chaos-devel Mailing List for The chaos Operating System

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

Hi everybody,

Are you ready? ;)

Best regards,
Per

Hi cha...@li...,

Add me as a friend on Last.fm so we can share our music taste :) 
Check out what I'm listening to: http://www.last.fm/user/joeljose420/?invitedby=joeljose420&tp=ff_tp_b

--------

Signing up is free and takes less than a minute.
Just click the link to automatically become my friend.
http://www.last.fm/join/?invitedby=joeljose420&tp=ff_tp_b

Visit my music profile and leave me a shout! I'll see you around,

Joel Jose

PS: I'm 'joeljose420' on Last.fm

--------

You received this message because someone (Joel Jose) who knows you sent you an invitation to join them on Last.fm. Your address was not saved and we will never contact you unsolicited. For more information, see our privacy policy at: http://www.last.fm/help/privacy.php

Hello Johannes! Great that someone finally answered my email. :-)

On Wed, 2008-01-02 at 15:11 +0100, Johannes Lundberg wrote:
> Some quick thoughts:
> 
> * Qemu can be used instead of VMware Player, and is fully
> redistributible. I have some hackish code I could use for wrapping it
> all into a single .EXE for easier testing. I will look into that.

Great - have you had any time to look into this further? I have no
first-hand experience with Qemu actually, I remember when it was
launched by the developer of Bochs... or perhaps that was plex86 rather
than qemu. :)

Anyway, I get the impresion that Qemu is less mature than the
"commercial" alternatives such as vmWare and VirtualBox (I use both of
them at work and they work great - except I've had some strange problems
with vmWare networking but they somehow disappeared.  :)

But, if Qemu works well then we might as well use it. The importance is
to make it simple to "bundle" chaos along with a suitable VM
implementation.

> * A ready-to-use developer virtual machine image sounds like a great
> thing indeed. All things we can do to make chaos development and
> testing less time consuming is good.

Yeah, we should just be careful so we don't spend all the time on the
sub-projects (such as that one :) since it can easily "eat" up the time
from the "real" development or whatever you wanna call it. I thought
about the developer VM for example, we could do a real cool one and make
it network-updatable so that we only distribute a "stub" which then
downloads the Debian packets needed (xemacs, gcc, so forth) over the
network every time you boot up. But, I doubt that there is any real
benefits to that approach except for the coolness/geekness factor. :-)

> * TCP would be fun. We need amd pcnet for the emulators though. Any
> good specs out there?

Yeah, I thought about pcnet actually before writing that email but I
forgot to mention it. However, Bochs doesn't need pcnet - Bochs emulate
an ISA ne2k card, and that works great with chaos. Noah Williamson wrote
an ne2k server a while back and I fixed some problems with it last year
(or perhaps 2006, unsure of which :) so that it worked pretty well with
bochs - I could boot chaos, and set up an IP address (perhaps even using
DHCP). I could then ping the chaos machine from my Linux host. :)

Pretty cool feeling I must say.

Anyway: it would be great if you could help out with the stuff you
mentioned. I think it is important that chaos is not a solo project - if
we can have an active commmunity of developers working on it, it will be
much more motivating for everyone to commit time and energy to work on
it. I think we are all pretty occupied with other things nowayears...
--
Best regards,
Per Lundberg

Some quick thoughts:

* Qemu can be used instead of VMware Player, and is fully
redistributible. I have some hackish code I could use for wrapping it
all into a single .EXE for easier testing. I will look into that.

* A ready-to-use developer virtual machine image sounds like a great
thing indeed. All things we can do to make chaos development and
testing less time consuming is good.

* TCP would be fun. We need amd pcnet for the emulators though. Any
good specs out there?

Regards,
  Johannes Lundberg

On Jan 1, 2008 1:15 AM, Per Lundberg <pe...@gm...> wrote:
> Hi guys, and a happy new 2008 to all of you!
>
> (All of the rest of the email is referring to "chaos-old" -- the
> original (and best? ;) chaos, the one that was most "feature-rich" at
> one time. What happened to chaos.net you might ask? Well, development
> stopped quite early after some days of intensive hacking in early 2006,
> partially because Henrik Hallin was attacked by a malicious stranger in
> his home city. He can tell you more about this if he likes to but
> nevertheless, his part of the development was never really taken up
> again after the accident. As far as I was concerned, I didn't really
> have very much spare time to spend on the project.)
>
> As some of you (probably only really Henrik since he's the only one
> reading that list ;-) might has noticed, I've done some basic
> development on chaos lately. Nothing really new and fancy, but what I've
> done is tried and fix some obvious problems with the storm kernel and
> the libraries (the servers and programs remain unfixed) that prevented
> them from compiling with the current gcc toolchain (4.1) that is
> supplied with Debian stable (etch); it only worked with gcc 2.95 which
> was "bleeding edge" in 1998-2000 when we were hacking a lot on chaos but
> nowadays is rather old... :-) (FYI, Linux couldn't be compiled with 2.95
> back then, it required the "old" gcc 2.7.2 series, so we were in fact
> "cooler" than they. :P )
>
> Anyway, when all of chaos compiles on a recent Debian installation, I
> think some of the issues below should be looked into. I'm not saying I
> personally will do all of them, not at all -- I have one 10-month old
> daughter and another (soon) 3-year old daughter to take care of, in
> addition to working full time as a software developer... so, my spare
> time is quite limited to say the least.
>
> Alas, this is what I think would be worthwhile pursuing:
>
> * Look into the problem with TCP connections not being ACK:ed properly.
> When we developed the IPv4 server back then, we never got TCP
> connections (from chaos to Linux) working, for some reason. Our SYN
> packets where never ACK:ed properly by Linux. We followed the RFC as far
> as we knew it but something still prevented it to work. This is an
> interesting thing to look into and I think I can gladly volunteer to
> take upon this ask. I will try too look into it more before the day of
> 2010-12-31. :-)
>
> * Create an ISO image for booting the chaos system, loading the servers
> in the proper order and starting up a basic cluido + log console. The
> ISO image can then be burnt onto a CD-RW and played around with simply.
> It would be great if the current chaos code would be tested on as many
> machines as possible since there might be issues with it being old and
> all. (My Core2 Duo caused an infinite hang in storm for example, I've
> fixed it now but there could be other problems with unsupported
> hardware. USB keyboards is an obvious thing, chaos only supports XT/AT
> and PS/2 keyboards for the moment)
>
> * When that has been done, we should also create a vmware Player image
> with chaos included. I don't know but I think we can distribute it all
> as one .exe file -- great for being able to test chaos in the simplest
> possible way.
>
> I think the work on virtualization platforms is one thing that differs
> most between the "early chaos era", around the year 2000, and now,
> around the year 2010. Back then, vmware was new (I remember when I first
> tried it out and it was really a "wow" piece of software). Now, vmware
> is even available as "free beer" (vmware server) and there are other
> options too -- VirtualBox is a great example that I use at work. Bochs
> also works well for testing out chaos, so a Linux inside vmware + Bochs
> inside that Linux is probably the easiest way to set up a chaos
> development environment. Which leads to the next point:
>
> * Create a vmware Player Image with a chaos development environment, set
> up using Debian etch (or whatever is current at that time) with bochs
> set up inside the Linux machine. Ideally, this should be done so you can
> keep the image on a USB memory or similar (so you can bring along a full
> chaos development environment wherever you go, at work for example :-)
>
> Over and out from me for now. Any spontaneous comments/flames?
> --
> Best regards,
> Per Lundberg
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2005.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> chaos-devel mailing list
> cha...@li...
> https://lists.sourceforge.net/lists/listinfo/chaos-devel
>

Hi guys, and a happy new 2008 to all of you!

(All of the rest of the email is referring to "chaos-old" -- the 
original (and best? ;) chaos, the one that was most "feature-rich" at 
one time. What happened to chaos.net you might ask? Well, development 
stopped quite early after some days of intensive hacking in early 2006, 
partially because Henrik Hallin was attacked by a malicious stranger in 
his home city. He can tell you more about this if he likes to but 
nevertheless, his part of the development was never really taken up 
again after the accident. As far as I was concerned, I didn't really 
have very much spare time to spend on the project.)

As some of you (probably only really Henrik since he's the only one 
reading that list ;-) might has noticed, I've done some basic 
development on chaos lately. Nothing really new and fancy, but what I've 
done is tried and fix some obvious problems with the storm kernel and 
the libraries (the servers and programs remain unfixed) that prevented 
them from compiling with the current gcc toolchain (4.1) that is 
supplied with Debian stable (etch); it only worked with gcc 2.95 which 
was "bleeding edge" in 1998-2000 when we were hacking a lot on chaos but 
nowadays is rather old... :-) (FYI, Linux couldn't be compiled with 2.95 
back then, it required the "old" gcc 2.7.2 series, so we were in fact 
"cooler" than they. :P )

Anyway, when all of chaos compiles on a recent Debian installation, I 
think some of the issues below should be looked into. I'm not saying I 
personally will do all of them, not at all -- I have one 10-month old 
daughter and another (soon) 3-year old daughter to take care of, in 
addition to working full time as a software developer... so, my spare 
time is quite limited to say the least.

Alas, this is what I think would be worthwhile pursuing:

* Look into the problem with TCP connections not being ACK:ed properly. 
When we developed the IPv4 server back then, we never got TCP 
connections (from chaos to Linux) working, for some reason. Our SYN 
packets where never ACK:ed properly by Linux. We followed the RFC as far 
as we knew it but something still prevented it to work. This is an 
interesting thing to look into and I think I can gladly volunteer to 
take upon this ask. I will try too look into it more before the day of 
2010-12-31. :-)

* Create an ISO image for booting the chaos system, loading the servers 
in the proper order and starting up a basic cluido + log console. The 
ISO image can then be burnt onto a CD-RW and played around with simply. 
It would be great if the current chaos code would be tested on as many 
machines as possible since there might be issues with it being old and 
all. (My Core2 Duo caused an infinite hang in storm for example, I've 
fixed it now but there could be other problems with unsupported 
hardware. USB keyboards is an obvious thing, chaos only supports XT/AT 
and PS/2 keyboards for the moment)

* When that has been done, we should also create a vmware Player image 
with chaos included. I don't know but I think we can distribute it all 
as one .exe file -- great for being able to test chaos in the simplest 
possible way.

I think the work on virtualization platforms is one thing that differs 
most between the "early chaos era", around the year 2000, and now, 
around the year 2010. Back then, vmware was new (I remember when I first 
tried it out and it was really a "wow" piece of software). Now, vmware 
is even available as "free beer" (vmware server) and there are other 
options too -- VirtualBox is a great example that I use at work. Bochs 
also works well for testing out chaos, so a Linux inside vmware + Bochs 
inside that Linux is probably the easiest way to set up a chaos 
development environment. Which leads to the next point:

* Create a vmware Player Image with a chaos development environment, set 
up using Debian etch (or whatever is current at that time) with bochs 
set up inside the Linux machine. Ideally, this should be done so you can 
keep the image on a USB memory or similar (so you can bring along a full 
chaos development environment wherever you go, at work for example :-)

Over and out from me for now. Any spontaneous comments/flames?
--
Best regards,
Per Lundberg

Johannes Lundberg wrote:

Hello,

Pretty cool idea. :-) Are we speaking Quake 1, 2 or 3? I presume Quake 1 
for the rest of this email.

I think that would actually be an awesome objective. It is much easier 
to work with something when there is a clear goal, an objective of where 
you are heading. Porting Quake and make it run in 320x240 resolution 
should really be a piece of cake.

What's needed is:
- network support - we already have working IPv4 + UDP, Quake only uses 
UDP as far as I know. Of course, the IP stack is not feature-rich enough 
yet to support everything Quake needs but the foundation is there.
- graphics support. 320x240 works (worked) and Quake 1 would work in 
this resolution with no problems. We could try to have a VESA 
framebuffer and 640x480 (or higher) resolutions as well. OpenGL or 
similar is obviously much more complex.
- sound support. SB16 used to work, we would have to hack it up to work 
with something more modern too.
- file system support. We had VFS working + initial ramdisk and 
FAT/Minix. We could do it by just loading Quake in the initial ramdisk 
but better off would be ISO 9660 filesystem support so we can read the 
files from a CD-ROM. (or DVD?) I think we have floppy code as well so if 
it would fit on a 1,44 meg disk (obviously a Quake rip with only 1 level 
or something like that), that could be used.
- mouse support. We had USB support as well, at least probing the 
devices... oh no, wrong, it was PCI probing that worked. Or perhaps even 
USB. Anyway, we would need to have OHCI + USB HID support to be able to 
read mouse events from the mouse.
- keyboard support. This one is a piece of cake.

When we have all of this, what we need is the equivalent of GNU libc to 
be able for Quake to do what it wants. We can do it the easy way and 
just start working on exactly the POSIX/ISO functions needed by Quake 
and nothing else. ;-) The simplest way is to write stubs first and then 
retry the Quake compile over and over again, to see new functions 
needed. And then, start implementing them.

The question is just, how do we line this up with the recent (well...) 
development in the area of chaos.net...? Henrik?

Best regards,
Per Lundberg

> Heh.
> I am hereby challenging you on a game of Quake running on chaos 2010!
> And well, the odds are a lot in my favour, but it will a fun game indeed. :)
>
> Regards,
>   Johannes Lundberg
>
> On 8/22/07, Per Lundberg <pe...@ha...> wrote:
>   
>> Regarding the chaos development, things have been a bit slow the last
>> months (no kidding???) but what I thought about tonight would be to
>> register the site chaos2010.org. That would give us some time to prepare
>> for the all-stunning, wonderful launch of chaos 2010, the KILLER x64
>> operating system that works in systems with barely 1 gig of RAM! (: And
>> only requires 100 gigs of hard drive space...
>>     
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc.
> Still grepping through log files to find problems?  Stop.
> Now Search log events and configuration files using AJAX and a browser.
> Download your FREE copy of Splunk now >>  http://get.splunk.com/
> _______________________________________________
> chaos-devel mailing list
> cha...@li...
> https://lists.sourceforge.net/lists/listinfo/chaos-devel
>   

Heh.
I am hereby challenging you on a game of Quake running on chaos 2010!
And well, the odds are a lot in my favour, but it will a fun game indeed. :)

Regards,
  Johannes Lundberg

On 8/22/07, Per Lundberg <pe...@ha...> wrote:
> Regarding the chaos development, things have been a bit slow the last
> months (no kidding???) but what I thought about tonight would be to
> register the site chaos2010.org. That would give us some time to prepare
> for the all-stunning, wonderful launch of chaos 2010, the KILLER x64
> operating system that works in systems with barely 1 gig of RAM! (: And
> only requires 100 gigs of hard drive space...

Hello!

I don't know if anyone else has noticed but I have... those pesky
spammers of the Internet had bloated the chaos web site so some pages
were full of porn links etc... :-( I have cleaned up the mess and made
sure the current site is checked in to the CVS repository.

For security reasons, I have set an edit password on the wiki. The
password is: chaos4ever. You enter it under the Preferences (in the
Administrator Password field) if you want to be able to edit stuff.

Hopefully, this will keep us from similar problems in the future. I kind
of like the idea of a wiki, it is innovative collaborative thing. :-)

Regarding the chaos development, things have been a bit slow the last
months (no kidding???) but what I thought about tonight would be to
register the site chaos2010.org. That would give us some time to prepare
for the all-stunning, wonderful launch of chaos 2010, the KILLER x64
operating system that works in systems with barely 1 gig of RAM! (: And
only requires 100 gigs of hard drive space...

Best regards,
Per Lundberg

Hi guys,

This is awesome... take a look at this:

http://en.wikipedia.org/wiki/Windows_PowerShell

It seems to be much like what has been discussed on this list
previously: the GUIs (administrative ones for Exchange Server 2007 etc)
are built on top of the shell, just like Johannes has been proposing.

Also, the idea of instead of piping text (like in Unix and compatible
shells) you pipe objects from one command to another... simply awesome.
Microsoft has done it again, and I dare to say that nowadays, the
technically superior system is not the open source/free software branch
with BSD and GNU/Linux, but rather Microsoft and it's .NET technology.
--
Best regards,
Per Lundberg

Henrik Hallin-UU wrote:

> I have included themeing into my planning of tornado, so you guys need 
> not worry about your 31337 skins not working in tornado. ;)

Good... I almost thought you forgot about that essentiality. ;-)
-- 
"stormG3 is obsolete"

Best regards,
Per Lundberg

Daniel Andersson wrote:

Hi Daniel,

> without being a os developer nor a chaos(.net) developer, but having 
> experienced both the amiga-way (may it rest in peace!) df0:, warez: etc, 
> the xp-way c:, d: etc and the unix-way, i would opt for the unix-way 
> since it offers the most flexibility

It might be so. Perhaps we could mix them and have:

/ - the root
// - the virtual root
//disks/etc - partition in the virtual root

...and be able to assign //disks/0/partition/2 to both /home/ull or the 
labeled way //ull. //ull would be accessible from the VFS but not a part 
of the "root filesystem" like in traditional Unix.

This would give the greatest flexibility, it might be a bit confusing to 
the lamers/newbies though. :-)

> and then you have the question of how you actually mount it
> 
> do you do
> //disks/0/partition/2.mount /warez   (OO?)
> or
> mount //disks/0/partition/2 /warez

I think the latter one is just confusing. :-)

> everything in the fs being an object is an interesting thought..

I think version-control on everything is a cool idea, can be implemented 
with Subversion fairly easily. It will make things a lot slower but the 
cool part about it is that if the system is cracked, you can just revert 
to the previous version of that rootkit-replaced binary. ;-)

Obviously, a serious cracker would change the versioning system as well. 
But if chaos.net is done in a proper way, cracking it will be much 
harder than a Unix or Windows system because privileges will not be 
escalated as easily (there will be no root account). Of course, in the 
practical reality, the sysadmin will probably set up a root-like account 
anyhow. :-P
-- 
"stormG3 is obsolete"

Best regards,
Per Lundberg

without being a os developer nor a chaos(.net) developer, but having =20
experienced both the amiga-way (may it rest in peace!) df0:, warez: =20
etc, the xp-way c:, d: etc and the unix-way, i would opt for the unix-=20=

way since it offers the most flexibility

and then you have the question of how you actually mount it

do you do
//disks/0/partition/2.mount /warez   (OO?)
or
mount //disks/0/partition/2 /warez

everything in the fs being an object is an interesting thought..

/ d

On 4 Apr 2007, at 17:00, Per Lundberg wrote:

> Tjena!
>
> (Sorry for you internationals about this mail being in Swedish, I =20
> started
> writing it to Henrik personally but I post it to the list as well. If
> someone replies to the list, please reply in English at least or =20
> better yet,
> translate my mail altogether into English)
>
> Tycker detta =E4r coolt, vi kan satsa p=E5 att n=E4sta release (0.0.3) =
=20
> har en
> minimal version av Tornado med en liten knapp man kan trycka p=E5 som =20=

> g=F6r att
> det h=E4nder n=E5got eller s=E5...
>
> Sedan m=E5ste man ju ha n=E5gon windowmanager eller liknande och det =
=E4r =20
> ju lite
> mer jobb. Men det k=E4nns coolt detta! Vi borde =E4ven satsa p=E5 VFS =
och =20
> HostFS
> s=E5 att man kan s=E4tta en fixed katalog (typ C:\chaos) att vara en =20=

> virtuell
> chaos-root.
>
> Jag har en del tankar om hur VFS skall funka, f=E5r f=F6rs=F6ka rota =
fram =20
> lite tid
> n=E5gon g=E5ng och speca hur jag t=E4nker. Men vi hade ju en tanke om =20=

> detta f=F6rut:
>
> // - virtuell root
> //disks/0/partitions/0 - ett filsystem
>
> mount //disks/0/partitions/0 / - assigna den som rootpartition.
>
> Men faktum =E4r att man skulle ocks=E5 kunna fundera p=E5 det h=E4r =
med
> enhetsbokst=E4ver. Eller snarare enhetsnamn. I Windows =E4r det ju C:, =
=20
> D: osv. I
> XP =E4r det ganska smidigt med hyfsat dynamisk tilldelning av =20
> enhetsbokst=E4ver
> ocks=E5.
>
> Vi skulle kunna g=F6ra p=E5 ett liknande s=E4tt fast s=E4tta s=E5 att:
>
> mount //disks/0/partitions/0 root:/
> mount //disks/0/partitions/1 home:/
> mount //disks/0/partitions/2 waReZ:/
>
> Det =E4r ju ingen sj=E4lvskriven princip att vi m=E5ste ha en =
UNIX-liknande
> filsystemslayout med en root / och allt mountat i ett katalogtr=E4d =20=

> under det.
> Men nackdelen med "labels" liknande detta =E4r f=F6rst=E5s att man =
inte =20
> har en
> rot... :) Eller? Skulle det funka i praktiken?
>
> mvh,
> Per
>
> ----- Original Message -----
> From: <sf...@us...>
> To: <cha...@li...>
> Sent: Wednesday, April 04, 2007 11:36 AM
> Subject: [chaos-svn] SF.net SVN: chaos: [1338] trunk/chaos.net
>
>
>> Revision: 1338
>>          http://chaos.svn.sourceforge.net/chaos/?rev=3D1338&view=3Drev
>> Author:   sf_hal
>> Date:     2007-04-04 02:36:35 -0700 (Wed, 04 Apr 2007)
>>
>> Log Message:
>> -----------
>>
>>
>> Modified Paths:
>> --------------
>>    trunk/chaos.net/Applications/Cluido/Program.cs
>>    trunk/chaos.net/BUGS.txt
>>    trunk/chaos.net/Library/Chaos.Console/Console.cs
>>    trunk/chaos.net/Servers/Server.Tornado/Program.cs
>>    trunk/chaos.net/Servers/Server.Tornado/Server.Tornado.csproj
>>    trunk/chaos.net/Servers/Wrappers/Server.Console/=20
>> ConnectionHandler.cs
>>    trunk/chaos.net/Servers/Wrappers/Server.Console/Console.cs
>>    trunk/chaos.net/Servers/Wrappers/Server.Console/MainWindow.cs
>>    trunk/chaos.net/chaos.suo
>>    trunk/chaos.net/start.bat
>>
>> Added Paths:
>> -----------
>>    trunk/chaos.net/start Server.Tornado.bat
>>
>>
>> This was sent by the SourceForge.net collaborative development =20
>> platform,
>> the world's largest Open Source development site.
>>
>> ---------------------------------------------------------------------=20=

>> ----
>> Take Surveys. Earn Cash. Influence the Future of IT
>> Join SourceForge.net's Techsay panel and you'll get the chance to =20
>> share
>> your
>> opinions on IT & business topics through brief surveys-and earn cash
>> http://www.techsay.com/default.php?=20
>> page=3Djoin.php&p=3Dsourceforge&CID=3DDEVDEV
>> _______________________________________________
>> chaos-svn mailing list
>> cha...@li...
>> https://lists.sourceforge.net/lists/listinfo/chaos-svn
>>
>>
>
> ----------------------------------------------------------------------=20=

> ---
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to =20
> share your
> opinions on IT & business topics through brief surveys-and earn cash
> http://www.techsay.com/default.php?=20
> page=3Djoin.php&p=3Dsourceforge&CID=3DDEVDEV
> _______________________________________________
> chaos-devel mailing list
> cha...@li...
> https://lists.sourceforge.net/lists/listinfo/chaos-devel

Another question one should ask is wether the traditional filesystem=20
concept should be ripped straight off. I was thinking more along the=20
lines of files being objects in a sense. Files can be created, copied,=20
moved, changed, deleted, recovered etc etc. One should be able to=20
identify what files are identical, what the files' histories are etc=20
etc. But obviously a tree hierarchy makes a lot of sense...

I am currently implementing some test drawing stuff in the Console=20
library, the console server and adding test commands to cluido. After we=20
have some working base code here, I will start hacking at a minimal=20
tornado setup.

I have included themeing into my planning of tornado, so you guys need=20
not worry about your 31337 skins not working in tornado. ;)

Henrik

Per Lundberg skrev:
> Tjena!
>=20
> (Sorry for you internationals about this mail being in Swedish, I start=
ed=20
> writing it to Henrik personally but I post it to the list as well. If=20
> someone replies to the list, please reply in English at least or better=
 yet,=20
> translate my mail altogether into English)
>=20
> Tycker detta =E4r coolt, vi kan satsa p=E5 att n=E4sta release (0.0.3) =
har en=20
> minimal version av Tornado med en liten knapp man kan trycka p=E5 som g=
=F6r att=20
> det h=E4nder n=E5got eller s=E5...
>=20
> Sedan m=E5ste man ju ha n=E5gon windowmanager eller liknande och det =E4=
r ju lite=20
> mer jobb. Men det k=E4nns coolt detta! Vi borde =E4ven satsa p=E5 VFS o=
ch HostFS=20
> s=E5 att man kan s=E4tta en fixed katalog (typ C:\chaos) att vara en vi=
rtuell=20
> chaos-root.
>=20
> Jag har en del tankar om hur VFS skall funka, f=E5r f=F6rs=F6ka rota fr=
am lite tid=20
> n=E5gon g=E5ng och speca hur jag t=E4nker. Men vi hade ju en tanke om d=
etta f=F6rut:
>=20
> // - virtuell root
> //disks/0/partitions/0 - ett filsystem
>=20
> mount //disks/0/partitions/0 / - assigna den som rootpartition.
>=20
> Men faktum =E4r att man skulle ocks=E5 kunna fundera p=E5 det h=E4r med=
=20
> enhetsbokst=E4ver. Eller snarare enhetsnamn. I Windows =E4r det ju C:, =
D: osv. I=20
> XP =E4r det ganska smidigt med hyfsat dynamisk tilldelning av enhetsbok=
st=E4ver=20
> ocks=E5.
>=20
> Vi skulle kunna g=F6ra p=E5 ett liknande s=E4tt fast s=E4tta s=E5 att:
>=20
> mount //disks/0/partitions/0 root:/
> mount //disks/0/partitions/1 home:/
> mount //disks/0/partitions/2 waReZ:/
>=20
> Det =E4r ju ingen sj=E4lvskriven princip att vi m=E5ste ha en UNIX-likn=
ande=20
> filsystemslayout med en root / och allt mountat i ett katalogtr=E4d und=
er det.=20
> Men nackdelen med "labels" liknande detta =E4r f=F6rst=E5s att man inte=
 har en=20
> rot... :) Eller? Skulle det funka i praktiken?
>=20
> mvh,
> Per
>=20
> ----- Original Message -----=20
> From: <sf...@us...>
> To: <cha...@li...>
> Sent: Wednesday, April 04, 2007 11:36 AM
> Subject: [chaos-svn] SF.net SVN: chaos: [1338] trunk/chaos.net
>=20
>=20
>> Revision: 1338
>>          http://chaos.svn.sourceforge.net/chaos/?rev=3D1338&view=3Drev
>> Author:   sf_hal
>> Date:     2007-04-04 02:36:35 -0700 (Wed, 04 Apr 2007)
>>
>> Log Message:
>> -----------
>>
>>
>> Modified Paths:
>> --------------
>>    trunk/chaos.net/Applications/Cluido/Program.cs
>>    trunk/chaos.net/BUGS.txt
>>    trunk/chaos.net/Library/Chaos.Console/Console.cs
>>    trunk/chaos.net/Servers/Server.Tornado/Program.cs
>>    trunk/chaos.net/Servers/Server.Tornado/Server.Tornado.csproj
>>    trunk/chaos.net/Servers/Wrappers/Server.Console/ConnectionHandler.c=
s
>>    trunk/chaos.net/Servers/Wrappers/Server.Console/Console.cs
>>    trunk/chaos.net/Servers/Wrappers/Server.Console/MainWindow.cs
>>    trunk/chaos.net/chaos.suo
>>    trunk/chaos.net/start.bat
>>
>> Added Paths:
>> -----------
>>    trunk/chaos.net/start Server.Tornado.bat
>>
>>
>> This was sent by the SourceForge.net collaborative development platfor=
m,=20
>> the world's largest Open Source development site.
>>
>> ----------------------------------------------------------------------=
---
>> Take Surveys. Earn Cash. Influence the Future of IT
>> Join SourceForge.net's Techsay panel and you'll get the chance to shar=
e=20
>> your
>> opinions on IT & business topics through brief surveys-and earn cash
>> http://www.techsay.com/default.php?page=3Djoin.php&p=3Dsourceforge&CID=
=3DDEVDEV
>> _______________________________________________
>> chaos-svn mailing list
>> cha...@li...
>> https://lists.sourceforge.net/lists/listinfo/chaos-svn
>>
>>
>=20
> -----------------------------------------------------------------------=
--
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share=
 your
> opinions on IT & business topics through brief surveys-and earn cash
> http://www.techsay.com/default.php?page=3Djoin.php&p=3Dsourceforge&CID=3D=
DEVDEV
> _______________________________________________
> chaos-devel mailing list
> cha...@li...
> https://lists.sourceforge.net/lists/listinfo/chaos-devel

--=20
Henrik Hallin
Casper Minisystem AB
Tel. kontor: 018-10 00 18
Tel. direkt: 0707-958 959

Tjena!

(Sorry for you internationals about this mail being in Swedish, I started 
writing it to Henrik personally but I post it to the list as well. If 
someone replies to the list, please reply in English at least or better yet, 
translate my mail altogether into English)

Tycker detta är coolt, vi kan satsa på att nästa release (0.0.3) har en 
minimal version av Tornado med en liten knapp man kan trycka på som gör att 
det händer något eller så...

Sedan måste man ju ha någon windowmanager eller liknande och det är ju lite 
mer jobb. Men det känns coolt detta! Vi borde även satsa på VFS och HostFS 
så att man kan sätta en fixed katalog (typ C:\chaos) att vara en virtuell 
chaos-root.

Jag har en del tankar om hur VFS skall funka, får försöka rota fram lite tid 
någon gång och speca hur jag tänker. Men vi hade ju en tanke om detta förut:

// - virtuell root
//disks/0/partitions/0 - ett filsystem

mount //disks/0/partitions/0 / - assigna den som rootpartition.

Men faktum är att man skulle också kunna fundera på det här med 
enhetsbokstäver. Eller snarare enhetsnamn. I Windows är det ju C:, D: osv. I 
XP är det ganska smidigt med hyfsat dynamisk tilldelning av enhetsbokstäver 
också.

Vi skulle kunna göra på ett liknande sätt fast sätta så att:

mount //disks/0/partitions/0 root:/
mount //disks/0/partitions/1 home:/
mount //disks/0/partitions/2 waReZ:/

Det är ju ingen självskriven princip att vi måste ha en UNIX-liknande 
filsystemslayout med en root / och allt mountat i ett katalogträd under det. 
Men nackdelen med "labels" liknande detta är förstås att man inte har en 
rot... :) Eller? Skulle det funka i praktiken?

mvh,
Per

----- Original Message ----- 
From: <sf...@us...>
To: <cha...@li...>
Sent: Wednesday, April 04, 2007 11:36 AM
Subject: [chaos-svn] SF.net SVN: chaos: [1338] trunk/chaos.net

> Revision: 1338
>          http://chaos.svn.sourceforge.net/chaos/?rev=1338&view=rev
> Author:   sf_hal
> Date:     2007-04-04 02:36:35 -0700 (Wed, 04 Apr 2007)
>
> Log Message:
> -----------
>
>
> Modified Paths:
> --------------
>    trunk/chaos.net/Applications/Cluido/Program.cs
>    trunk/chaos.net/BUGS.txt
>    trunk/chaos.net/Library/Chaos.Console/Console.cs
>    trunk/chaos.net/Servers/Server.Tornado/Program.cs
>    trunk/chaos.net/Servers/Server.Tornado/Server.Tornado.csproj
>    trunk/chaos.net/Servers/Wrappers/Server.Console/ConnectionHandler.cs
>    trunk/chaos.net/Servers/Wrappers/Server.Console/Console.cs
>    trunk/chaos.net/Servers/Wrappers/Server.Console/MainWindow.cs
>    trunk/chaos.net/chaos.suo
>    trunk/chaos.net/start.bat
>
> Added Paths:
> -----------
>    trunk/chaos.net/start Server.Tornado.bat
>
>
> This was sent by the SourceForge.net collaborative development platform, 
> the world's largest Open Source development site.
>
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share 
> your
> opinions on IT & business topics through brief surveys-and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> chaos-svn mailing list
> cha...@li...
> https://lists.sourceforge.net/lists/listinfo/chaos-svn
>
> 

After some more days of hacking, we are proud to release the second 
pre-alpha release of chaos.net (the re-engineered chaos using the .NET 
architecture, written in C#).

Most important visible change is that Cluido + the rest of the system 
has now been improved to the point where you can type commands in Cluido 
and make the system do something. This is a great leap forward and it 
also makes debugging the system tremendeously simpler!

Get it at http://prdownloads.sf.net/chaos/chaos.net-0.0.2.zip (source)
or http://prdownloads.sf.net/chaos/chaos.net-0.0.2-bin.zip (binary)
-- 
"stormG3 is obsolete"

Best regards,
Per Lundberg

On 3/28/07, Henrik Hallin-M1 <he...@ca...> wrote:

> *) Cluido now reads commands and echoes them.

Two words: way cool! This means we can make Cluido do something useful. :-)

Great work, Henrik! I'll test the code ASAP.
--
Best regards,
Per Lundberg

-- 
Best regards,
Per Lundberg

Hi,

I've tried to run the chaos.net stuff on mono and partly succeeded. The 
console server appears to run, and cluido also starts up (an exception 
is fired because it can't connect to storm.net). Storm.NET does not 
start however, and no exception is thrown. There might be some debugging 
functionality in mono, but I haven't tried it...

I suspect the reason is that storm.net uses richtextbox which doesn't 
seem to be implemented in their Windows.Forms-stuff. Since there is 
nothing "rich" about the way i use the rich text box, I could probably 
use some other control and make it work on mono.

This is cool, since it allows people to work on chaos.net regardless of 
operating system.

Henrik

-- 
Henrik Hallin
Casper Minisystem AB
Tel. kontor: 018-10 00 18
Tel. direkt: 0707-958 959

Hi everyone,

For those of you who aren't reading the svn list, here's a short update:

*) The IPC mechanism has been updated and reworked a couple of times. :)

*) An event concept has been added for servers to use to notify their 
clients about state changes (this is used for keyboard events by the 
console server).

*) The console server now supports multiple consoles with keyboard event 
handling.

*) Cluido now reads commands and echoes them.

What needs to be done I think is:

*) Improve stability of the whole system, e.g. allow clients and servers 
to die without storm dying too. Allow clients to exit cleanly if the 
service goes away. Allow servers to correctly dispose of resources 
associated with a client when the client goes away. In short, make 
things not crash. :)

*) Improve cluido and make it do something useful. I think we will need 
some kind of TTY server in the middle somewhere. But that can come 
later. First we need VFS, but that can come later as well.

*) Discuss libraries and their structure.

*) Security.

*) Fullscreen and graphics drawing support in the console server/library 
+ tornado! :)

-- 
Henrik Hallin
Casper Minisystem AB
Tel. kontor: 018-10 00 18
Tel. direkt: 0707-958 959

On 3/23/07, Anders =D6hrt <and...@gm...> wrote:
> One of the main issues I had with chaos last time was it took so long
> to set everything up so that once I could start development and
> booting, I'd already gotten tired of it. This should help with this
> issues very well.

This can not be said too much. We should not underestimate the
importance of having a very straight forward way of getting started.
And not just for getting chaos up and running, this should be for a
working development environment too.

I think the biggest reason for the success of PHP is it's ease of
getting started. Create a file and write code.. No more, no less.

Regards,
  Johannes

per...@us... wrote:

Hi Henrik et al,

IPC now works again, it seems. :-) However, the question has arised (if 
I understand you correctly Henrik) as to whether our connections (which 
are similar to named pipes in Unix/Windows) should be unidirectional or 
bidirectional.

What I mean is this: you can have the connection work in such a way that 
the initiator of a request is always the client. And the server handles 
the request and possibly sends something back to the client, depending 
on the type of the request.

Perhaps I'm misunderstand you Henrik but you seem to be proposing that 
the IPC should work in another way than this? Perhaps what you are 
thinking about is asynchronous calls, where the server needs to initiate 
the call back to the client.

What I've also realized that we should have also is something akin to 
Unix/Posix signals.

Signals are a way for the operating system to notify a process that 
something unpredicted happened. For example, that the process is 
terminating. :-) (SIGTERM) Another useful signal is SIGHUP, which is 
sent if the modem line is being hangup (or if Putty is closed, I 
presume, in a modern scenario).

We could implement signals using the service system. By default, 
Storm.NET has default handlers for all "signals". But, you can create a 
service with a certain class name (signal) and you are then assigned the 
service path /signal/<pid> which Storm.NET and other processes in the 
system can send requests to (think "kill 12345" and similar).

> Revision: 1284
>           http://chaos.svn.sourceforge.net/chaos/?rev=1284&view=rev
> Author:   perlundberg
> Date:     2007-03-24 14:28:52 -0700 (Sat, 24 Mar 2007)
> 
> Log Message:
> -----------
> Fixed the IPC to work since Henrik is too lazy to do it himself. ;-)
> 
> I think this is similar to how you intended to solve it, right?
> 
> Modified Paths:
> --------------
>     trunk/chaos.net/Chaos.IPC/IPCMessage.cs
>     trunk/chaos.net/Chaos.IPC/IPCMessageArgument.cs
>     trunk/chaos.net/Chaos.Service/ServiceConnection.cs
>     trunk/chaos.net/Chaos.Service/StormConnection.cs
>     trunk/chaos.net/Servers/Server.Console/ConnectionHandler.cs
>     trunk/chaos.net/Storm.NET/ClientConnection.cs
>     trunk/chaos.net/Storm.NET/Connection.cs
> 
> Added Paths:
> -----------
>     trunk/chaos.net/release.sh
> 
> Property Changed:
> ----------------
>     trunk/chaos.net/Chaos.Console/
> 
> 
> This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
> 
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share your
> opinions on IT & business topics through brief surveys-and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> chaos-svn mailing list
> cha...@li...
> https://lists.sourceforge.net/lists/listinfo/chaos-svn
> 

-- 
"stormG3 is obsolete"

Best regards,
Per Lundberg

Johannes Lundberg wrote:

> I'm not saying this is easily done, but it would certainly be
> interesting to try out. :)

Well, feel free to implement it. :-) I also think it is a very 
interesting idea and in line with the chaos philosophy of being kind of 
"revolutionary" in our ideas. There is no point in doing a new operating 
system just to emulate another operating system, too many people have 
already done that. (Hello, Linus, nice to see you're sneak-peaking at 
what we're doing. ;)
-- 
"stormG3 is obsolete"

Best regards,
Per Lundberg

On 3/22/07, Per Lundberg <pe...@gm...> wrote:
> > I'd like this to be combined with some sort of memory-presistance too.
> > Kind of a new philosophy for building applications. Instead of writing
> > applications that create everything (memory-wise) from scratch
> > everytime you start the application (like reading the configuration
> > file, loading icons etc), you do the setup once and then just resume
> > from a post-setup state when a user starts the application. I talked
> > quite a lot with Per about this a year ago or so. If you do an strace
> > of a small Linux shell utility, there is a lot going on under the hood
> > that could be done a more effective manner.
>
> I remember this and I definitely think it's something that ought to be
> considered further. There might be disadvantages with that approach but
> if we can make it work, we can probably have applications launching
> much, much faster on chaos than on any existing operating system in use
> today - which is good. I think Windows XP bootup on my machine is too
> slow, for example, and part of the reason is that all the startup
> applications need to be ready before you can do much anything.

I think such an approach would work very well with todays disks too.
High throughput but high seek times. Instead of doing 35 seeks to open
all libraries, config files etc, you just do one seek and sequential
read to get the application going. This could be implemented by having
startup-methods on the applications, which are called dynamically the
first time, if the application havn't been run before.

One concern though is how to do dependancies. When system or file
changes are done, which invalidate the "startup state memory image"
(or whatever it should be called), it needs to be flagged properly as
outdated, and recreated on the next run.

One cool thing would be if it would be possible to combine a
system-wide startup image with a user-specific startup image (maybe
for his configuration etc). With .NET/CLR, I think this could be
doable. Basically, the system-wide image have references to the
user-specific image (and a diffrent image / "collection of objets" are
loaded for each user).

I'm not saying this is easily done, but it would certainly be
interesting to try out. :)

Regards,
  Johannes Lundberg

On 3/23/07, Anders =D6hrt <and...@gm...> wrote:

Hi Anders & welcome back! :)

> I'm impressed! I like the direction the project has takes, which feels
> like an exciting development. I've not looked closed into C#, since I
> now believe everyone should switch to Visual DataFlex (shameless
> promotion: http://www.visualdataflex.com/features.asp?pageid=3D679), but
> it's interesting non the less.

C# is really a great programming language to work with, methinks (and
I think Henrik Hallin agrees with me strongly). It has everything we
need and few negative aspects. One is possibly that you cannot specify
default parameters:

class Test
{
  public int Test123(int i =3D -1)
  {
  }
}

...that would be invalid C# code. But you can achieve the exact same
effect using method overloading (making one method with no parameters
in this case and let that method call the other) so it's not really a
problem... They probably removed it to make the code we write less
obscure. :-)

> One of the main issues I had with chaos last time was it took so long
> to set everything up so that once I could start development and
> booting, I'd already gotten tired of it. This should help with this
> issues very well.

It sure does. I mean, developing the system on Microsoft Windows
(rather than GNU/Linux) is a very good thing since it makes it much
easier for most people to try out the system and perhaps start hacking
around.

Please comment my other email about Chaos.Security if you like,
perhaps you have some valuable comments to add?
--=20
Best regards,
Per Lundberg

> After some days of fairly intensive hacking evenings, we are proud to
> release the first "version" of chaos.net (the re-engineered chaos using
> the .NET architecture, written in C#). Eventually, this system will be
> self-hosted on top of the storm kernel, but until we are ready for that,
> we will develop the system as a set of applications communicating with a
> fake kernel (Storm.NET) running as a user process on top of the .NET
> runtime.

I'm impressed! I like the direction the project has takes, which feels
like an exciting development. I've not looked closed into C#, since I
now believe everyone should switch to Visual DataFlex (shameless
promotion: http://www.visualdataflex.com/features.asp?pageid=679), but
it's interesting non the less.

One of the main issues I had with chaos last time was it took so long
to set everything up so that once I could start development and
booting, I'd already gotten tired of it. This should help with this
issues very well.

// Anders

Henrik Hallin-UU wrote:

Hi Henrik,

> What we can start doing now is to implement a real shell (Cluido). This 
> ought to easy debugging of servers. Anyone up for the task?

Johannes volunteered and I might also be able to help with some bits there.

The Chaos.Security is also something that needs to be implemented soon, 
as you wrote in another email. As I mentioned to you, we should make 
Storm.NET launch a set of servers on startup (and give them the 
capabilities it feels like). Then the Server.Boot should launch Cluido, 
dropping capabilities as needed.

The capability system can be fairly easy but flexible. Perhaps just a 
set of regular expressions or something. The important point is: 
Storm.NET has full capabilities (.* in regexp talk). When launching a 
new server, it can set any capability less than or equal to its own 
capabilities on the new process.

Likewise when the Server.Boot will launch Cluido, it can give it less 
than or equal to its own capabilities.

If we go the regexp road (maybe a bit overkill really) I think the "less 
than or equal" formula might be a bit hard to implement, to put it 
mildly. ;)

Perhaps a form of simpler globbing is more suitable:

Service.* - access to all services.
Service.Console - access to a specific service.
Storm.CreateService - access to the specific system call in Storm.NET.

The problem here is not the implementation really but rather to find a 
suitable level of fine-grainedness. If you make it too specific, it 
might get over-complicated to add capabilities to a user etc. (you need 
to find all the specific capabilities that he needs - can obviously be 
partially remedied by having capability groups, preferrably with a form 
of inheritance so that a capability group can contain other capability 
groups). If you make it too non-specific, we have reimplemented Linux, 
welcome to the 70's again. :-P

Back to where we are now: As soon as the IPC works again :) (it is 
currently broken in SVN but you are already well aware of this) I think 
we release chaos.net 0.0.2, to make sure people see that things are 
proceeding quickly in the right direction.

Now, bedtime...
-- 
"stormG3 is obsolete"

Best regards,
Per Lundberg

2006	Jan	Feb	Mar (1)	Apr (139)	May (47)	Jun (9)	Jul (21)	Aug (6)	Sep (2)	Oct	Nov (1)	Dec
2007	Jan	Feb	Mar (53)	Apr (5)	May	Jun (1)	Jul	Aug (3)	Sep	Oct	Nov	Dec
2008	Jan (3)	Feb	Mar	Apr	May	Jun (1)	Jul	Aug	Sep	Oct	Nov	Dec
2017	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct (1)	Nov	Dec

chaos-devel Mailing List for The chaos Operating System

chaos-devel — chaos {operating system} developers list