[Oberon] Re:... inet-fetches &Linux/N-O/wiki

easlab at absamail.co.za easlab at absamail.co.za
Mon Dec 4 07:28:52 MET 2006


> > Q - how many users now exist to amortise the debugging cost
> of Linux / ETH Oberon V2=2E3=2E4 =3F
> 
Your email too, looks like garbage since you changed your ISP.
This time I'm not going to edit out the funny "=2E,=3F..etc.",
stuff.   Since you had some deep investigation into 'char
seqences which are escaped', if I post you "=2E", do you see
"=2E" or "<dot>" ?

P. Easthope wrote:
> I've never tried it=2E  Wasn't that released years ago=3F
> 
Since [linux ETH Oberon]
it's quiet good, it must have evolved over several years 
and versions.  It just happened that I was getting 
desperate to have better than present N-O capabilities 
for inet connectivety, while still wanting to use N-O for 
the text once fetched, when this version of Linux-ETH 
Oberon was announced.  Since interchanging files between 
N-O, linux & even DOS is easy, especially if you use some 
mc 'terminals' under linux, you can easily do eg. download
*.pdf, and if need be > pdf2ascii > move to N-O format
for colouring etc. while being digested. If I wanted to I 
could easily spell check this N-O entered text with Linux.

The final stage, where human consumption occurs [and
not the technology] is what's important and where the
value is added.   Analagously people concentrate on the
gee-wizz of how quickly they can get across the ocean 
to another continent; instead of considering the whole
journey from source chair  to destination chair.

> Perhaps use the recent release of Aos and=20
> access a favourite "other system" by VNC=3F
> 
Has Aos got the N-O good chording etc. interface ?
Did they fix the underscore char ?

> > Also collaboration needs a suitable CVS-like framework=2E
> A wiki-facility should suffice=3F
> 
> Alternatively, adopt this proceedure=2E
> Notice bug=2E
> Understand bug=2E
> Implement repair=2E
> Post repair in Web site=2E
> Report repair to this list=2E

No!! Many principles applicable to a one-man-show don't
scale to a multi-person enterprise. Have you seen wikipedia ?
IMO it's the next revolution after google.
BTW I fetch 10 wikipedia [or other] pages via a linux ->
lynx script, while I'm paying for a 3 minute connection to
fetch/test my email.

A concrete example of collaboration where individuals may
fail is:
pop-11, an open source, algol family like syntax language
suitable for AI, is well supported by a very active prof, on
news:comp.lang.pop [the group server has crashed lately].
He reported a rare bug related to array out-of-bounds,
which had been around for a decade.
The prof. doesn't do C or asm. language programming, but
has brilliant writing/explaining/managing skills.
>From his good description and material posted from
UK, I was able to speculate that the unstructured asm code
caused unbalanced stack push/pop depending on whether 
an asm-routine was exited via a return or a jump.

I didn't have C compling software nor experience to confirm
my conclusions, but a collaborator from Poland was abe to 
confirm by adding an extra NOP to balance the push/pop
stack mechanism, and compile/test the prof's sample.

Such conditions, where collaborators from UK, S.Africa and 
Poland can use division of labour to efficiently solve a problem 
in a few hours are rare; but if the evolving solution can be 
stored with common access, tasks can be acheived which the
individual can't do.   Even for an individual, although trivial 
tasks are best directly attacked - like making a cup of 
coffee - often an indirect approach is more efficient.  
Subtle problems are often more efficiently solve by indirect
probing.  

 With the inet, these days, it seems that everything has 
 already been done and is published. So 'just do it' is the 
 worst approach.  Producing bananas in Canada is not a 
 good strategy; foreign trade to get them is much better ?
 
== Chris Glur.


More information about the Oberon mailing list