osfree.org

#1018 Re: [osFree] 269 members but very few posts?
Expand Messages

Alan Duval
Jul 19, 2004

On Sun, 18 Jul 2004 19:40:43 -0000, Tom Lee Mullins wrote:

>It is odd that there is 269 members in this group but
>very few posts. Have so many dropped out but not un-
>subcribed to this list/group or are they just inactive?
>
>BigWarpGuy

Hi,
I guess there's a lot of people like me who watch what's happening but are not programmers. I think there has
never been enough programmers who have enough experience to work on this project. Lyn has documented
what needs to happen but no one seems to support his efforts. I can't see that any other approach will work.
As he says, unless the development process can be improved substantially by development of appropriate
software tools, the effort wouldn't be worth it. I believe he is still working on the development of such tools.

IBM seems to be rapidly shifting all it's support to Linux, so probably Linux is the way to go, although I still
prefer eComStation.

Regards,
Alan Duval

#1019 Re: [osFree] 269 members but very few posts?
Expand Messages

Tom Lee Mullins
Jul 19, 2004
....

>
> IBM seems to be rapidly shifting all it's support to Linux, so probably
> Linux is the way to go, although I still
> prefer eComStation.
>
> Regards,
> Alan Duval
>

It would be nice if some of that support was given
to the OSFree project. I have suggested it a couple
of times (at least - will continue to do so in the
future) via their 'feedback' section of the download
area of their site. I told them that supporting this
project would not only show support for open source
projects (which seems to be their current thing) plus
enable them to be free of OS/2 (eventually).

BigWarpGuy
* * * * * * * * * * * * * * * * * * * * * * * * *
* * * * eComStation * * * * * * * * * * * * * * *
* * the os for the internet generation * * * * * *
* * * * * * * * * * * * * * * * * * * * * * * * *
* * * http://www.ecomstation.com * * * * * * * * *
* * * * * * * * * * * * * * * * * * * * * * * * *
* * * * http://www.os2world.com * * * * * * * * * *
* * * * * * * * * * * * * * * * * * * * * * * * *
* * * * * best site for OS/2 Warp Users * * * * * *
* * * * * * * * * * * * * * * * * * * * * * * * *

#1020 Re: [osFree] 269 members but very few posts?
Expand Messages

Lynn H. Maxson
Jul 19, 2004
Despite all the fears surrounding OS/2 when IBM switched its
focus, moving much of its user community into panic mode, we
have since seen some enhancements in eCS and from Innotek,
OpenOffice, and others. In reality we have a "breathing
zone" of some years of a working, reliable OS along with a
significant array of applications. With the Innotek runtime
improvements we will probably see even more applications.

Now frankly I don't know how many programmers subscribe to
this list. I consider myself a developer even though I have
been programming some 48 years. My experience says the
last thing I need to do is to start programming, producing
actual code, before having thought through a system. It's the
maintenance, i.e. program changes, which occur in
development that run its costs and time up, a process which
continues once the first version is released.

In my view we have enough breathing space to do what no
other OS group has been able to do: to do it right. To take all
concerns, resolve them in a systemic manner, avoid conflicts,
and create an efficient, effective OS without reservations or
compromises or apologies.

Let's be clear on one point mentioned earlier. It isn't
programming development that kills you. It's the maintenance.
It's the maintenance that begins in development and continues
to increase during its lifecycle. You can't eliminate
maintenance as that reflects the dynamics of an OS's
environment, i.e. its using community.

You can, however, significantly reduce it, reduce the number
of people needed, reduce the time needed, and reduce the
aggregate cost. This all comes from changing the tool(s),
primarily the programming language and its
implementation...with the emphasis on the implementation.

You don't have a desktop OS today not built on using some
variant of C and OO. If there are two things to avoid, I just
mentioned them. You avoid C because it's a third generation
language, that started out crippled and despite all surgical
operations over the last 30 years has remained so. You avoid
OO, because all programming from its inception has been
object based. OO is a restricted discipline that has both
increased processor resources consumed, people resources in
development and maintenance, and software costs.

Instead you move to a fourth generation language, a
programming language which is your initial specification
language, i.e. the formal language into which the informal
language of your requirements get written. It's a
characteristic of fourth generation languages that once you
have the "unordered" specifications the software assumes
responsibility for the analysis, design, construction, and
testing. It's also a characteristic that has been unrealized
because fourth generation programming languages like Prolog
have started from a C basis, thus embedding themselves with
the same crippling effects.

The plain fact of the matter is that you have three pre-C
programming languages--APL, LISP, and PL/I--that contain all
capablities and more than all other third generation,
programming languages combined. You have the operator
richness of APL. You have the operation on aggregates
(arrays and structures) of APL and PL/I. You have the
"complete" data type support of PL/I. You have the
"complete" exception handling of PL/I. You have the list
aggregate, i.e. a native list, of LISP.

All you need to do is combine them in a single language. To
that language you add the fourth generation "assertion"
statement to complement the third generation "assignment"
statement. You implement it with a fourth generation logic
engine, that which does the analysis, design, construction, and
testing. You allow the implementation to operate natively in
interpretive mode with an option to produce compiled output
for production version of the software.

Whether you buy this approach or not it doesn't change the
fact that its independent of the design of the OS. If you
separate the design of an OS from its programming, the cost
of maintenance, i.e. changes, in one versus the other, you
come up with an irreducible cost in design. However, it is one,
which distributed over the time available to you, you can
afford. You cannot, however, afford the programming
associated with your design with current C-based tools.

So if I don't seem unduly upset by the lack of programmers on
this list, I will be less upset by the loss of wasted
programming in the course of design. I therefore place
emphasis on the importance of design first, which once
complete to a functional detail level you can implement
however you choose at a much less development cost.

Perhaps by that time a small group of us, local to my area,
will have implemented the design language.

#1021 Re: OS/2 Follow-On
Expand Messages

Lynn H. Maxson
Jul 19, 2004
John Baker writes:
"I am one of those individuals who follow the osfree mailing
list, but who rarely actually post anything.

Your posts seem to be some of the few that actually
contribute ideas."

Because he may have left in disgust some time ago, I miss the
posts from Frank Griffin. We differed significantly in
approaches, but I had tremendous respect for him.

I would hope that the OS/2 community has risen above its
initial "panic" mode and the accompanying emotional release
that followed IBM's withdrawal of OS/2 from the retail market.
We were in no danger then and even less now. As we have in
many areas we assumed responsibility through various
volunteer efforts to right our own ship.

Ultimately we need an OS/2 replacement, not to disparage
Serenity Systems and eCS, but to insure that we have
complete responsibility for our future. "Ultimately" implies
several years, possibly a dozen or so, out. That means we
have something that other OS developers do not: time to do
an OS "right", i.e. without compromises and extensible with
full backward compatibility.

We have the ability to produce an OS with no equal on the
desktop. No equal in performance. No equal in function. No
equal in lower maintenance time and cost. That's what we
have and can keep if we do not engage in a rush to
programming. Nothing will kill or run up the cost and time of a
project like programming before it's time.

The problem lies in procedural or imperative programming
languages, first, second, and third generation. The solution
lies in declarative programming languages or fourth
generation based on logic programming.

For reasons not clear to me we have chosen to implement
each generation separately as languages. This separation had
little effect in first (actual) and second (symbolic assembly)
generation which basically maintained a 1:1 ratio between the
two. However, this disappeared with the advent of third
generation, which basically means that no language was
capable of implementing itself entirely. At some point you had
to inject first or second generation programming.

Now there's no real reason why the second cannot contain
the capabilities of the first and the third, the second, and so
on. Every programming language has four components: (1)
syntax, (2) semantics, (3) proof theory (code generation), and
(4) meta-theory (e.g. compiler options).

Now imperative languages differ from declarative in terms of
the proof theory. Imperative languages rely upon the
"assignment" statemen, e.g. "a = b;", which can only fail if the
compiler cannot generate the code which assigns the contents
of "b" to "a". Declarative languages rely upon the "assertion"
statement, e.g. "[a] = b;", which involves additional proofs
above simple code generation. Placing "a" in brackets says
that [a] is not an variable, but a potential of a list of
variables. In fact as an assertion it may fail the proof process,
returning the value "false", an empty list. Thus a list may
include 0, 1, or more entries.

Now all of logic programming relies on list aggregates, the [a]
in this instance. The most common example of this is an SQL
query which returns no entries, i.e. which fails. The SQL query
itself is an assertion whose possible outcome is "false" or one
or more "true" instances, which comprise the list aggregate.

Unfortunately no available programming language supports
both assignment and assertion statements, although in every
implementation an assertion must decompose into
assignments, i.e. the imperative language of the instruction set
of the computer. Thus no existing declarative langauge is
capable of implementing itself entirely.

Now the most complete and powerful programming language
available today, bar none, is PL/I. It offered capabilities prior
to 1970 greater than all of the languages offered since put
together. I realize that the statement is strong, but realize
that it is true.

However, it is a third generation language, meaning its main
processing occurs through assignment statements. To
enhance it to fourth generation means to add an assertion
statement. To add an assertion statement means to have a
native list aggregate. It also means that any implementation
involve a two-stage proof engine, a completeness proof
(which includes code generation as well as analysis and
design) and an exhaustive true/false proof for testing.

The beauty of fourth generation lies with ability to accept
unordered input of logical segments and as part of the
completeness proof organize the logical segments optimally.

That says that maintenance occurs only on the unordered
input and never on the optimized organization. That means
you do not use the old version (optimized organization) as the
input for the new optimized organization, as is done currently
in imperative languages. Therein lies the major cost and time
in maintenance in going from one version to the next.

So you generate each version anew from the unordered input,
at a cost and time thousands, if not millions, of times faster
and cheaper than currently. That reduction puts it well within
the range of the resources available to the OS/2 community.

It's no minor task completing a tool which allows this,
considering that the tool is written in the same language.
Thus you must essentially begin from scratch. You have to
have a version of the tool not written in the language initially.
That ends with the second version which is entirely
implemented in the language. Thus it is self-defining and
self-extensible.

Therein you have my current pursuits in tool building. In the
meantime I'm more than willing to engage in a design process
whose specification language matches that of the tool. You
may argue that this is in fact programming. You would be
correct. However, you do not engage in optimally organizing
source code across logical segments. Thus you avoid entirely
the major cost and time of maintenance. You leave it up to
the software.

I see no reason why we cannot design the most optimal OS
possible which includes as a subset the OS/2 API. We can
effectively code it while designing it, throwing away or
changing source code without ever incurring the costs and
time in doing so currently.

Why not design the "perfect" OS as long as we have the time.
We will have the "perfect" language to do so all along prior
to its implementation.

#1022 Re: 269 members but very few posts?
Expand Messages

pinoozzyid
Jul 21, 2004
--- In osFree@yahoogroups.com, "Alan Duval" <amoht@o...> wrote:

> On Sun, 18 Jul 2004 19:40:43 -0000, Tom Lee Mullins wrote:
>
> >It is odd that there is 269 members in this group but
> >very few posts. Have so many dropped out but not un-
> >subcribed to this list/group or are they just inactive?
> >
> >BigWarpGuy
>
> Hi,
> I guess there's a lot of people like me who watch what's happening

but are not programmers. I think there has

> never been enough programmers who have enough experience to work on

this project. Lyn has documented

> what needs to happen but no one seems to support his efforts. I

can't see that any other approach will work.

> As he says, unless the development process can be improved

substantially by development of appropriate

> software tools, the effort wouldn't be worth it. I believe he is

still working on the development of such tools.

>
> IBM seems to be rapidly shifting all it's support to Linux, so

probably Linux is the way to go, although I still

> prefer eComStation.
>
> Regards,
> Alan Duval

Well, I don't care much about what IBM is doing on the operative
system ground because I won't follow their lead. To me we have such a
good platform to work on with many advantages over other new OSs but
we lack will. Maybe its because IBM still "supports" OS/2 and we are lazy.

But think about these things. OS/2 have thoudsands of applications and
I mean lots of them (most outdated true, but they are there). Tell me
how many of the new platforms have; say DB2, MQSeries, WebSphere,
mySql, Mozilla, PIM applications, Maple, Mesa, Lan Server products
(many of those), LOTS of shareware and freeware, etc, etc.... SkyOS?,
NO, BeOS? NO, AmigaOS? NO, MOrphOS? NO, then why in the hell they were
able to create these systems and we are not, if we have a much bigger
chance to gather a crow?.

If we were able to create such a beast then we could show IBM of the
idiotic decision of backing Linux, instead of creating a "free" OS/2,
it would of cost them much less, since the applications were there,
but instead, they dropped OS/2 and reinvented the wheel creating all
these new application for an OS that wasn't even theirs. (clever, very
clever (being ironic there)). The only reason I could think of, is the
stigma of the OS/2 name, but there are many reasons, that stigma,
could of been overcome (offering OS/2 for free for example).

Well enough ranting

Leonardo Pino

#1023 Re: [osFree] Re: 269 members but very few posts?
Expand Messages

Lynn H. Maxson
Jul 21, 2004
We really don't know the "full" OS/2 story regarding IBM's
decision. The only thing we know is that the PC division
responsible for OS/2 was losing about a billion dollars a year
for several years before IBM pulled the plug. Part of that
came from loss of marketshare in PCs due to the $35 increase
in cost per Windows license IBM had to pay because it also
offered OS/2 on the same machine. We can reasonably ask
why IBM did not sue M$ for this provably anti-competitive
contract or even the more infamous non-disclosure contract
which discouraged multi-platform support from application
vendors.

Nevertheless Leonardo Pino is correct in that OS/2 did and
does have a number of applications. If OpenOffice is any
indication of how close ODIN is to meeting its objective, OS/2
stands to gain quite a few more "Windows-only" apps.

The challenge lies not so much in growing the number of OS/2
applications, originally the greatest fear dominating the OS/2
community, but growing the number of OS/2 users. We
certainly can buy copies of Warp4 cheap enough. When
coupled with fixpak 15 we have a viable OS. Thus it's not the
cost of the OS holding back our growth.

While the discussion here has focused on the replacement
OS/2, the kernel, the major software challenge lies in OS/2,
the package...which includes the kernel. The difference
between the OS/2 and Linux communities is that Linux has
one. Linux has a dedicated and independent group, organized
to enhance it and support it. OS/2 does not.

So we represent by and large a group of OS/2 users, but by
no stretch of the imagination anything close to Linux as a
community. That lack stands between us now as it did way
back then in getting the respect from IBM that we felt we
deserved. Until we get our act together we need to stop
claiming community rights and privileges until we suck it up to
do what is necessary to become one.

#1024 Re: [osFree] Re: 269 members but very few posts?
Expand Messages

Gregory L. Marx
Jul 22, 2004

Hide message history
On Wed, 21 Jul 2004 14:48:08 -0700 (PDT), Lynn H. Maxson wrote:

<snip>

>While the discussion here has focused on the replacement
>OS/2, the kernel, the major software challenge lies in OS/2,
>the package...which includes the kernel. The difference
>between the OS/2 and Linux communities is that Linux has
>one. Linux has a dedicated and independent group, organized
>to enhance it and support it. OS/2 does not.
>
>So we represent by and large a group of OS/2 users, but by
>no stretch of the imagination anything close to Linux as a
>community. That lack stands between us now as it did way
>back then in getting the respect from IBM that we felt we
>deserved. Until we get our act together we need to stop
>claiming community rights and privileges until we suck it up to
>do what is necessary to become one.

Hear ! Hear !

This is truly an insightful observation. And so very true.
It's ironic IMO that we have everything BUT a community.

Who's the OS/2 equivalent of: Torvalds, Raymond or Stallman ?
Linux has many supporters in the media who have the plateform to push and comment on Linux.

OS/2 has no one pushing in the popular media. We used to have a few people but they have gone now.
Occasionally they will write something and OS/2 will get mentioned, but it's always in passing. A historical context.

The shame is IMO that we do have some good people but they are flying solo and doing their thing independently.
I'd like to see ALL the non-United States people work out of NetLabs. Make that be a real portal showing the non-US strength of our OS.
For the US I think Bob St. John and eComStation.com would be the best site to make a US portal.

We do have a handful of developers but they see far-flung and scattered about.
Group everyone together under a common site and the overall community would appear much more viable.

Gregory L. Marx

#1025 Re: [osFree] Re: 269 members but very few posts?
Expand Messages

Lynn H. Maxson
Jul 22, 2004
Gregory L. Marx writes:
"...The shame is IMO that we do have some good people but
they are flying solo and doing their thing independently.
I'd like to see ALL the non-United States people work out of
NetLabs. Make that be a real portal showing the non-US
strength of our OS. For the US I think Bob St. John and
eComStation.com would be the best site to make a US portal.
..."

Well, we do have one group, Innotek, receiving income from
IBM and enterprise OS/2 users, ostensibly to ease those users'
transition from OS/2. As a byproduct of that effort we have
gain several "strategic" software advances.

I would agree on a portal in Netlabs. Serenity Systems,
however, is a "for-profit" enterprise. As long as it perceives a
profit in its marketplace it will continue to develop eCS.
Though it may package open source software into and with
eCS I don't believe it wants to drain any of its "meager"
resources in support of a "non-profit" effort. In my mind that
indicates that Netlabs become an all inclusive portal.

I belong to SCOUG, the Southern California OS/2 User Group,
whose Programming SIG recently has come to focus on open
source support for OS/2. I don't know on what we will decide
to focus our effort. I would like to see us at least examine
something like Mozilla as the IBM-sponsored team on that
project no longer exists.

I've attended and presented at enough Warpstock's to know
that pockets of talent abound for OS/2 support. VOICE once
represented (and may still) a viable portal. I had my
differences with its leadership and to prevent those
differences from detracting from its good intentions with the
Warpdoctor project I withdrew.

I would just like to make the point that even if IBM could offer
us complete rights to OS/2 as open source code we, as well as
Serenity Systems and others like Innotek, could not sustain its
further development with any less sources than IBM, Apple, or
M$ devote to their efforts. That the LINUX community
overcame this hurdle, requiring more in people resources and
time than IBM, Apple, or M$, lies in not having to recoup the
costs of its "volunteer" labor. Open source development
remains less efficient in terms of organized effort than closed
source.

OS/2, the kernel, doesn't require more than a handful of
developers. We could easily produce and maintain an OS/2
replacement kernel. In OS/2, the package, as it does in the
Linux open source arena, runs into the thousands. This
number doesn't lie in those who write the source, which are
relatively few if you believe the complaints of those who do
about those who don't, but in those who test the code, i.e. its
using community.

Whenever you see anyone talk about an open source project
that receives daily updates or changes be aware of the need
for "sacrificial lambs" for testing, those hardy souls willing to
ride the bleeding edge in which the rest of us care little to
participate.

Therein you see the major flaw in third generation languages
largely overcome in fourth due to its "automated" testing, the
exhaustive true/false proof, the second stage of its two-stage
proof engine. We've complicated it by the restricted form of
OO technology. To formalize this OO form we've normalized
on UML, the Universal Modelling Language, which has jacked
up the front end costs of analysis and design through
excessive manual documentation. This OO form does not
allow compatibility among class libraries due to its
"restrictive" implementation of inheritance. This OO form
requires more in terms of system resources in execution than
the one it replaced.

C is bad enough, C++ is worse, and JAVA is out of
consideration. Not because they will not work, they will. But
because of the human resources required. You don't have to
look any further than Linux and the open source community
to get a feeling for the numbers.

That's why I'm happy that IBM cannot release the source code
for OS/2. That means that we can write an OS/2 replacement
in something other than a third generation language like C,
C++ or C#. We can use a fourth generation language and
bring our people cost, by far the principal cost of software
development and maintenance, down well within our ability.

Closed source or for-profit enterprises do not deliberately
invest in more resources than needed. In general they
operate more efficiently than open source. So if you look at
the people numbers they have engaged in maintenance, the
majority of whose time is spent in testing, the open source
community, if it uses the same tools, with have to come up
with that number plus, i.e. more than that number.

For those who do not understand why I focus on the
implementation tools, that which you use to write an OS, and
not the implementation, the OS itself, they need look no
further than this number based on people resources required
as well as the time consumed in testing.

Unfortunately the fourth generation language and its
implementation do not exist currently. I think I have the
language. I'm currently working on the implementation.

That's why I emphasize engaging in the documentation and
design of an OS/2 replacement prior to concerns about its
coding. That design includes its formal specifications. Those
formal specifications in an implementable language. I'll offer
you a clue to the source of that language.

#1026 Project Definition
Expand Messages

John P Baker
Jul 23, 2004

I have begun the process of creating a project definition within Microsoft Project Professional for the creation of an OS/2 follow-on.



And before anyone returns a snide comment, yes, I know that it is a “Microsoft” product, but it is what I have, so we will just have to live with it.



I am going to setup a very vanilla web-site over the weekend where I can make a web version of the project definition available.



As additional items to be added are identified, the project definition and the web version thereof will be updated.



Please note that this is an outline which will address both code and documentation. Many entries will occur multiple times. For example, an API definition will occur under a “Code” grouping as well as under a “Documentation” grouping.



There are several items of information for which I need to identify appropriate sources. Your help will be appreciated.



For a filetype of “AVI”, where can I find detailed information for the internal structure thereof? Are there multiple implementations (versions) of the internal structure of this filetype? If so, I need to identify appropriate sources for each implementation (version).



For a filetype of “BMP”, where can I find detailed information for the internal structure thereof? I am aware that there exist multiple implementations (versions) of the internal structure of this filetype.



If there available an “INF” viewer that runs on Windows?



If you have an interest in contributing to this effort, please drop me a note, indicating your specific area of expertise.



If your are looking for a hand-out, and have no interest in making a contribution, please don’t waste your time nor mine.



Please understand that I do not intend to write the first line of code until a reasonably comprehensive project definition is completed and the pre-requisite documentation is in a useable form. You may take that to mean that no code development will begin until next year.



John P Baker

Software Engineer

#1027 Re: [osFree] Project Definition
Expand Messages

michael
Jul 23, 2004
Hi,

my name is Michael and I've reading this discussion since mounths/years.
I'm interested in documentation.
My knowledge based on Warp 4 and Windows NT/XP.

Regards,

Michael

Hide message history
>
> If you have an interest in contributing to this effort, please drop me a
> note, indicating your specific area of expertise.
>