
The different combinations make the probability of a single breach
running through most of the machines in the world like a grass fire
on dry prairie much smaller.
>SuSE does it one way, Mandrake does it another. Can you imagine the
>development nightmare of installing on just two systems that have different
>configuration structure?
>
Unix vendors have been doing it for years.
I does mean that crapware will fail but most of us look forward to
that.
>Is that why we see a KDE 2.1 RPM for Mandrake, SuSE and RedHat? Why doesn't
>one work for all? Is this what choice brings us to?
>
It could be that they each have different administration tools.
Why is it a big deal ? Are you planning to try loading the SuSE one
on Mandrake as your next Linux failure ? Don't bother none of us
will sympathize.
>> All those choices at those and many other points are why we will always
>> have multiple distributions and will eventually see multiple offerings
>> from each distribution vendor.
>
>Standardisation works to eliminate the variations and make development life
>easier.
>
That was the point of my original rant.
Easy times for developers are about to end.
Only smart ones who are willing to try providing more than their
competitors will survive more than a few months.
Profiteering is no longer acceptable. A vendor who does not work
hard to please me and lots of others does not deserve our money.
>> We have seen only the tip of the diversity landscape so far. The 2.5
>> kernels will bring with them a language allow opening up whole new
>> territories of combinations which would have required advice from a
>> kernel developer before.
>
>Ack! More confusion, here we come!
>
>> What terrifies the WinTrolls is that knowledgeable choices by consumers
>> means that they cannot be forced into three to five piles like the
>> ignorant ones can.
>
>There's no terror here! With Windows it becomes rather easy to configure
>things, there in one place. Is that true of Linux? Let me see:
>
Unlike you many people like to decide for themselves what they want.
Most Linux advocates do not want to turn off our brains. We prefer
to do the thinking and leave Redmond to their own criminal
activities which the courts will soon put a stop to.
>Mandrake:
>/etc/rc.d/init.d/smb restart
>
>SuSE:
>/etc/init.d/dhclient restart
>
There is a symbolic link which makes SuSE respond to
/etc/rc.d/init.d as well.
What has a DHCP client got to do with Samba ?
>> That is why they are so adamant that MS is not really a monopoly,
>> because their complicity in fleecing the public might get noticed.
>
>Microsoft _is_ a monopoly. The only people who aren't seeing this the last
>time I looked was the court.
>
You are the first WinTroll to admit that in this forum that I
recall.
The court of Judge Jackson has convicted them and they are appealing
the remedy he proposed not the monopolization charges themselves.
The conviction will stand and we will be finding out about the
remedies before summer.
>> Yes the WinTrolls are strongly attempting to deny that their gravy train
>> is about to end and they will need to *work* for a living. Anything but
>> that !
>
>And how does supporting Linux help work for a living? How do you get paid?
>Selling distributions? Not enough! Writing software - most of it is free!
>
Most of the software programmers are paid to write is customization
of generic packages and in house stuff to support their business.
Those will still be needed with Open Source, not just Linux,
software.
Having customers who can call up and point to the bug or ask for
particular changes to be made to better fit their intended use will
require that salesmen not attempt to design the software but
companies like Cygnus have thrived in that environment.
Having sales people soap them up prior to their next screwing will
not survive so most of the "software" vendors of today are looking
forward to bankruptcy I think.
That is what scares the WinTrolls, that the victims are wise enough
not to submit to abuse a second time and that they must keep coming
up with new things because many of their most lucrative scams have
been cloned under the GPL.
The "software production" profiteering is wheezing its last gasp.
W2K has proven to be the biggest flop since MS Bob.
MS is planning to cripple it on purpose and force its installation
on every new computer under the label XP. That is attempting to use
their desktop monopoly to force customers to accept what they have
already rejected.
I hope we see the start of another antitrust trial the week it is
released.
--
Linux -- The Unix defragmentation tool.
------------------------------
From: Charles E. Hill <[EMAIL PROTECTED]>
Subject: Re: Could Linux be used in this factory environment ?
Crossposted-To: comp.os.linux.hardware,comp.os.linux.misc
Date: Mon, 16 Apr 2001 05:24:17 GMT
Franek wrote:
> Brent R wrote:
>> No way man, you're wrong ... everything that's computer-related must
>> look like a web-page... if it doesn't then you're just behind the times.
>> The factory worker's will be soooo much more productive if they feel
>> like they're searching the web,
> Oh yea, yeah, now I see I was wrong, of course, yeas, just think of it,
> the factory workers will be able to shop on-line while operating their
> favorite lathe! Kewl, dude. To check how their stocks are doing, transact
> their banking online, make a reservation at a favorite restaurant in
> Seattle, stare at some nekkid broads while them forklifts are running wild
> around. Read some unimaginative crap on Salon-dot-com. Productivity will
> soar that's for sure. That's the end of the shop floor as we know it.
>
Be serious! There are several ways to block this -- no default (0.0.0.0)
route on the terminals is the one I used when setting up web-clients in an
electronics (automotive elec) facility.
How about having the firewall deny outgoing packets from the shop floor's
subnet?
There are numerous, effective ways to restrict the floor systems to showing
only work-related items. Many manufacturing facilities require that a
terminal show only info related to the STEP PERFORMED AT THAT POINT, and
not just general work info.
HTML control with PDF documents is a great way to do centralized document
distribution in a document-controlled environment.
--
Charles E. Hill
Artek New Media
------------------------------
From: Mark Robinson <[EMAIL PROTECTED]>
Subject: Re: Something cool in gcc
Date: Mon, 16 Apr 2001 05:27:00 GMT
mlw wrote:
>
> Erik Funkenbusch wrote:
> > You think it's cool that a compiler violates the C++ standard without even a
> > warning? What if you're trying to write compliant code?
>
> The feature is cool because it can be recognized and used in a controlled way.
> It can make code much more efficient and robust. It is a very good feature of
> GCC if used correctly.
> >
> > In any event, I think this may be an optimization. C++ allows you to use
> > constant variables as array declarations, since you're not changing the
> > value of cb, it is probably optimizing it to a const value.
>
> "constant variables" are not variable, they are constant and thus do not do
> what you think they do. For instance:
>
> const int len = strlen(string)+1;
true, but
const int len = (strlen(string)+1);
should be a-ok since you aren't modifying a const.
Mark
------------------------------
From: "Bill Todd" <[EMAIL PROTECTED]>
Crossposted-To: comp.theory,comp.arch,comp.object
Subject: Re: Blame it all on Microsoft
Date: Mon, 16 Apr 2001 01:39:11 -0400
"Dennis Yelle" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> Charles Lyttle wrote:
> [...]
> > I agree with the lawful part, I'm not too sure about the ethical part.
> > Intel has pulled some pretty dirty, but legal, tricks.
>
> Where can I find out more about these tricks?
A good start seems to be the book Inside Intel by Tim Jackson.
- bill
>
> Dennis Yelle
> --
> I am a computer programmer and I am looking for a job.
> There is a link to my resume here:
> http://table.jps.net/~vert/
------------------------------
Crossposted-To: alt.destroy.microsoft,comp.os.ms-windows.advocacy
Subject: Re: NT is stagnant while Linux explodes
From: [EMAIL PROTECTED] (Norman D. Megill)
Date: Mon, 16 Apr 2001 05:48:03 GMT
In article <9bcuqk$qcq$[EMAIL PROTECTED]>, Mafoo <[EMAIL PROTECTED]> wrote:
>>>The number of bugs _found_ is less than in Windows... but is the number of
>>>bugs that _exist_ less? What are the comparitive number of Unix systems
>>>(server and client) in use compared to the number of Windows systems (server
>>>and client) in use? Doesn't Windows clock in with just a few more
>>>installations?
>
>considering most of the code is open source, and thus people can actually see
>the bugs, as opposed to trying to generate them... you make your mind up
Yes, to find a bug in source code requires a linear scan (line by line
reading) by motivated intelligent people, who look at the computer
program in the same way as they would verify a mathematical proof. This
is how for example OpenBSD acquired its now famous rock solid security
and stability.
Finding a bug without source code is a hopelessly exponential process.
As an extreme example, a bug might be activated if say a login name has
a certain precise sequence of 40 characters. The age of the universe
wouldn't be long enough to find it by exercising the program with test
inputs. (It is easy to purposely create "bugs" with this property, and
they are called "backdoors".)
An alarming number of "hidden" bugs are revealed just by peppering
software with random inputs, exposing conditions the programmers
overlooked or checked for incorrectly. See
http://www.cs.wisc.edu/~bart/fuzz/fuzz-nt.html
This study showed that essentially every major Microsoft product crashes
or hangs under random testing. Many other vendors' products don't fare
so well either.
(These current results are only for Windows NT/2000. In 1995 the Unix
results weren't so great either - it also has been traditionally
closed-source. No Linux results are given, but they do say "Our 1995
study found that applications based on open source had better
reliability than those of the commercial vendors.")
In the exponential process I described for discovering closed-source
bugs, these random-testing bugs might be just the tip of the iceberg.
These were just bugs exposed at the "surface" user-interface layer.
Imagine what might lurk in the core.
--Norm
------------------------------
From: "Les Mikesell" <[EMAIL PROTECTED]>
Subject: Re: Linux = CHOICE!
Date: Mon, 16 Apr 2001 06:04:01 GMT
"Pete Goodwin" <[EMAIL PROTECTED]> wrote in message
news:0dbC6.21050$[EMAIL PROTECTED]...
> There's no terror here! With Windows it becomes rather easy to configure
> things, there in one place. Is that true of Linux? Let me see:
>
Just because there is only one place in windows doesn't make it easy.
Try installing the 21 security patches for IIS 5.0 that each force
a reboot as part of the install.
Les Mikesell
[EMAIL PROTECTED]
------------------------------
From: "Les Mikesell" <[EMAIL PROTECTED]>
Subject: Re: So much for modules in Linux!
Date: Mon, 16 Apr 2001 06:04:01 GMT
"Pete Goodwin" <[EMAIL PROTECTED]> wrote in message
news:GzxB6.13900$[EMAIL PROTECTED]...
> Roy Culley wrote:
>
> > Of course he can't if you won't look for help in the appropriate place.
> > Did you ask for help on the SuSE english mailing list? Of course not as
> > then you would have nothing to complain about. All the problems you have
> > are trivial. One problem you have is that you somehow manage to screw up
> > any Linux installation you do. Then you just complain in c.o.l.a. Grow
up.
> > You would learn a lot more if you actively sought help in newsgroups and
> > mailing lists that are relevant to your problems.
>
> Lest we forget - I am not here asking for help. I pointed that out in my
> first post. You did read that, didn't you?
>
> My point is Linux doesn't make it very easy to configure. It gets harder
> when a distro manages to shoot itself in the foot. That's what I'm here to
> complain about.
Has Suse done away with the ability to control the steps taken at
startup by editing a set of scripts and linking them into the correct
rc.n directory (where n is the runlevel) using names that alphabetically
sort into the right execution order?
> If the problems were trivial, seems to me I would have fixed them by now.
Editing a text file is trivial. Making a link is trivial. If you can make
something work when executing it by hand, making it work by the above
two steps is trivial.
Les Mikesell
[EMAIL PROTECTED]
------------------------------
From: "Les Mikesell" <[EMAIL PROTECTED]>
Subject: Re: So much for modules in Linux!
Date: Mon, 16 Apr 2001 06:04:01 GMT
"Pete Goodwin" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> In article <[EMAIL PROTECTED]>,
> [EMAIL PROTECTED] says...
>
> > Did you notice that the dhclient isn't named S05* to start
> > with? The intent is to move the dhclient after the network
> > startup, which I thought was exectly what you're talking
> > about.
>
> I moved dhcp to boot.local which I thought would run after everything
> else. It did but it still ran before the network modules were loaded.
If you need a module loaded, why don't you just add
insmod module_name
or
modprobe module_name
at the point where you want it? Or rename the script so it will
execute later if that accomplishes the same thing? This is
approximately as difficult as knowing when to right-mouse instead
of double-click in a GUI.
Les Mikesell
[EMAIL PROTECTED]
------------------------------
From: Franek <[EMAIL PROTECTED]>
Subject: Re: Could Linux be used in this factory environment ?
Date: Mon, 16 Apr 2001 06:22:36 GMT
"Charles E. Hill" wrote:
> Franek wrote:
> > Brent R wrote:
> >> No way man, you're wrong ... everything that's computer-related must
> >> look like a web-page... if it doesn't then you're just behind the times.
> >> The factory worker's will be soooo much more productive if they feel
> >> like they're searching the web,
> > Oh yea, yeah, now I see I was wrong, of course, yeas, just think of it,
> > the factory workers will be able to shop on-line while operating their
> > favorite lathe! Kewl, dude. To check how their stocks are doing, transact
> > their banking online, make a reservation at a favorite restaurant in
> > Seattle, stare at some nekkid broads while them forklifts are running wild
> > around. Read some unimaginative crap on Salon-dot-com. Productivity will
> > soar that's for sure. That's the end of the shop floor as we know it.
> >
>
> Be serious! There are several ways to block this -- no default (0.0.0.0)
> route on the terminals is the one I used when setting up web-clients in an
> electronics (automotive elec) facility.
>
> How about having the firewall deny outgoing packets from the shop floor's
> subnet?
>
> There are numerous, effective ways to restrict the floor systems to showing
> only work-related items.
Oh you must be one of them fascist right-wing republican rich bastards who want to
limit
human workers to only work-related items! What about diversity, multi-kulturalizm,
don't
the workers have their needs, don't they need to check out their stocks on line? Play
with
the mouse? Verify latest news in their kommunity?
> Many manufacturing facilities require that a
> terminal show only info related to the STEP PERFORMED AT THAT POINT, and
> not just general work info.
>
> HTML control with PDF documents is a great way to do centralized document
> distribution in a document-controlled environment.
>
> --
> Charles E. Hill
> Artek New Media
------------------------------
** 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 by posting to comp.os.linux.advocacy.
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
******************************
