[lug] Why Linux will win and Micro$oft will lose

BOF bof at pcisys.net
Mon Nov 12 07:29:16 MST 2001


Hello,

The comments in this article show me why it will be a long time before 
Linux triumphs over Microsoft.

    http://www.linuxgazette.com/issue72/field.htm

The quality control of many Linux distributions, exacerbated by the 
attitude of many companies in taking responsibility for their products 
leaves much to be desired. There is simply too much of the attitude that 
"we sold you the software, now you have to get it to run" going around. 
As an example, look at Red Hat's customer support. While they state that 
the purchaser is entitled to support, by the time they caveat out all 
the exceptions (including laptops), they end up supporting almost 
nothing other than a straight off-the-CD onto a desktop system. 
 Furthermore, their attitude towards problems leaves much to be desired: 
my message to them that my laptop was freezing up under installation of 
7.1 when it had never frozen on previous versions of 5.1, 5.2, 6.1, 6.2, 
and 7.0, and asking what they had changed in the installation program 
got no response at all -- no even an acknowledgment that they had 
received the message. Of course, since they did not support laptop 
installations (although the installation program specifically had a 
laptop-specific choice), they were under no obligation to answer! My 
already high opinion of Red Hat's customer service was furthered by the 
messages D. Stimits got from them about his problems with the ipchains 
and iptables. Not to mention the ten security advisories that RH has had 
to issue since 7.2 was released less than 30 days ago. Again, the 
attitude seems to be one of "how quickly can we blow you off now that we 
have your money" or, even worse, "you must be stupid because you can't 
get it to work -- RTFM!"

The recent kernel fiascoes with 2.4.11 show a similar attitude towards 
QA: it's out one day and declared an evolutionary dead end. However, 
these problems did not stop with the 2.4.11 kernel. Saturday, when I 
downloaded the 2.4.12 -> 2.4.13 -> 2.4.14 patches, applied them and then 
tried to compile a 2.4.14 kernel, I got stop errors during "make 
bzImage." The error messages indicated that one of the .o files was 
trying to access an undefined function. Thinking that I had done 
something wrong (since this was the first time I tried upgrading via the 
patch process), I tried again, and failed, and then downloaded the 
entire 2.4.14 kernel source and compiled it. Each time I got the same 
error messages.

Apparently the solution was to deleted the two references to the 
function in one of the source files. After I did this, the kernel 
compiled as it should have and so far has been working. But it took me 
an hour to track down the problem (which was listed as a known bug), 
apply the solution and recompile. 

The point that I am trying to make here in my rather long story is that 
QA is extremely poor. The problems that I had were completely 
unnecessary. The function in question had apparently been included in 
2.4.13 but not used in 2.1.14 and was known to be a bug. So why was it 
not corrected before 2.4.14 was released? Why did not anyone remove it 
before releasing the source code? One wonders if the Linux gods that do 
kernel development bother to take a final copy of the code to a new 
machine and try and compile it before releasing it and seeing if it 
really does work or not. From my experiences, it would seem that they do 
this only as a Einsteinian thought experiment:"Yeah, that will work now 
-- no reason to test it." The result of this is that the user wastes 
their time in trying to get something to work that should have worked in 
the first place had someone else done their job right to begin with.

Now don't get me wrong here: I love using Linux and will never go back 
to MS products again. But until this rather cavalier attitude towards 
its users changes, Linux will never make it to the desktop: it is simply 
to hard for the average user to cope with. The reasons for this "stupid 
user syndrome" will make another good rant, so I will leave them for 
another day. (Hint: it has to do with training and indoctrination of the 
way computers are viewed as very simple devices that everyone should be 
able to use).

BOF

Al wrote:

>That was me. Cost is the only factor that drives my company's interest
>in Linux, from the management's viewpoint. 
>




More information about the LUG mailing list