Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - Roderick Klein

Pages: 1 ... 43 44 [45] 46 47 ... 49
661
Hardware / Re: Motherboard woes
« on: December 25, 2014, 02:34:46 pm »
I settled on the MS! Board.  Set the Bios to Legacy+EFUI (no secure boot option available). and tried to boot OS/2.  Got as far as the recovery screen, pushed F5 and moments later got the dreaded black screen.  Tried booting ECD 1.2 and 2.1 from the cd drive and got trapped out while loading the files in to ram.  I haven't had any time to play with it but my first thought is that I may have a bad memory stick.  I usually use Corsair brand but Frys was out of stock and had to use Viking which I never heard of.  I bought an extra hard drive and installed Windows 7 so I'd have a computer. (M/S always sends me a GA release of their software. They must think I'm a IT pro). After the first of the year I'll play tinker with it some more. Sure do miss OS/2.  Windoze is a pain, makes me appreciate OS/2 more.

When you provide a posting you realy need to provide more information if you want any assistance. Its not a garentee for a resolution but it helps to increase your changes.
Based on what you write my guess if that your system is crashing because the driver screen01.sys is making a call into the OEMHLP driver thats in the OS2LDR.
In turn the OS2LDR makes a call to the BIOS.... And a lot of modern BIOS'es do not support this call.

Do not worry more device drivers make this call into the OEMHLP function that causes a TRAP. Thsi defect can be fixed by

Installing your eCS 2.1 on an older machine and get the latest ACPI from Arca Noae. Then install that on the system and transfer the the disc to your new system.
I would forget trying to load even eCS 1.2 as the danis506.add is very old (heck that eCS version is 10 years old).

Also the OS2AHCI driver included in eCS 2.1 is pretty old and has bugs...

Roderick Klein

662
Along with the Dell mini 1012 that I was given by my in-laws came a venerable Dell desktop (venerable indeed, it uses PS/2 keyboard and mouse!)

Spec wise I think it would suit eCS well (dual core Celeron, 2GB of RAM, 40GB hard drive), so I had a go at installing eCS 1.2R on that after cleaning out the internals of the copious cobwebs and dust.   It currently has XP on it and I ran a full checkdisk using that - all came back fine

However, when I get to the eCS Create Volume step in the setup, it keeps reporting a corrupt partition 0 and won't let me do any changes, i.e. I cannot create Boot Manager, cannot rewrite MBR, cannot add a new volume

I can boot back into XP and again no errors reported using checkdisk.

I had the same problem with an old Dell laptop a few months back - ran XP fine, Linux installed fine if sluggish, but eCS complained of an error in partition 0

Is this a known eCS 1.2R bug? Is it fixed in 2.1 or is there a way around it? Would a bootable DFSee be the way to go - is there some step I need to do to make a hard drive eCS friendly?  I don't recall any such thing when I last used eCS several years ago, but it could be that I've forgotten

Any advice gratefully received, otherwise Linux will be going on there instead!

PS - I don't get this error when installing under VirtualBox, it only seems to be with real hardware

If you would have put the exact English error discription in google it might have already popped up.


    To work around this installation problem, download DFSee (available at http://www.dfsee.com/dfsee/download.php).

Get dfsos2.exe from the zip-file and put it on a USB-stick (FAT16) or on a floppy. Boot from the eComStation 1.2R CD-ROM and press Shift-F3 again to get a command prompt.

Start dfsos2.exe, press F10 and goto the menu option Mode=Fdisk.

Choose the menu option at the bottom:
  "Manage OS2 LVM information"

Then select:
  "Add default LVM-info (VCU)"

After having executed this command, DFSee will restart the system.

Boot from CD again and start minilvm to create your partitions.

This error occurs because VCU.exe from IBM is not run during installation. VCU is a utility to assign drive letters to all partitions. In some cases, running VCU under MCP1/MCP2 can lead to data loss. In eComStation 2.0 and above this problem is resolved with the installation CD. This checks the hard disc (if there are no problems) it reports in the text mode phase of the boot phase there are no errors or otherwhise offers to resolve the detected problem.

663
Hardware / Re: AMD 8-Core, UEFI and eCS 2.1 a No Go
« on: July 31, 2014, 04:49:05 pm »
With eCS 2.1 you are most likely stuck!

You need to boot the machine in legacy mode and not UEFI.

Second you need to load ACPI. But could try to load with he switches /!NOD.
The problem is its an old version of ACPI that is in eCS 2.1 and the newer version has made a lot of headway!

Also the OS2AHCI driver included in eCS 2.1 its realy the question is it will work...

So I am afraid you need eCS 2.2 beta 2...

Roderick

664
Setup & Installation / Re: Re-organizaing OS/2-eCS Directories.
« on: July 24, 2014, 06:58:34 pm »
Nope, no documentation...

That was the verdict at the time when we looked into this... And the question is if you want to break stuff yes or no...
Roderick, I understand completely your intention not to move OS/2 system dirs. But, from the current POV it might have been a good point of time to do so in the early days of eCS. That would have created a new standard, which would be fully accepted in this time.

Do you have similar thoughts (just out of curiosity)?

Years ago when we worked on eCS 1.1 or 1.2 we looked at the directory structure. Sadly I only remember the conclusion we where wondering if we would not open up Pandora's  box.
We had some cases where we found fixed paths... One example I remember now that has floated around in my head this topic.

Please check:
\ecs\install\rsp\migrate.cmd
Comment: /* remove syslevelfiles to make sure IBM installer doesn't choke */

What I remember is that the peer installer when you have it installed on the boot drive and then do a miigrate of the network installation to another drive the peer installer would blow up.
The reason was it was also checking the BOOT drive. So your bootdrive for example is drive C: and you reinstall networking support to drive D:.
The instaler would allways check boodrive\IBMLAN. While not a path fix its an example of fixed path code left...

Martin already bumped into two problems one with MPTS and the other with UNICODE.SYS. My only point is how many other older applications this could simply break...
Apps  you might not have access to...


665
Setup & Installation / Re: Re-organizaing OS/2-eCS Directories.
« on: July 24, 2014, 06:43:00 pm »

That was the verdict at the time when we looked into this... And the question is if you want to break stuff yes or no...

Yes !!! I'm doing this on a VM, not at someone else home's machine or a business critical environment. I want to experiment and produce some kind documentation (this forum and OS2World Wiki), what is wrong with that?
At least I plan to share the results (failure or not) with the community instead of having it locked in my mind.

What is it wrong with the collaborative spirit? People had tried differnt stuff with OS/2-eCS and never talked about it... and when someone tries it the only thing they say is "NO", without reasons, without documentation, without explaining what went wrong in detail. We need to change that. Use the Wiki, use the forum, this is not twitter, anybody is free to write more than 140 characters.

Go to what I wrote and read it carefully. First of all the thing is you can test this. The only detail is you simply do not have access to all software and scripts to test this. So question is if your tests are sufficient enough to conclude that this directories moving around is something that will work  reliable.

Second as I said you might simply make the LIBPATH & PATH longer and its the question if that is something that is worth it...

The only thing that happens dis that we looked at this years ago at Mensys. And as is not a public surprise at Mensys we did not have a whole team to put conclusions we researched in a wiki of some sort. I simply remember the conclusion and wish I could have remembered the details from about 8 years ago, but thats ife..

So while I want to change it. In this case I can only tell you the details of this if I go back in time and can hear the discussion again

666
Setup & Installation / Re: Re-organizaing OS/2-eCS Directories.
« on: July 22, 2014, 08:24:48 am »
About 8 years ago I looked at this and found about 4 to 5 items that would be broken. Most of the stuff does not use fixed paths. But why run the risk.

Roderick, did you documented this? did you posted somewhere? Please share the link.

Nope, no documentation...

That was the verdict at the time when we looked into this... And the question is if you want to break stuff yes or no...

Roderick

667
Setup & Installation / Re: Re-organizaing OS/2-eCS Directories.
« on: July 22, 2014, 08:22:46 am »
What I want it is not necessary a priority of the rest.  But at least I can try to do it and document what it is getting broken, so in the future it can easy the job of the guys that "really" want to do it.

Nothing can go wrong if we try it on a VM and in a non-production machine.

The point is you can not test it really. Its mostly down to dirty coded apps, sciripts that use hard coded paths. Besides depending on how you move this stuff around you might end up making the path/libpath length even longer... Unless you merge directories...

Roderick

668
Setup & Installation / Re: Re-organizaing OS/2-eCS Directories.
« on: July 21, 2014, 09:08:05 am »
Hi

I still dream of a next version of eCS (or anything else) that improves the directories structure to have a more clean C:\ root. Some years ago I wrote some suggestions like this one (Recommendation #8 – Directory and File Structure ).

I want to know how hard will be to move some folder from the standard structure. I still dream of moving:
- DMISL
- IBMCOM
- IBMGSK
- IBMGSK40
- IBMGSK50
- IBMLAN
- LANGUAGE
- MMOS2
- MPTN
- MUGLIB
- PSFONTS
- TCPIP

Possible a good site for those folder can be under "/OS2/System".

What do you think.? Do you have any experience moving those directories? Which ones can be moved easily and which one of those are hardcoded and can not be place somewhere else?

Regards

I think you asked this question some time ago and from what I remember this was never done in eCS to prevent things from being broken.  Scripts and other programs.
I think this is also one of the last priorities to worry about, the directories in the OS...
And what it could break. About 8 years ago I looked at this and found about 4 to 5 items that would be broken. Most of the stuff does not use fixed paths. But why run the risk.

Again a low priority....

Roderick

669
Setup & Installation / Re: Is eComStation DEAD?
« on: July 03, 2014, 01:47:30 pm »
Hi Martin,

open source is great: in times when things go well, many contribute (and things go even better).  But the truth is that  in the long haul (and the nitty-gritty of tedious work is required), someone must be paid to do the work.  Let us look to Linux and Android. Linux is mostly still run on a voluntary basis, but with important commercial partners/off springs like Suse, RedHat and Mandrake, that all have helped advance Linux.  For Android it is even more clear - Android had not been where it currently is without the vast resources put in by Google.  The Android business model is a bit different from that of Linux -- the "economics" of Android is based on making other software (like Chrome) more palatable, and by lowering costs to consumers for buying Android devices, which in turn increases traffic, including the use of Google's services.

eCS is in a different situation.  There are probably less than 100 K private eCS users left in the world.  Open source routes, like those of Linux or Android, are unlikely to create the revenues needed to fund the boring polishing work (I assume many computer interested people enjoy working (and succeeding) on the fun parts, like cracking problems like ACPI etc.), but documenting and making stuff available for other hardware than one's own, is not fun.  And to undertake "not fun" tasks (or tasks that are not rewarding in other ways), people usually require some pay.  This is where licensing is important. 

Licensing is also the same business model applied by academic societies (where I am member of many, and have been president of one): membership fees cover for costs.  To keep the membership base (and preferably to increase it), the expected benefits of the membership must exceed members' costs.

This is where eCS (Xeu/Mensys) currently struggles.  Too little progress, uncertain deliveries etc. make users doubt the future. For me, and many other eCS users, it would be a loss if eCS were to vanish.  Therefore, we gamble (hoping sufficiently many others also gamble) by paying the modest license  fee to create revenues that cover the costs of keeping eCS operational, and by being a good platform, attract new users.

I wish I were wrong, and that it would be possible to keep eCS for free.  But I doubt it.  Somewhere, there must be some revenues to cover for the boring, but necessary part of making software run on hardware different from ones own. 

Have a thoughtful day
Eirik

I understand Martin his drive for open source. But when it comes to large projects I told him many times it takes time to develop something. And large projects mostly requires paid people.
And do not take my word for it. This guy has presented at LNFW (LinuxFest Northwest) for some years he did so already http://www.youtube.com/watch?v=Sh-cnaJoGCw.
I think it was this video I watched... The presenter mentions somewherre in that video that Linux has big money behind it. If its for the kernel or the different graphic user shells.
Its far from just volunteers that develop code in spare time.

So while open source does a lot for OS/2 and would be dead without it. Open source has primarly allowed a lot of people to work and port stuff from primarly Linux to OS/2.
But the work is mostly (big projects) done by paid or low paid people.

So I understand the will to see stuff have open source as so much nice OS/2 software died because it was closed source. But open source is not a warranty to keep OS/2 alive.
Just like with Linux (which I recently looked into more), more and more money flowed into the product over the years. A lot of critical mass came from large cooperate funding.

As for eCS I can not state much. I like OS/2 a lot and type this message on my laptop running eCS. But as you can see on my linked in profile and have possibly heard I have left Mensys.
I am still president of VOICE and will remain so. I will also be in the forums.

What Mensys is currently doing with eCS I do not know. I have seen messages from Lewis Rosenthal popup he is starting a new company. What will happen from here I do not know.

I am staying tuned to forums and mailing lists.

Roderick Klein

670
Then put in the bugtracker where it belongs. If Pasha has a lot to say.

"Well, maybe, not enough knowledge. How exactly is Pasha going to write a GRADD driver you will know if he gets hired."
You know if something can offer something in the future is the argument to not use Multimac it should be made more clearly in the offer.
Not after you hire somebody.

Thats why I ask. And without further details how this driver frame work can provide video support without DDK/GRADD support I would like to know how this works and I asked.

Boris wrote:
"I always thought that making an open source! hardware-accelerated video driver with 3D support is what makes sense. I'm quite disappointed that a person from Mensys disagrees with me and wants to roll back to multimac..."

I am not against accelerated drivers. But I certainly do not see the benefit of 3D accelerated drivers for PM.

Apart from that David has been working on Multimac more updates can be expected in the next few weeks. Less rewrites on the code from Linux shoudl be needed.
But if Pasha needs 2 months to write the 8169 driver and then move to Wifi driver and then implement security.

Then the scope of this project should also describe how that video support might be provided. The scope if the project is to write a Wifi driver.
And without further details how the video support would be done. This is reinventing the wheel...

And if Pasha has so much information to tell then let him put that in tickets to enhance Multimac.

Roderick 

671
Of the 1 GB drivers are not useable on a LAN then open a bug. I think its better then to completely reinvent the wheel with a new driver skeleton.
So far Mulitimac provided a proper working Gigabit Realtek and Intel driver. And it has also provided updated driver.
The Nvidia driver was also updated.

No Multimac will not provide this video support. But how will this project do that without the DDK as vast portions of the DDK are needed to my knowledge to get it done.
Or is this somehow going to be a GRADD plugin ? Some more details would certainly be interesting.

Its not the point that Pasha got paid for it. Its the question what makes sense.
Reinvent the wheel or use Multimac driver skeleton ?

What I ment with the video driver stuff. How would it work ? How does it hook in the video sub system ? A new screen01.sys etc ?


Roderick

672
Hi

I have another bid for using this funds. Pavel Shtemenko proposed an Network/Wifi BSD driver port project.
The idea is to port the Network/Wifi BSD driver.

The basic idea of this project is:

Stage 1 or Bounty 1:
1. Base network driver
2. Ported from FreeBSD some of DevHlp (this part can also be used for BSD Video driver port too (ATI-NVIDIA) in the future)
3. FreeBSD driver (compiled from source)

Goals: PCI RTL816x driver working
Time: 1.5 Months
Cost: USD $ 1.200.

Stage 2 or Bounty 2:
4. Security subsystem
5. Other Wifi adapters ported
6. Suspend/LowPower support (with ACPI 3.18)

Goals:
- Security system,
- other Wifi adapters (Any WiFi driver which FreeBSD 7.x has to support.r like Atheros WiFi  , some other one)
- Suspend/Low power supported.
Time: 1 Month
Cost: USD $ 800.

Conditions:
- All source code produced most be open source (BSD license)
- The Common Fund and Bounty Rules apply to it.
- Development will be done in the open at sourceforge or github.
- Source code from IBM's DDK can not be used in the project, since it have a restrictive license that does not allow to share the source code or derivate works.
- The driver most works on Warp 4.52 and eComstation 2.1 with their respective kernels.
- He can not accept Paypal, we are trying to see if we can get another alternative.
- Payment will be done after completing each stage.

The idea of this is making it an open development and the source code will be open source for everyone to use.  Since this is an open community and the idea is to have an open conversation about this, please do not send me e-mails in private. Give me you feedback on this forum.

Please state what do you think about this offer.

Regards

I am a bit puzzled by this proposal ?!

Why work on such a skeleton if we already have something that such as Multimac. Its not to port FreeBSD drivers but it is to port Linux drivers:
http://svn.ecomstation.nl/multimac/browser/trunk

The stage 1 of the bounty:
Ported from FreeBSD some of DevHlp (this part can also be used for BSD Video driver port too (ATI-NVIDIA) in the future)
http://svn.ecomstation.nl/multimac/browser/trunk/Porting.txt

While its true it requires a DDK to work on this. How is that going to work the video drivers ?!

Next to that I do not know what is ment to implement "security  subsystem"?
Is that the API for a something like WPA supplicant and and XWLAN widget ?

Apart from that I do not know how mature the current version of the software is. The Multimac skeleton David has written has already provided support for the current Realtek gigabit and Intel gigabit driver. New drivers for Broadcom wired are in the pipeline. While the Mutlimac project has still not provided a wireless driver why with such little human resources reinvent the wheel again if we already have Multimac. And of course its going to take time to get the bugs out of the code...

Roderick


673
Applications / Re: Firefox 24 vs OS/2 World
« on: June 01, 2014, 09:21:10 pm »
I got the new Firefox 24 beta running here. But it's claiming every word I type in the composition box is misspelled. I know I'm not the greatest speller, but how can I correct my mistakes if all of them come up?

Come on Neil you should no better. I did not read the bug list on Gidhub but check if the bug report exits already or open a bug report in the proper place if that has not been done yet.

Roderick Klein

674
This.
Is.
Starting.
To.
Be.
So.
Damn.
Annoying.

How could one write such a messed up installation system?..

Okay, I've tried eCS 2.1 installation disc, which ended up with the same message in the same place.

Then I've figured out this is because of the PCI PATA controller I use. So I've temporarily connected the hard disk drive to the motherboard one. The first stage has finished (I've returned back to 2.2).

And here I already have questions. How exactly did it manage to fail? The target installation drive is ACCESSIBLE from the management console. I can read, write and even reformat it with no apparent problem. No, really, WTF?

Okay, then I've assured myself NOW it's 100% gonna be okay and rebooted the machine from the hard disk drive to finish the installation process. But with eComStation you can never be sure. The second stage has started, then the progress bar has reached the 50% mark. Cool for now. One more minute and it ... no, silly me, it didn't finish the installation, it has JUST SUDDENLY REBOOTED. Just in the middle of something it has INSTANTLY, without any message, started from BIOS POST... what else can I say?

Mensys, HOW!?

Of course, the halfway installed OS/2 was kind of corrupted (although it has booted to the desktop), so I went to the installation disc over again, filled all that forms and checked the needed check boxes, completed installation stage 1 and booted from the HD again. Again just to be sure it will fail to instant reboot the second time in the very f*cking same place.

No comment.


You where kind of on your way...
As Alex suggest check \var\log\base_1.log.

It seems the behavior of the good old IBM RSPINST.EXE blowing up is the same on eCS 2.1 and 2.2.

Your conclusion that this is caused by ACPI is doubtfull as the system is able to boot from CD ROM with ACPI!
What is more likely the cause is that RSPINST crashes half way and you end up with a half installed OS/2 base operating system.

Roderick

 

675
General Discussion / Re: Sources of DBExpert?
« on: May 22, 2014, 10:59:34 am »
When Rendall Flint was still alive I talked to him about source code for DB expert. Some things make it very difficult. Right now I know from people who know Carla that its better to leave Carla alone.
Dispite the fact its a few years ago Randall died its emotional for here. At least for this 2014 you should not count of getting the sources for DBexpert.

Pages: 1 ... 43 44 [45] 46 47 ... 49