Re: what to do after "Touretzkey's book"?

Ulrich Hobelmann <u.hobelmann@xxxxxx> writes:

Tim X wrote:
Ulrich Hobelmann <u.hobelmann@xxxxxx> writes:

drewc wrote:
And of course, RedHat, IBM, Google and others are making millions
(partially) from Free Software, and in turn paying programmers to
contribute.
I don't count those. IBM has some Linux hackers, but Linux isn't
strictly necessary anyway (there's BSD and proprietary Unices).
What do you mean? This doesn't (to me) make sense - I don't mean the
statement is silly or worng, it just doesn't make sense i.e. I can't
see what your point is and how you discount IBM's open source contributions.

No, it's definitely good that IBM has some employed hackers, but it
doesn't make a huge difference for most people. And most open-source
hackers aren't really employed I think. I wouldn't know how to get a
job in that area, even if I were to spend the next two years hacking
some cool desktop GUI, or some kernel device driver...

While there is a lot of open source which is done by people in their
own time and for which they are not paid, there is alot which is
generated by people who are paid to do the hacking. For example, if
you have a look at a lot of the specialised drivers for things like
network cards and even som eof the inner kernal features you will find
a lot of it was written by people who were getting paid to write the
source. Many applications which are open source have had substantial
parts contributed by people who were paid to develop the
contributions.

As pointed out by others, the majority of software writing is not done
by people developing commercial applications for sale. In fact, most
programmers are employed "in-house" and the software they develop is
rarely packaged for sale as a commercial product. In these
environments, programmers are regularly required to enhance existing
systems or fix problems with these systems. A growing number of both
public and private companies are taking advantage of open source
aplications because they get the source code and often there is a very
active community backing up the development of the software. In these
situations, when a company improves or fixes something in the
software, they generally contribute it back to the open source
community.

My own experience has been along these lines. I've been programming
for nearly 20 years and I'd estimate 80% of the code I've written has
been released under either a GPl or BSD/MIT type license. Only a small
part of the code I contributed to has been closed source and even a
smaller part has been sold commercially. Most of my employers have
been quite happy to release code which they have paid the development
costs on because this isn't how they make their money. The code was
developed to support their business and as long as it does what they
want and they have the source, they don't really care who else gets
some benefit from it.



RedHat produced IMHO the crappiest piece of software ever (yes, worse
than Windows 98), even though it ended up as the most-used Linux
distro; tastes differ.
Can't follow you here either? Which bit of Red Hat's very meagre
software contributions are you referring to? Its not loke they are a
software development company. To say that Red hat is worse than windows 98 is a
pretty radical call - especially as Red Hat's version of Linux is only
marginally different to any other version and most of those diffeences
are pretty well cosmetic. In what specific ways was any distribution

That's what I thought when I bought BadRat 6.2. I had spent a few
weeks or months with my first Linux (Debian 2.1, which was counter
most people's claims very easy to get running), and spent maybe 40€
on BadRat. Well, it didn't bother to boot the graphical installer on
my laptop (i.e. my computer back then), and when I booted into the
"expert installer (text mode)" it simply hang. When I installed it on
another computer later, the system sucked - it was DOG slow, and I
couldn't see any advantage over Debian. This was simply the worst
system ever, and while RedHat didn't write it themselves, they put it
all together in a big, slow, expensive distro, and they deserve lots
of credit for that.


I'm sorry, but to me, this just doesn't add up. The most critical part
to any OS is the kernel. Now, if your comparing apples to apples,
debian to Red Hat, etc, I would assume you are comparing systems with
the same kernel version. Now, there are very minor differences between
distro kernels of the same version - for example, Debian removes some
driver modules which Red Hat includes by default (this is because
Debian has a much stricter policy on licenses than Red hat and
therefore remove some drivers with licenses which don't comply with
their policy). As only the modules which are needed are loaded, I
don't see how you would see a major difference in performance between
the two.

If you believe it was because Red hat had a lot more daemons running,
then if you don't need them, why have them running. If you are
comparing a minimal Debian distribution with only a few daemons
running to a full Red hat distribution with every possible daemon
running you could have, well thats not a very fair comparison and a
rather unjustified condemnation of Red Hat isn't it?

I have lost count of the number of Linux distributions I've installed
over the past 14 years. However, my experience without fail has been
that performance under Linux compared to Windows has always been
better regardless of whether it was Red hat, Debian, SuSe, Slackware
or Mandrake. The most striking improvement I ever saw was with a sparc
station when I replaced solaris with Red Hat's version of Linux for
the sparc - that difference was amazing. A sparc server which I
thought was really slow and had blamed on the hardware suddenly became
a rock solid reliable and fast server with Linux compared to solaris.

In all the years I've run Linux, there has only been three occasions I
had problems getting the software to install. My first Linux
installation was a pain because it consisted of around 25 floppy disks
and I had quite a few disks with faulty sectors. I had problems with a
mandrake install, but that was because I was using a cheap clone video
card which claimed to be 100% compatible with a well known brand, but
of course wasn't and I had problems with my first Debian install due
to the fact the network card I had was brand new and not supported by
the stable distribution kernel (but was supported with the testing
kernel). I'd also have to admint the sparc station install was tricky,
but that was more due to my lack of low level knowledge of sparc
hardware more than anything else.

This isn't to say Linux doesn't have problems with some hardware.
There is no doubt that Linux does not support the range of hardware
Windows supports - but thats no shock. Lots of hardware vendors have a
very proprietary view and in the past more than now, were unprepared
to provide the necessary information to assist device driver
development for their hardware. Things have really improved over the
last 5 years or so, but anyone who doesn't first verify their hardware
is supported by the distribution they plan to install is asking for
trouble.


I was totally angry when two weeks later Debian 2.2 was released, and
I could buy it for much less - maybe €25. Debian 2.2 rocked; the
package manager (dselect) was much slower than in Debian 2.1, but it
had all the new packages and worked just as perfectly.

Later I tried some Mandrake (Redhat derivative), and it couldn't
connect to the 'net. I found out that it ran WVDial, but unlike my
Debian where I did the same, it didn't work.

You cannot assume all Linux based distributions are the same - this is
how they get their market difrerentiation. They all follow different
philosophies and I would have to say I find it a great shame that many
of the distributions are trying to win the battle by attempting to
mimic the point and click installation of windows.

Is the fact that wvdial was configured differently under madrake than
debian and didn't work the same as on debian mean that madrake was not
good or that you were caught out by making assumptions that turned out
not to be justified. If you had never run Debian and started with
Mandrake, you wouldn't have assumed how the modem dialing worked,
would have read the instructions and it would likely have worked out
of the box. It seems to me you place the blame for things not working
in the wrong place.


Much later, when 9.0 was just out, I decided to see if things changed
for the better. It was still dog slow, it didn't have any switch to
move the picture on my screen to the left, unlike other Linuxes or
BSDs (CRT; if you have a high frequency the picture in X11 tends to
move right, don't ask me why). I didn't have a modem anymore, and at
least IIRC it worked more or less in general. But with my screen
picture totally moved to the right, it wasn't exactly what I'd call
usable or user-friendly.

There is a big difference between useable and user friendly. I find it
incredible how people often make the criticism of linux along the
lines that it isn't user friendly like Windows, but totally over look
they fact they are comparing a commercial product which sells for
considerable money with a free product (yes, I realise you probably
paid something for it, but in reality what you paid for was mainly the
packaging and CD as you could just download it and burn your own CD
for nothing). I'm also quite surprised that Red hat didn't have a tool
you could use to tune your video timings - while its been some time
since I ran Red Hat on a desktop with X, my recollection was that it
came with a couple of tools for this process. I've also found that if
you provide the correct details regarding the refresh rates supported
by your hardware, the screen is usually pretty spot on and if its out,
it is easily adjusted with the monitor's controls. If you find the
screen moves right by a full half a screen, then this is an indication
you definitely have things misconfigured and have either provided the
wrong values when configuring X or the hardware is unrecognised and
the installer was not able to auto detect the correct settings.

However, it is interesting to see how quickly we become accustomed to
simple idiot proof configuration. When I first started running X on a
PC, you had to calculate all the timing and modeline settings by hand
- it was difficult and you could easily damage your hardware. However
at the time it was definitely worth the effort as MS was still stuck
back with Windows 3.11 and compared to X, it was a joke (and lets not
even consider network support).

of Red Hat worse than Windows 98? (Note that I don't personally like
Red hat as a distribution, preferring Debian, but I would certainly
find it interesting to see how you can justify such a claim with any
actual facts.

Its obvious your experiences with Linux were not great. There is a
saying that its not that Linux isn't user friendly, its just fussy
about who its friends are. There is no argument its not the right
OS for everyone and if asked by someone what OS to run, I certainly
don't just automatically say Linux because it depends so much on the
individual and what their expectations are and the type of use they
are looking for. I'd say I've probably recommended Windows and Mac far
more often than Linux. However, for some users, particularly those who
have an interest in understanding and possibly development rather than
just surfing and e-mailing, I think its often a good choice. However,
you do have to recognise and be prepared for a complete paradigm
shift. If your after a windows clone, you will likely be disapointed.


The interesting thing is that back then everybody described Debian as
some monster from hell that would scare newbies away, when in reality
RedHat was that monster (and far worse than I could have imagined!),
and Debian was really easy to install, even for someone like me who
got his first PC (with Win98, which annoyed the hell out of me, so
that I decided to try that "Linux" thing a year later) in 1999! I
guess the Debian catch is that you have to be able to read, and hit
return sometimes. ;)

I suspect if Red hat had been your first exposure to Linux, your
opinion would likely be exactly the reverse. Its like debates
concerning editors - most people find the first editor they ever
remains their favorite regardless of features etc. Its very much what
you get accustomed to that sets your expectations and what you look
for and value.


I bet you only ever used the basic or free distribution and never one
of the enterprise ones either.

No, I paid lots of money for it, as much as for two Debians combined.
Actually later the free versions, RedHat 9.0 and Fedora Core 1 were a
lot better than their ugly ancestors, but still not really optimal,
unless you want your computer to start about 500 different servers and
daemons and take five minutes to start up.

Not a very valid criticism IMO - all of that is configurable and
depends largely on what choices you make in the initial installation.
BTW, I was also referring to the enterprise version of Red Hat
(desktop), which is comparable in price to Windows and which has been
made very user friendly. Not my cup of tea but .....


I'd say the growth of Linux in corporate environments is very much
down to at least Red Hat and IBM (plus others like SUSE/Novell). These
companies have made significant contributions in areas such as
clustering and clustered file systems (though some of that technology
they purchased from other developers initially), significant kernel
improvements, such as in the areas of SMP, major architecture porting
contributions to allow Linux to run on other platforms etc. In fact,
many of the larger software companies are now joining/contributing to
open source - even Sun has now released an open source version of
Solaris.

Yes, they did. So effectively they do just what IBM and Sun used to
do: employ people to build server systems and sell hardware+software
to companies.

That doesn't mean that open source in general is a great business
case. How many hackers that developed good, useful, open source
projects are employed for that now? I bet, maybe 20%. The rest found
a "normal" software job, if they're lucky/good.

Even after their recent cutbacks, Novell employs a lot more than 20
developers working on open source software. I suspect Red hat probably
does as well. However, you have missed the point. The business case
for open source is not about the software development - its about the
service. this is what companies like Red hat are doing to create a
business based on open source.



The point I think a lot of people who feel threatened by open source
miss and one which I think drew aluded to is that the market is
rapidly moving away from a situation in which you can write a bit of
software and then earn lots from selling licensed closed source
copies to a model where the emphasis is on long term on-going service.
The actual code is less important.

I don't feel threatened by open source. I like it, or used to, back
when not all popular software was crappy (KDE and Gnome were crappy,
but they weren't THAT popular yet). Today it's all just cloning
Windows software, so I'm not interested anymore.

but how is that got anything to do with it being open source. The fact
people have had their imaginations limited by what they have been
exposed to has nothing to do with whether the code is open source or
not. It isn't like your seeing anything any better out of the closed
source proprietary world.


I simply don't feel that open source is a scalable business case (i.e.
for all programmers), like ESR and RMS claim. As I mentioned, really
GREAT software doesn't need servicing, or very little of it, and most
open source isn't really great in terms of usability.

I have never and would never argue that open source is what everyone
should be doing. Thats totally up to the individual and what they
want/expect from what they do. All software needs considerable amount
of servicing and maintenance. I know nothing about your situation and
experience, but from comments like that, I am almost certain you have
had no or little commercial programming experience. If you are lucky
enough to get to the point where you have developed some software you
are able to sell and make a living out of, I think you will be very
surprised at how much of your time ends up being taken in supporting
and maintaining that software.

One of the reasons it is very very difficult to be successful in
software development as a small developer is that businesses simply
are not prepared to gamble their future on software which may not have
a future. From a business perspective, you have a lot of investment
tied up in your IT. Being forced to change a critical part of your IT
infrastructure because your vendor has gone out of business or is no
longer maintaining the software etc is often extremely expensive. You
have to find an alternative, find some way to transfer data which
often is not cmpatible or has been stored in some proprietary format
and more than likely you will need to change your business processes
to fit with the new infrastructure, which can be very expensive
because people generally don't like change. Consequently, we find
business is vary wary of small operators without a long track record,
even if they do have a superior product.

The irony is that if you are lucky enough to have some software which
you wrote and can build a market for (often its in some very
specialist field where competition is limited etc), if your
successful, you will need to employ more people to keep things ticking
over, provide support, maintenance and upgrades. Suddenly you look
around one day and find out that your now managing a company, worrying
about marketing, advertising, staff, tax, liability and all the other
hassles associated with running a business and no longer get to do
what it is that got it all started - being creative, writing good code
and creating interesting new products.


For me the difference used to be that you had to be a wizard to be
able to configure Windows so everything worked; but it didn't even,
reliably. With Linux you had to be a wizard too, but at least it
worked reliably.

Thats funny. I've always said the reason I got into Linux was because
it was simple, straight-forward and logical and I don't do Windows
because I'm just not smart enough to get it to work reliably!


Now I have a Mac where I don't have to be a wizard. It's interesting,
because I used to be a little Linux zealot in 2000-01, but now I fully
appreciate that paying money can well be worth the time, and nerves,
saved.

Well I agree to a point. I've always believed that generally you get
what you pay for. However, open source isn't about free software.
People often get confused about this. Open source is about access to
knowledge and information - freedom rather than free. The fact its
free is really just a side benefit. In fact, there is no reason you
couldn't sell your software and make it open source.

However, I do have to say that I find your above comment at best
ironic or at worst contridictory given your earlier statement that you
can't see how open source has really benefited many peole. The new Mac
is the first real challenge MS has had at the desktop level. Even
apple have been surprised at the numbers who have switched from
Windows to the new Mac OS. However, wouldn't it seem to you that since
OSX is based on BSD and BSD was open source that you and many others
have benefited from open source?


Interestingly, I think this will continue even more as software
development moves to more "economical" labor markets, like asia. While
you can certainly send development off shoure, its harder to do the
same with more service oreinted activities (though advances in
telecommunications and cheaper rates may change that to some degree).
What businesses want more than just a bit of software is to know that
if they have problems or need changes/enhancements etc, they can get
someone to come to their offices and help them resolve problems,
enhance business process and/or assist them in providing a better/more
competitive service.

Exactly, and the current situation, where you have to buy
super-expensive consultants who end up developing some EJB
architecture for you, isn't really solving problems for businesses,
because I don't see where most problems would ask for EJB and high
consulting and implementation fees. This is evidenced by the fact
that many small and medium businesses don't really have good IT
investments, because they only software+services that there are, are
too expensive for them and so don't really bring a financial *benefit*
for the company (which is usually the point of investments).


My earlier comments address this a bit. I think you need to broaden
your view a little. for business, the IT infrastructure is about a lot
more than just a piece of software being purchased and installed. A
lot of small and medium businesses use consultants precisely because
they cannot afford to employ permanent IT experts. they don't have any
"in-house" expertise and therefore must bring it in when required. You
cannot sum up all consultants in the same way and nor can you assume
any consultant or software company is going to create some EJB monster
to solve a simple problem. these are all wild exagerations without any
real foundation. Sure, Java has been the flavor of the month for the
past few years, before that it was C++ and before that C. Next month
it may be python, ruby or whatever. Its really what the market will
support and at the moment, java is it until something else looks
better. However for something to look better, it takes a lot more than
a powerful language. We might not like this and we may feel its
inefficient or wrong, but its reality. We can work to change it, but
to do so, we need to recognise what the realities of business are and
realise that its not just about the technology. While it may seem
insane to many of us, technology is less critical to success than any
of us would like to admit. We can run around shouting til we are blue
in the face that Lisp (or any other technology) is the better choice
and it won't do any good unless we can prove that what we are arguing
for is able to address the fears and concerns the prospective customer
has, even if most of the resistance is based on FUD.

Tim

--
Tim Cross
The e-mail address on this message is FALSE (obviously!). My real e-mail is
to a company in Australia called rapttech and my login is tcross - if you
really need to send mail, you should be able to work it out!
.

Relevant Pages

  • Red Hat: Open Source News (fwd)
    ... Red Hat Open Source News ... To challenge the dominance of Microsoft's Windows PC operating system, ... Japan is promoting the Linux system in public schools. ... Open Source Law: Open source and the IT trade deficit ...
    (comp.os.linux.announce)
  • Re: Check that out: THe Bottom Line: Software and copyright
    ... Undisclosed source code did NOT originate with Microsoft. ... to get Adobe Photoshop to work on Linux. ... Open source can be a misnomer, but in general, open source is a product of ... SCO can't be sold short ... ...
    (linux.redhat.misc)
  • Re: Check that out: THe Bottom Line: Software and copyright
    ... Undisclosed source code did NOT originate with Microsoft. ... to get Adobe Photoshop to work on Linux. ... Open source can be a misnomer, but in general, open source is a product of ... SCO can't be sold short ... ...
    (comp.os.linux.misc)
  • Re: Future of IT in Lebanon
    ... Because India has the lowest per capita GDP? ... So Longhorn is not an experiment and Linux is an experiment? ... Developers can take an Open Source product and build on it, and that's great, but a couple of issues: i) They still have to target an environment; ii) none of the products providing real revenue to corporations today are Open Source. ... The Novel Linux integrates the Linux kernel product with other propreitary products to come up with a new derived product. ...
    (soc.culture.lebanon)
  • Re: GPL vs non-GPL device drivers
    ... If Linux becomes closed to external drivers, ... You don't contribute to the kernel ... you aren't willing to open source your kernel code. ... are unwilling to play by the rules those who have donated to you set out. ...
    (Linux-Kernel)