Semi-OT: Database for "average" users

[Peter]

On Mon, 2 May 2005, David Mayerlen wrote:

> I don't think that non-techie use is the goal. The goal is to "normalize"

The goal of the borg is to sell pretty animations for good money after 
making the user believe that he will be able to do with it what the big 
guys are doing, by clicking on a few buttons. The feature set that 
accomplishes that is known as the 'least common denominator' and it has 
become the standard in computing, after destroying all the parts of the 
industry that did not migrate to India or China or other one-diskette 
countries.

I find it funny that certain relatives of mine spend hours on the phone 
explaining me how they fixed XP for the 4th time this month and that 'it 
works' (I have been running linux since 1996 as a desktop and I remember 
two crashes with data loss in this period: one hard disk and one SDRAM, 
about 5 years apart, and I have been building computers since about 1987 
- by that I mean embedded to PC, including software and hardware). Imho 
a games psychologist is a must have consultatnt for any successfull 
computer project. Give the user a couple of animations to click on and 
he will feel *accomplished*. Add a couple of small hurdles to cross 
(intentional or not) and you have him hooked. You see, now he is an 
expert. He is confident that by clicking here and there he can fix his 
problem. It's brilliant. He will start his own blog and mailing list, 
tell others how to do things (without having a clue about why it 
worked), and be happy. And that's the point of it. His google ranking 
will grow. He will start selling advertising on his site. But it's not 
CS, it's user management and marketing.

> the amount of detail work. Funny how we rarely program in assembler
> anymore. Thats the whole point of what Upstart has done. Is a bit

Speak for yourself. Some people program in assembly all the time.

> ridiculous to program some user/login tables. That has been done so many
> times before surely we can wrap that in a higher level package. We benefit
> in many ways not the least of which is security ... betcha at least half the
> programmers who build it from scratch don't get it right.

Bugs are a part of programming. If something has 100 options and it is 
programmed by someone who has 98.5% error rate then it will have 1.5 
bugs. How many options has your program got ? A 10,000 line program has 
about 400,000 characters. Assuming it is written by geniuses who manage 
99.999% error rate it has only 400 typos, logic errors not counted. Who 
reviewed your code ? Is it open source or did you test it in house, 
between 12 pairs of eyes ? php for example is new and complex enough 
that there is not one security focus digest that does not mention it, or 
an application using it. About the same score as IE and IIS. Coincidence 
? Reinventing the user friendly wheel and hot water comes at a price. 
Someone on another list has a sig along the lines of 'you can't ignore 
20 years of experience'. I like that line very much.

> Ever build a reusable code library in "C"? Given the opportunity most
> programmers seem to want to program everything from scratch and things
> take way longer than they should and then become unsupportable messes.

Unless one programs by contract after an analyst determines what is 
really needed and lays down a roadmap, interface specifications, and a 
testing framework. Which is how CS is supposed to be done according to 
certain outdated books that were superseded by pritty multi-colored 
400-page doorsteps that show plenty of screenshots of Java UIs for users 
in case they cannot find the 2 buttons on a form, or read the text in 
the screen font. Of course the bugs are still there, but now you can't 
fix them, since you have no access to the source ... it takes about 10x 
the lines of code to do something with a gui as opposed to without. 
Considering that no-one could afford to do complete testing of command 
line programs 20 years ago, draw your own conclusions. The pritty gui is 
a container for somebody other's bugs. The closed source kind being the 
kind that cannot be fixed if broken. But you can always purchase 
upgrades.

> Find common requirements. Wrap them in little black boxes that meet those
> requirements.

And join the lemmings who feel so secure seeing so many others running 
in the same direction as they do. Want to stop and take a leak ? Can't. 
You'd get run over. Want to run faster ? Can't. Left, right ? What, are 
you antisocial or something ? Up ? Can you fly ? Stock market went to 
h at 11 because everyone was using the same futures prediction algorythm ? 
Oh well, blame it on the makers. Should have randomized it or something, 
since everyone was using that same particular black box, with its 
rounding errors and estimate bias and all that.

You have a nice system and you have clients for it, and it works. That's 
good. That's what people call 'a solution'. Please don't try to sell 
fridges to the penguins. We got plenty of ice to cool things like that.

Peter
--
The Toronto Linux Users Group.      Meetings: http://tlug.ss.org
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://tlug.ss.org/subscribe.shtml