Wed, 25 Mar 1998 17:33:17 +0100 (GMT+0100)
>>>>> "Chris" == Chris Bitmead <chrisb@Ans.Com.Au> writes:
Chris> Rodrigo Ventura wrote:
Chris> Of course you can get simpler. You just eliminate them.
>> Baa. It is implicit one wants to simplify without affecting
>> the functionality!
Chris> Yeah, so what functionality are you worried to lose?
I'm sorry, but I'm afraid this conversation is getting
nowhere, unless there is indeed a point you are trying to make, and I
was unable so far to grasp it. What is really your point? If it goes
beyond LispOS, please, let's talk by emails rather than using the ml.
Going to the point: if you eliminate the UNIX API, what is
left? You ask what functionality am I worried to lose? Simply the
possibility to access any UNIX device by the means of the syscalls
API. If someones puts a new device into the machine, along with a
device driver that registers a char-device, how is it supported on the
LispOS side? With syscalls it becomes trivial!
>> Because simplicity (or cleaniness) of an API is usually an
>> under-rated factor for broad acceptance.
Chris> So why not simplify the UNIX way, instead of wanting a "similar
Chris> API in LispOS"?
Yeah! That would be great! I assumed it was difficult. Do you
see a way of doing that without jeopardizing all UNIX flexibility and
functionality? Let's hear it.
>> Even if a certain API is the
>> best in some technical sense, if it is not simple, if people cannot
>> easly grasp how it works, it becomes useless as far as broad use is
>> concerned? Imagine for instance the Xlib API. It's fast, is flexible,
>> but nobody uses, because everybody prefers a Motif-like API where you
>> have widget pointers and callbacks --> simplicity!
Chris> Xlib is actually quite simple. It's just that you've got to do
Chris> lots and lots of stuff, just to get say a button up on the
I concede that Xlib can be simple after reading tons and tons
of docs. And after several months without thinking about Xlib, is it
possible to get back to it easly? Although I confess my ignorance on
>> Ok, I probably didn't explain myself good enought. This idea
>> came up while I was thinking about whether to use (format)-like
>> semantics or a printf()-like one. Then I had an idea: why not a much
>> more complete and powerful semantics, for instance based on LaTeX
>> syntax, eg "\int". It's much more readable than "%5d", don't you
Chris> Hmm. The whole concept of formatting strings seem a bit flawed to
Chris> me. \int is the invention of a sub-language that is not
Chris> Lisp/Scheme. That is what I thought we are trying to avoid.
Good point. To make the formating string, not a string but a
list, is a neat idea! I like it! It can be some sort of sub-language,
but based on lists. And it could be easly expanded. Nice idea. Really
*** Rodrigo Martins de Matos Ventura, alias <Yoda>
*** firstname.lastname@example.org, http://www.isr.ist.utl.pt/~yoda
*** Instituto de Sistemas e Robotica, Polo de Lisboa
*** Instituto Superior Tecnico, Lisboa, Portugal
*** PGP Public Key available on my homepage
*** Key fingerprint = 0C 0A 25 58 46 CF 14 99 CF 9C AF 9E 10 02 BB 2A