port I/O abstraction macros

"Le Chaud Lapin" <unoriginal_username@yahoo.com> wrote in message 
news:1149685713.011841.53460@h76g2000cwa.googlegroups.com...
> Tom Lucas wrote:
>> I see little point in graduating knowing how to write a compiler or 
>> design
>> an FM radio unless you are directly going into Sony's Radio Compiler
>> division. Knowing what a compiler does and the sort of tricks current
>> optimisers play will be more useful in the real world than knowing how to
>> parse an arithmetic expression. Similarly knowing about RF, demodulation 
>> and
>> amplifiers is more useful to the real world than an rudimentary FM radio
>> kit.
>
> There seems to be a bit of a contradiction here.  You seem to be saying
> that shallow practice is not good (FM kit), but you also seem to be
> saying that basic theory (being able to parse arithmetic expressions)
> is not good either.

What I'm saying is actually implementing a solution is not necessary and 
moves away from the theoretical into the practical. I would also class the 
mechanics of parsing an expression as an implementation challange and not 
part of the theory of a compiler (although it is an important part of the 
solution.)

> A person cannot design an FM radio without understanding many aspects
> of electronic theory.  In the case I was speaking of, the electronics
> course gives a bag of components, and a lot of theory.  You're expected
> to make everything, including the amplifiers, filters, phase-lock loop
> (from a pre-packaged VCO), etc.  Anyone who had not been paying
> attention to the theory would simply not be able to make the radio.
> The same course (or the next one) introduces Kalman filtering.  It's
> not a connect-the-dots type project.

There is certainly value to prove the knowledge gained on the course but 
this is simply a practical way of verifying that the exams are setting the 
correct questions. My point is that I would regard the radio (and compiler) 
construction as a polytechnic activity whereas the universities will be 
happy to impart the knowledge academically and test it by exams.

> For the compiler, I think it is very important to be able to parse
> arithmetic expressions.  Knowledge of automata theory can be
> invaluable.  For my current job, I recently had to make a small
> compiler for a stack machine.  Here's the tokenizer for lexical
> analysis:

Again, I think it's important to understand what the compiler is going to do 
and have an idea about how you would approach an expression parser but very 
few people actually write a compiler. I would say that for the majority of 
graduates the theory behind an abstract generic compiler is enough and 
actually writing one is time that could be spent on something else (although 
there is some merit in taking on a complex software design task)

<snip code>

> I  have no clue about optimization, but what I needed to know to make
> this compiler can be found in the first few chapters of any compiler
> book.

Hence why there isn't a lot of value in writing one unless you actually need 
to.

> Self-reliance is achieved by focusing on a small bit and
> learning that well, than reading the whole book casually and not
> remembering anything.

I agree that's true for me but people learn in different ways. I learn by 
doing but I know people who can read the book while I do the examples and we 
both end up with the same knowledge.

> I started dong this long ago after a
> converstation I had with a professor just before my graduation.  I
> asked, "There is so much to learn in the world...is it better to hurry
> up and grab what you can and depend on other people for deep insight or
> is it bet..."  he finished my sentenced quickly and said, "It's better
> for you to know."  I asked, "But does one person have time in his life
> to...?" he says, "Yes."  I asked, "Are you sure?" he said, "Yes."

Sure a professor will have plenty of time to know it all - he doesn't have a 
manager baying for blood because products in the field crash every 12532th 
time they are turned on :-) Philosophically it is better to know the answers 
yourself but often it is sufficient to know where you will find the answers. 
The more you know the more complete a solution you will find but realism 
will always intrude and take away your learning time. 

Paul Keinanen wrote:
> On 6 Jun 2006 09:31:30 -0700, "Le Chaud Lapin"
> <unoriginal_username@yahoo.com> wrote:
>
>> I just had a protracted argument about hiring someone who claimed to
>> have device driver writing experience, but did not know how to divide
>> and integer by 32 using bit shifting.
>
> How do you elegantly divide a _signed_ integer by 32 using shifts in
> systems that are using 2's complement representation ?
>
> Paul

#define    DIVBY32(i)    (i<0? -(abs(i)>>5):i>>5)
// Will not work with the lowest negative number

i = DIVBY32(i);


-- 
Best Regards,
Ulf Samuelsson
ulf@a-t-m-e-l.com
This message is intended to be my own personal view and it
may or may not be shared by my employer Atmel Nordic AB