http://aplawrence.com/Opinion/nntp.html


It seems that this forum is already dying from the above described effect.
IMO we need a backup/alternative.
Wikis seem to be growing.
I've tested an existing one, dedicated to ETH-oberon, but I'm
uncomfortable in having to learn yet another syntax.
It doesn't give immediate WYSIWYG.
In particular 'new lines' need 6 key-strokes "[[BR]]" !
Perhaps this is how wikis work ?
Also since I won't be having 'always on' connectivety, the ability to
compose and read off-line as NNTP allowed, is essential.

Q - What are other users ideas ?

Thanks,

== Chris Glur.

comp.lang.oberon

I do not agree with this. NNTP outperforms any methods of web access or
web based forums. NNTP clients provide many forms of threading, scoring,
and sorting of articles which are not found in HTTP-based interfaces.

And, yes, we are still running our own NNTP server: news.in-ulm.de.


This is not a forum but a USENET group

The real problem is that the community gets smaller, newcomers know less
about USENET groups, and that many subscribers (like me) are just lurking.


Then you will lose subscribers like me. I do not have the time to regularly check web pages.


Wikis are not really a replacement for USENET groups.

Discussions in wikis need quite some discipline and experience as you
have to take care of indentation and signatures. Users get easily lost
over lengthy threads: What has been added recently? Do I need to read
the old stuff I skipped over already several times?

Wikis are best for a collaborative form of do***entation.


Well the good thing about some Wikis is that you can still use a
vi or any other good text editor for them. (I am typing this right
now using vim and I also type my Wikipedia entries with vim under
the elinks text browser. I also use vim to edit my Oberon program
texts. Using always the same powerful tool for typing and editing
makes my life easier.)

This is one of the good things of standard protocols (like NNTP)
as interface in contrast to WYSIWYG-based interfaces: It gives me
a free choice out of so many available clients. I am not enforced
to use bloated web browsers or the mouse.


Please do not fraction the community any further.

Andreas.

august karlstrom

no-top-post skrev:


I find the newsgroup interface in Mozilla Thunderbird really good.

I don't see your point -- Google Groups is just a web interface to
newsgroups and comp.lang.oberon is (of course) accessible from Google
Groups.

Why? The only problem with comp.lang.oberon is the low activity and the
Chinese spam.

In general, what's missing, as I see it, is a free (as in freedom) high
quality multiplatform standard conforming Oberon-2 command line
compiler. A front-end to gcc would be ideal. I'm sure this would
increase the Oberon user base.

Being free software I find OO2C to be the only current alternative
among the Oberon (command line) compilers. What I don't like, however,
is its lack of focus (non-standard/experimental features), its poorly
designed libraries and its incomplete and out of date do***entation.

As far as I know Wikis are good for do***entation, not for (lengthy)
discussions.


August

stewart greenhill

Same here. If the group postings are not there on the same page as my
regular email and RSS feeds, I'm not likely to read them. Thunderbird is
great at bringing everything together in one place.

Low activity probably means low level of interest. It could also mean a
high level of satisfaction, but if this were the case I think there
would be more posts.

Why? Oberon-2 lacks any real useful standards, so every implementation
has to define its libraries and platform interfaces. The result is many
"niches" within the Oberon community and poor interoperability.

The only way to increase the Oberon user base is to build community.
This means using informed advocacy and enthusiasm. It means sharing
interests and ideals in a supportive and non-judgemental way. Based on
what I see in this place I would be surprised if there were more than a
handful of Oberon users left. But perhaps we are all living in our own
little niches and not having much to do with each other.

I guess this is part of Chris' original point. The forum seems to be
dying, but I really don't think that technical issues are the problem.
Its the nature of the Oberon community: too many niches and not enough
common interest. Also: many of the features that once made Oberon-2
attactive over languages like C++ are now provided by mainstream
languages like java and C#. You can't use technological arguments to
answer these issues. You must use idealism.

OOC actually has very good compliance with the Oberon-2 language. Sure,
it implements many language extensions but you are free not to use these
features. The do***entation is certainly out of date in some areas (eg.
the user manual), but with the V2 compiler all modules can be made
self-do***enting using a system similar to javadoc.

Remember that OOC was built entirely from dontations of free time, so
its strengths and weaknesses directly reflect the interests and
abilities of the contributors. The original development team seems to
have dwindled over time in line with the decline of Oberon itself. Of
course, we still welcome all contributions so if something really irks
you please feel free to fix it.

Cheers,
Stewart

august karlstrom

Stewart Greenhill skrev:

Agreed. However, I don't see how a different discussion forum, as
proposed by Chris, would increase the Oberon user base.


As far as I know, the Oberon-2 language report and the Oakwood library
specification are sufficiently precise. Other libraries can be distributed separately.


and (too many) research projects ;-)

If the great real world tools are there, I'm sure the people will come.


When it comes to safety: yes. But these two languages are still bloated
and slow compared to Oberon. Oberon combines the safety of the mentioned
languages with the speed of C and C++.

I see in Oberon a nice blend of idealism and pragmatism. It may be a bit
more idealism (and soul) than in some other languages though.

I agree with that but it's a shame the Oakwood modules were removed,
making it harder to write portable programs.


August

If technical excellence was the determining factor, then NOT 95% of
users would be using M$hitWindows.


If the dreaded/warned-of transformation occurrs it will not be instantaneous.

A forum is a more general term which includes USENET and old ladies meeting for coffee.

No ! 'Backup' is not the default method. Having accident insurance doesn't
mean you must have an accident.

Having safety insurance is not really a replacement for avoiding an accident.

Yes I was shocked/dissapointed that one needed to learn yet
another syntax.
As the URL above refers, most http-clickers just top-post to the
last comment. The ability to conduct multistage threaded dialog is disappearing.


Yes, but I'm predicting what is coming [and hoping than I'm wrong]
rather than telling what I'd prefer.


PS. I wrote this note originally for comp.lang.pop
which seems to have died. The prof confirmed that
the university was considering cutting the newsSever.

Then I read the URL mentioned, where it's expalined
how/why usenet is dying. As others have written, it's
not about technoloy, but rather about psycho-sociology.

A related example: recently my ISP requires TxAuthenticate
for email. The help desk can only advise me for M$-outsp00k
or the 'web-interface'. Fortunately I anticipated such
problems and had a 'backup' google-emailing facility.
Which I hate because of it's baroque web-interface and
slowness [I have to pay for dialup time], but allowed me to
post important emails while I'm patching the ETH-oberon
mailer.

== Chris Glur.

comp.lang.oberon

Which motivations, if not excellence, do you have to continue
to work with Oberon?

I do not use M$ Windows and I will not switch to a web-based forum.
I simply do not have the time for this.

Quite a number of ISPs (including the largest) in Germany provide
NNTP services. Most universities run NNTP servers. I have three news
servers I can use for free: news.in-ulm.de (my ISP), news.uni-ulm.de
and news.belwue.de through my university.

Well, it is important to distinguish between USENET groups and
web-based forums. Newcomers tend to confuse web-based USENET groups
access (Google Groups) with web-based forums, not knowing that there
exist a well-defined protocol for USENET which gives a wide choice of possible clients.

I do not agree. And others like Google do not agree as well.
Otherwise, Google would not have put money in it. (The archival
service of Google Groups is another big advantage of USENET, BTW.)

Then switch to news.individual.net (see http://www.individual.net)
which offer USENET news service for 10 Euros / year.

You do not need your ISP for sending emails. As long as your
ISP provides full connectivity, you are free to send your emails
directly or to use another SMTP service.


Then email transport by UUCP could be interesting to minimize dialup
times for the transport of emails. It is still offered by some providers.

Andreas.

aubrey.mcintosh

I have seen this idea since before the Oakwood meeting, and I never
understand it.

I see that Oberon, the system and compiler, is implemented on many
machines, with very different architectures. Not only that, but I have
seen the language implemented on a Motorola 6811 and implemented a
project on it, using only my background with the other Oberon systems,
and had a straight ahead time of it.

The two things that seem necessary for a new environment are 1) a
compiler and 2) the kernel. The libraries and applications move
unchanged, even across different byte *** machines.

I view Oberon as the most portable system that I know of. I have moved
programs and data files from Solaris to Win-NT just by FTPing the data
files and re-compiling my source.
I have seen epoch changes. There is the V4 --> S3 --> BlueBottle
evolution. But at the time that V4 was released, there were several
candidates for the C language, and you couldn't take your own C source
code from one 'nix machine to another and compile it in less time than
it took you to write it for the first time. You can't take a non
Oberon program written in the same time that V4 was released and move
it.

So, is there an example of non-portability of Oberon that does not
involve pulling the underlying 'nix or MS operating system into an
application?

noch

yes, you are completely right when talking about oberon system.
But we really have a portability problem now in case we don't use
oberon S3 or V4.
Try to compile any old oberon system source with oo2c?


I am not writing about Pow, which has completely different libraries as
well because I don't use it under unix
Without having a set of standard libraries, we have no portability!

Now when I write in oberon I use wrappers to libraries.
To port my program easier from one compiler to another.
In case of oo2c v2 (which is definitely better than oo2c v1) my file io
libraries have roots in TextRider, Files (idealogically differ from
oakwood Files module), Msg, etc. In case of xds compiler my wrappers
have roots in xds specific modula-2 libraries like FIO or more low
level.
I agree that removing oakwood libraries from oo2c was not a good idea
at all.

personally i believe that oo2c v2 way is rather good.
but we need a set of standard libs, where oakwood, iso, or any other,
it does not really matter.
we just need standard
that standard hopefully will include oberon system like libraries.
even assuming oakwood libs are standard we still have problems porting
oberon system sources to any other non oberon system compiler.
for instance, it is usual way for oberon system programmers to use not
oakwood libraries but set of other oberon system specific low level
libraries (texts, kernel, oberon) instead
and if one wants to use oberon system codebase, then he has to choose
whether to write his own wrappers for compiler he prefer or to modify
existing code.
Second solution may be not beautiful, but will result in faster
executables.
That is why I think that oberon standard libraries must include modules
like Texts.
I think that you believe me that what I wrote above really important.

I have to write here about very successful and high quality work again:
FreePascal compiler.
It is an excellent project. They did an excellent compiler, but they
also developed very good site, with docs, news, announecements.
That is why they have many contributions and growing community.

By the way, I noticed that oo2c developers preparing python and java
backends.
Last one could be useful for writing for instance j2me code, but i
don't really see any sence in first.
I think that if oo2c deals with ast then it should be rather easy to
develop a gcc frontend from it

Thanks

Norayr

august karlstrom

aubrey.mcintosh@gmail.com skrev:

I was mainly referring to the compiler specific libraries that comes
with the available freestanding compilers (oo2c, xds, pow etc.).


August