• Welcome to OS2World OLD-STATIC-BACKUP Forum.
 

News:

This is an old OS2World backup forum for reference only. IT IS READ ONLY!!!

If you need help with OS/2 - eComStation visit http://www.os2world.com/forum

Main Menu

eComstation Silver Release has been out since August 28th!

Started by MrJinx, 2009.08.30, 18:16:17

Previous topic - Next topic

MrJinx

I'm surprised this didn't make the site yet!

eComStation 2.0 Silver Release available
August 28, 2009
Serenity Systems International and Mensys BV are pleased to announce the immediate availability of eComStation 2.0 Silver Release for download.

eComStation 2.0 will offer unparalleled performance with Bootable JFS and support for multi-core CPUs.

eComStation is fully OS/2 compatible and will run your existing OS/2 applications!

Got your software subscription still?

Phil

Hi    Downloaded my copy on the 28th, installed and am in the process of try out.
  Yea it is some suprise of not on site yet, since my perception is the forum has been quiet this past week, I assumed that RC6a had run its course and everyone was waiting to pounce upon RC7 (Silver).

Regards 
Phil

warpcafe

Hi,

well... the Silver release was announced some time ago and now is available.
And I guess that "everyone" noticed about the availability. That's why I didn't ran around screaming "it's here! it's here!". :)

OTOH, what suprises me to some extent is the fact that Mensys announced the "soon availability" of RC7 and... they were correct and more or less on time. :))) (Sorry, I simply couldn't keep me from doing that joke. Apologies to the Mensys folks ;) Good job, guys! )

What I'm actually waiting for NOW is that someone posts a message like "I have RC7 installed without any flaws and it blew me away, kick-ass great."

Cheers,
Thomas
"It is not worth an intelligent man's time to be in the majority.
By definition, there are already enough people to do that"
- G.H. Hardy

StefanZ

Ok, then here it is:

"I have RC7 installed without any flaws and it blew me away, kick-ass great."
:D :D :D

I'm usually the one who is running, jumping & screaming here when the new RC is available, but this time I did not. I hoped somebody else would notice. You noticed, so good job guys! :)

And yes, it really IS the best release so far. A lots of bugs and tweaks fixed, some more things automated in an intelligent way, some new (and excellent!) functionality added, generally more compatible with the outside world.

Good job Mensys! Two thumbs up!!  :D  :-*

Saijin_Naib

Think this is worth a hardware install? I've got my 8gb xbox HDD formatted and waiting...

StefanZ

It definitely is. If RC6a worked for you, think of RC7 as just updated and enhanced RC6a.

I already migrated from RC6a to RC7 and not looking back. :)

DougB

Well, it isn't all good. I installed into Virtual PC 5.1 in eCS 2.0 RC6a host. That was the first time that ever worked right, for me. Good job.

Next, I installed to my IBM Thinkpad T43 (1871-W8M). Not so good. ACPI 3.17, which was obviously added at the last second, just won't work. I get traps when ACPI loads. Backing off to ACPI 3.16 allows it to work, as well as it ever did with ACPI 3.16 (which is not perfect). I am also not too sure about the UniAud 1.9.20 that comes with Silver. I think that 1.9.16 works much better.

Next, I installed to my Asus M3A78-EM, with quad core AMD Phenom processor. This is even worse. Phase 1 went okay, and phase 2 started, It ran for a few minutes, and got to a section that says "Merging additional driver selections". There it sat. I left it alone for about 20 minutes, but nothing was happening. I checked for hidden popups (none), and I tried 5 different install options. It still stalled at the same point. OK, I got the installer window minimized, and did Ctrl-Esc. That told me that HWMERGE was running, so I opened the window. It was all black, and obviously, nothing was happening, so I did a Ctrl-C. The window closed, and the install finished normally, up to the phase 3 boot. Then, ACPI Trapped (same as what I saw when i tested the last ACPI 3.17 from betazone). Okay, so I set it back to using APM, and the old OS2APIC.PSD (this is on the CD, but it is not put on the disk). Now, it booted, but I got the repeating startup sound. Okay, so I backed off to UniAud 1.9.16, and now it seems to be working good.

I find this somewhat concerning, because RC6a installed without problems, to both of these machines. ACPI 3.14, as installed by RC6a had some definite problems, but it was good enough to get the system installed, and running. ACPI 3.17 is just not working, on either of these machines, and they are very different. The odd thing is, that some users have reported that ACPI 3.17 is finally working for them. I will admit, that I thought that too, until I discovered that my quad core was running with only one processor. It turned out, that I had chosen some BIOS settings, that were not good, and when I set them to what should be working with ACPI, I got traps. The older ACPI didn't care what those settings were, but it would hang, solid, after 20 minutes to about 6 hours. When I use APM, with the older OS2APIC.PSD, it will run for 8 to 10 DAYS, before it starts to give me trouble, and then it is usually not a solid hang. For me, at least, Silver is a big step backward. If I was a new user, and didn't know what I know about eCS, I would be demanding my money back.

StefanZ

Hmmm, I'm sorry for that. Although sounds strange for me, because on my R51 this is working very well, even the S3 sleep state is working (with some incompatible drivers being manually removed from config).

Of course, ACPI is as always a hot spot here (and probably still will be for some time). It either works, or it does not. But generally speaking it is a big step forward for me.

In a couple of days I should receive a new toy (laptop) so we will see how Silver will be compatible with that one.  ;)

BTW: It is now possible to install without ACPI and then add SMP/ACPI via a wizard.

ivan

If all that I have been reading about ACPI is correct - and I assume it is - then it sounds as if the present development needs to stop and be scrapped and a new effort started - based on what works in linux.

I know this sounds hard and may put some people's noses out of joint but development that relies on a very small pool of hardware is doomed to failure.

warpcafe

Hi,

Quote from: ivan on 2009.08.31, 16:30:05
I know this sounds hard and may put some people's noses out of joint but development that relies on a very small pool of hardware is doomed to failure.

No, you are perfectly OK. However, the very small pool of hardware is the problem.
If more people would consistently test and report feedback with detailed information for the developers, it could only improve. Also, even if Mensys would manage to have a broad testbed of contemporary machines, it means nothing: You cannot cover everything available on a specific date, not to mention new hardware that comes out on a weekly basis.

When you have a storage controller driver, you can expect some kind of time frame where it exists and also some kind of backward compatibility (since it needs to run with existing storage devices and mainboards). When talking about ACPI for example, where each manufacturer implements his own flavor of more-or-less compliant chips, it's a different issue.

Not to mention that the board/chipset manufacturers make their stuff compliant to Windoze rather than compliant to an industry standard. If for example Microsoft fails to adhere 100% to an industry standard, they can lean back and give a damn, since the manufacturers will work around this to be able to sell their crap to the Windoze market. THIS is what I call power. ;)

I don't want to defend all possible things that Mensys perhaps simply failed to do. There was enough positive and also other kind of criticism about this. I just want to mention that Mensys is a rather small company and we are a small community. We can't expect them to invest like Microsoft or Sun and we can't expect to have hundreds of volunteers and reliable coders with sufficient expertise.
Looking at all this implications, I think we're still doing pretty good, also does Mensys. And it seems the Dutch are back on track somehow by now.

Cheers,
Thomas
"It is not worth an intelligent man's time to be in the majority.
By definition, there are already enough people to do that"
- G.H. Hardy

ivan

Thomas, I know what you are saying and agree with most of it.  I take my hat off to Mensys for the work they are doing in trying to advance OS/2 (eCS).

As you say the small pool of hardware available is the problem not helped by the small number of people that have access to the test builds.  There is a way round this but it requires radical thinking - open the testing to anyone with OS/2.  You will get a wider range of hardware because not everyone with OS/2 is using eCS.   For example, I have access to a large number of machines ranging from quad core AMD to a 386 that I would be willing to test ACPI on but I don't have eCS only OS/2 and am therefore unable to do so.

One thing that might mitigate the problems is to NOT install ACPI by default and have a simple way of backing out of the later install if it crashes the machine - something like booting to a command line from CD that removes the ACPI statements from CONFIG.SYS then reboots the machine.


StefanZ

This (not to install ACPI) is actually possible - with RC7 you have to choose the hardware "profile" in pre-boot menu; there is Legacy HW / Modern HW and ACPI mode / Safe mode.

Now it is even possible to change from UNI/nonACPI to SMP/ACPI system (and vice versa) with a newly created wizard.

warpcafe

Hi,

Quote from: ivan on 2009.08.31, 17:32:55
I take my hat off to Mensys for the work they are doing in trying to advance OS/2 (eCS).

Well, now, THIS is exaggerating. ;)

Quote from: ivan on 2009.08.31, 17:32:55
There is a way round this but it requires radical thinking - open the testing to anyone with OS/2.  You will get a wider range of hardware because not everyone with OS/2 is using eCS.

I agree, THAT is the way to go, and it's been my credo for a while that I kept repeating over and over:
Why not create an image that you can put on a blank CD for example... it would give you a bootable system that performs some kind of tests and allows you to store this on a thumb drive or send the results online back if connection is available... whatever...

But that would "only" ;) solve the issue with testing ACPI. It would not cover the test of "eCS".

Cheers,
Thomas
"It is not worth an intelligent man's time to be in the majority.
By definition, there are already enough people to do that"
- G.H. Hardy

DougB

I agree that Mensys, and the developers, don't have enough people helping. Of course, it would help those who do attempt to help them, if they would document their work in a comprehensible way (the current docs are a kludge of, sometimes incorrect, information, that means absolutely nothing to the average user). It would also help, if they would take the extra few minutes to package the updates into the existing WarpIn package, so install, and uninstall, is easy, and correct, every time. Trying to find all of the places to do the updates (or downdates) is a challenge, and it doesn't help when they add new programs, like ACPIEXEC.EXE, with no clue as to what it does, or where it should be put (the program is in the betazone distributions, but not in the eCS 2.0 Silver install). This is a two way street, but doing it the way that it is being done, is not working, and that is mostly because the average user is so afraid to try it out, that they refuse to do it. It would also help, if automatic log collection, and problem reporting, was available. Of course, all of those things take developer time, although someone with some knowledge of how it all works, and what information is needed to analyze a problem, could probably build something to do the log collection, and problem reporting. Building a WarpIn installer can be part of the program build process.

I don't think it would be right to open testing to those who don't own a version of eComStation. After all, Mensys does need something to justify the cost of eCS, and supply an incentive for OS/2 users to upgrade to eCS (and it is definitely an upgrade), so they can recover the cost of development. However, it may be a good idea to open testing, perhaps with a time bombed version of ACPI, to those who do have a license for eCS. A time bomb would also ensure that those who do use the beta versions, won't use them forever.

As for how the development is being done, I suspect that they are trying to do it properly. Unfortunately, nobody else is doing it properly, so they end up in a situation where they need to write special code for every machine, and BIOS level, that is built, to work around the differences. It appears that MS, and Linux, have taken the approach to totally ignore what the machine manufacturers put into their machines (which is often incorrect), and simply do it themselves. That is not the way that it is supposed to be done, but it seems that it is the only way to make it work on every machine, without having to accommodate the differences that exist. The fault goes back to poorly written specs (probably heavily influenced by MS), that can be interpreted in more than one way. Of course, every manufacturer manages to find a different interpretation, and then they change it when they find out that what they thought was right, doesn't work. So, the problem is, that trying to "do it right" has turned into a disaster, but it seems to be too late to change directions. The other problem seems to be that they are using the Intel version of how it is supposed to work. That, mostly, works with Intel hardware, but seems to be incompatible with AMD hardware. You can't just leave a good portion of the market without support. What can be done about that situation is unknown (to me). Hopefully, somebody can find an answer, soon.

warpcafe

Hey Doug,

two thumbs up!
This is by far the best and most professional summary I've read so far.

Quote from: DougB on 2009.08.31, 19:12:05
Of course, it would help those who do attempt to help them, if they would document their work in a comprehensible way (the current docs are a kludge of, sometimes incorrect, information, that means absolutely nothing to the average user).

Yeah, exactly: They should keep in mind that they want feedback from as much people as possible. And not everyone on earth was raised with Assembler as native language or understands dump files of a multicore machine's memory... So they MUST come up with a noob-proof, easy way of doing so. Why not click 1 program icon that will trigger logs collection and sysinfo etc. and zip the crap or even send it back to them immediately if a connection is available? That's the way to go.
Dumping a trunkload of inconsistent and undocumented (or with hard-to-lookup info) executables onto the user and later complain that noone is using it might not be the best approach...

Quote from: DougB on 2009.08.31, 19:12:05
This is a two way street, but doing it the way that it is being done, is not working, and that is mostly because the average user is so afraid to try it out, that they refuse to do it.

I guess it's not only because people are afraid (which is surely correct and the case for quite a lot of people). I think it's also because of the effort required: See, when you find yourself with a new release of your OS and the thing traps... who (except for someone focused on testing) has the will to walk thru readmes, download files, read another readme, rename/move/copy some files, do reboots, collect logfiles...

Quote from: DougB on 2009.08.31, 19:12:05
It would also help, if automatic log collection, and problem reporting, was available. Of course, all of those things take developer time, although someone with some knowledge of how it all works, and what information is needed to analyze a problem, could probably build something to do the log collection, and problem reporting. Building a WarpIn installer can be part of the program build process.

Yep. That would be the approach to choose if you had some kind of change/release management or even processes. -sigh-
Sometimes I have a picture in my mind of how these people work... and it is the same way that I worked... however 15 years ago. :)
Requirements were gathered on a short phone call or on a sticky note. When something wasn't clear I went to the business people who ordered the program and asked them what they meant. Then, the stuff was coded, more or less tested and went into production. One cycle of bug fixing perhaps. Done. And if the customer was asking (and paying) for it, we even took the time to do some kind of documentation...
I guess times have changed. :) But to some extent, looking at what it takes to do my work today, I envy those ACPI developers. :))

Quote from: DougB on 2009.08.31, 19:12:05
I don't think it would be right to open testing to those who don't own a version of eComStation. After all, Mensys does need something to justify the cost of eCS, and supply an incentive for OS/2 users to upgrade to eCS (and it is definitely an upgrade), so they can recover the cost of development. However, it may be a good idea to open testing, perhaps with a time bombed version of ACPI, to those who do have a license for eCS. A time bomb would also ensure that those who do use the beta versions, won't use them forever.

Correct. But it hasn't to be a full-blown retail version of eCS. It could be as easy as having a special distro without any fancy stuff... only the kernel, drivers, log/info collection and a streamlined shell (perhaps not even GUI) to make it as uncomfortable as possible for someone to turn this into a desktop/laptop operating system.
Anyway, if someone really wants to have a "cracked" version of eCS, there are some "warez" and "serialz" places to go to already now...

Quote from: DougB on 2009.08.31, 19:12:05
So, the problem is, that trying to "do it right" has turned into a disaster, but it seems to be too late to change directions.

I'm afraid you're right. Perhaps it could change in 2 or 3 years if there'll be the "next big thing" in computing hardware at the horizon... and someone will make an effort and get a thorough standard implemented. But then, thinking of the implications of "backwards compatibility" -phew- I guess I won't see that happen in my lifetime. ;)

Cheers,
Thomas
"It is not worth an intelligent man's time to be in the majority.
By definition, there are already enough people to do that"
- G.H. Hardy