Question:

Peter Skye wrote the article eComStation Version 1.0 Rolls Out which appears to be a review of the GA Rollout of eCS. There's some stuff I just don't understand. Does Mr. KIA have an opinion?

Answer:

Mr. KIA has lots of opinions. One of Mr. KIA's opinions is that everyone's got the right to have an opinion, even if Mr. KIA does not agree with it. Mr. KIA, noting that some readers don't think much of Peter's article, also has the opinion that one does not persuade someone to your way of thinking by calling them names.

To Mr. KIA, Peter's article appeared to be several themes combined into a single article. It was a review of what happened at the presentation. It was a review of what he believed the eCS product to be. It was also an editorial piece expressing his opinions on the purpose of eCS and its value to the OS/2 community.

Peter's description of what happened at the presentation was pretty accurate. Kim did the presentation using a new master track delivered just before the presentation. Unfortunately, it was not sufficiently tested. Kim did not know this and the result was a less than smooth install. This is what caused the Pre-boot screen display errors and the miscellaneous hangs during the install. All this indicates is that you can never do too much testing.

Peter, at the time, believed he was seeing the GA product that was ready to ship. Those of you on the eCS list know that Kim was asked repeatedly if SCOUG would see the GA. Kim repeatedly and emphatically said we would not know until the presentation. Peter was signed up to the eCS mail list and had downloaded all the articles, but never read any of them until after his article was published. I'm sure more than one SCOUG member believed they were seeing a GA product, based on the various announcements before the presentation. If Mr. KIA was under the mistaken impression that he was seeing a demonstration of a GA product, he would have been similarly unimpressed with the quality of the installation. Kim stated clearly that the errors would be corrected; however, he could have been much more emphatic that the code he was showing was not even close the what the GA would be.

Like many newcomers to eCS, Peter was confused about what eCS is and what the distinctions are between the bootable CD installer, the eCS demo CD and eCS the product. Mr. KIA has followed the evolution of eCS on the mailing lists and newsgroups for quite a while. It is his opinion that Serenity does a mediocre job of describing eCS to OS/2 users. The same questions come up over and over from new arrivals and there's no single URL that we can point to and say go here and do a little reading and then come back and ask questions. Instead, each new arrival gets individualized education. This is not terribly efficient or effective and contributes to the massive data overload on the eCS mailing list.

At the presentation, Kim focused on demonstrating the bootable CD installer and WiseManager. During the presentation, he also discussed the bootable demo CD and the customized CD's that ISV's could build to use as a sales tool or even as a ready-to-run product. WiseServer and RIPL booting clients were also mentioned in passing. It would be easy for someone not extremely familiar with eCS and Serenity to conclude the CD's were all one and the same product or to miss the subtle distinctions between them.

All of this probably contributed to the now infamous "no SCSI support" and "no networking" statements. Folks using SCSI with eCS clearly know SCSI works and the support is not going away. The "no networking" statement is both correct and incorrect. If one goes back far enough, IBM was not going to permit Serenity to put networking support of any kind on the bootable demo CD. Somewhere along the way this changed so that TCP/IP networking would be available on the demo CD, but LANServer/Peer networking would not be. Again, there's no place to go to find a product description of what the demo CD can and cannot do along with a definition of what Serenity considers to be networking support.

After publishing his article, Peter worked with the Serenity team to better understand the eCS product. As a result of what he learned, he made several updates to his article. There are still several parts of Peter's article that Mr. KIA disagrees with.

Peter states:

"eCS boots from CD and can be run from the CD without installing to a hard drive"

This is not exactly true. the eCS install CD is bootable as will be the demo CD and ISV custom CD's; however, what most OS/2 users think of as eCS will be installed to a hard drive and will boot from the hard disk not a CD. For these users, the bootable install CD does little more than avoid the need to create the 4 installation diskettes. The CD may also be able to replace utility diskettes for emergency maintenance, but this depends on exactly what is available on the installation CD. As of today, the exact content of the CD is not documented on a publicly accessible site.

Peter continues with:

"no networking is included in this mode"

This also is not exactly true. The planned demo CD will have TCP/IP networking available and the ISV custom CD's can support any form of networking the ISV wishes.

You have to define what you mean by eCS before making statements about networking support.

Peter makes statements that imply that determining the correct SCSI driver might be so difficult as to require opening the case. Mr. KIA believes Peter is overstating the problem by several orders of magnitude. First, the vast majority of SCSI users know what hardware they are using. Second, the vast majority of users and potential customers are and will be using IDE drives. When trying to get an unknown system to boot eCS, there's more likely to be a problem choosing between Daniela's and IBM's IDE drivers.

Peter states:

"Also, many apps today need networking and can't be demonstrated without it. Nor can the developer include a server in his onsite demonstration since networking isn't supported."

Given that the demo CD will have TCP/IP networking and that an ISV's CD can support any form of networking the ISV wishes, the above statement is just plain factually wrong.

Peter states:

"One user, however, searched for the document on the pre-release version and could not find it."

The document is in the root of the eCS Preview1 CD and, strangely enough, is titled "known.issues." As to why the two individuals Peter mentioned could not find this document, Mr. KIA could only guess. Interestingly, there's also a document titled "readme.txt" on the CD which refers the reader to a document titled "known.issues." As to why Peter did not verify these individual's statements with Kim or someone else since there was a direct contradiction, Mr. KIA can only guess.

Regarding "readme.txt," Mr. KIA would suggest that Serenity deliver word-wrapped text files. The lack of pre-wrapped lines in readme.txt makes it somewhat difficult to read in tedit and it prints poorly if you "copy readme.txt lpt1."

Peter states:

"These separate documents do make for a more complex install process. Kim noted several times that failure to follow the printed install documentation instead of relying on the screen displays could lead to a failed install."

Mr. KIA believes Peter misunderstood the context of Kim's statement. Kim was referring specifically to the networking install. The replacement for the networking install is not finished. All that eCS provides is a wrapper for IBM's existing networking installation programs, thus the eCS network install suffers from many of the same defects and failure modes as IBM's installer. The eCS wrapper goes a long way to ensuring a successful first time networking installation by providing SmartGuides which assist the user with the non-obvious parts of the network install. Serenity has completely reimplemented the base OS/2 install, so none of the IBM installer issues exist for the base install; however, the GA will be v1.0 code, so there are bound to be some new issues. Hopefully, they will be minor.

Peter states:

"Cheung was apologetic. "This code wasn't written by Serenity," he said, "it was a worldwide effort."

following his description of some of the installation glitches. This could be read as if Kim said this as an explanation of why the glitches existed.

This is simply not the case. The two statements were made at separate times. Kim was hardly apologetic. He was, and is, extremely proud of the world-wide effort. He was apologetic about the defects and explained that since the code was not written by Serenity and it was just delivered and that he did not have a chance to test it before the presentation.

The section "Does It Install Faster?" has also raised some controversy. Did I tell you Peter thrives on controversy? The presentation was a demonstration of the eCS installation and not specifically an installation speed test. Kim paused to answer questions, explained the screens and entry fields and had to suffer with some installation glitches. He also switched video to another system while the install was in progress so that he could describe aspects of the WiseManager package. To use the duration of this install as a measure of how long a typical eCS GA install will take is just plain bad science. However, fear not: once eCS goes GA, SCOUG has a plan to answer this question. Stay tuned for the results.

If you think the installation analysis was controversial, you should have seen what happened when folks read the section "The Road To Windows." Best I can recall, Kim's statement was either as Peter quoted it or very close. My recollection was that it was said nearer to the middle of the session, but I was not taking notes and Peter was. This is one of those cases that I don't think Kim meant what he said. Perhaps he was talking of hosting WiseServer on another OS and remote booting eCS clients. I can only guess. eCS, as most of us know it, is the OS/2 kernel, the WPS and OS/2 native applications with all the stability and functionality that that implies. This chunk of code is not going to get ported to another OS, much less another CPU architecture any time soon. Technically, it would be possible to run OS/2 as a client on a virtual machine. The host OS had better be as stable or more stable than OS/2 or I can't imagine many OS/2 users being very interested. Perhaps, large enterprises with legacy OS/2 applications might be more receptive to OS/2 as a client OS of a virtual machine. Mr. KIA can only speculate.

On Peter's statement that:

"eCS is a good fit for those current OS/2 users who are planning to migrate to Windows. eCS has been designed with that in mind..."

Mr. KIA can not understand how Peter could conclude this, even after long discussions with him. To resolve this apparent paradox, Mr. KIA refers the reader to his statement about opinions at the beginning of this article.

Peter states:

"The future of eCS appears to be a cross-platform product which supports various operating systems, most likely Windows and, possibly, Linux."

eCS provides tools to make it easier to interoperate with these platforms. HOBLink supports remote execution of Linux apps. SmartSuite provides reasonably good document interchange with MS products; however, none of this fits Mr. KIA's definition of cross-platform. Cross-platform means the application runs natively on multiple platforms. SmartSuite is cross-platform. Describe is cross-platform. VisualAge Java was cross-platform until v3.02. eCS in not cross-platform as Mr. KIA uses the term.

If anyone wishes to agree or disagree with Mr. KIA, he can be contacted at Mr. KIA.

SCOUG, Warp Expo West, and Warpfest are trademarks of the Southern California OS/2 User Group. OS/2, Workplace Shell, and IBM are registered trademarks of International Business Machines Corporation. All other trademarks remain the property of their respective owners.