VOICE Home Page: http://www.os2voice.org
[Newsletter Index]
[Previous Page] [Next Page]
[Feature Index]
|
VOICE Home Page: http://www.os2voice.org |
March 2003
[Newsletter Index]
|
An editorial view from Christian Hennecke.
Lately OS/2 users were surprised by two new announcements, which could affect the future of OS/2 and especially eComStation. They concern both Java and Virtual PC versions for OS/2. For a take on Microsoft's impending purchase of Connectix (developer of VPC), see Thomas Kline's "The Redmond Borgs keep assimilating" in this issue.
Innotek, and not Golden Code as announced last year, has released the first beta version of the Java 1.4.1_01 run-time environment and developer's toolkit. And surprisingly they are free for private use. According to Serenity Systems eComStation 1.1 will include a license for commercial use of the packages. With regard to IBMs support guidelines, the named system requirements are Warp 4 with fixpak 16 or later, or similar Convenience Pack systems (which includes eComStation). Use with earlier versions should be possible, but not supported. It is not clear which versions of Warp 4 or even Warp 3 are included in this statement. Nevertheless good news, since otherwise many may have found themselves locked out from using modern Java applications soon.
There are two points which have already broiled quite some minds. First, Innotek's Java is the Windows version which has been utilized for OS/2 via Odin technology. Second, Golden Code is still working on a second, native Java version. Now, what does that mean? Is it that bad, or that good, as the advocates of each position claim? Let us have a closer look.
Innotek lists a number of advantages of their approach. Those include the timely availability of new versions and bug fixes, since existing Windows versions would only have to be integrated with Innotek's separate »OS/2 Kit«. Furthermore, they list substantially improved loading times for the JVM on JFS drives compared to traditional ports to OS/2 due to the better loading performance of the Windows format for executable files. Last they claim excellent performance thanks to the use of optimized Windows binary modules and the Sun HotSpot JIT compiler and mention that a traditional port would typically have to suffer from the outdated compiler technology that is available on OS/2. Appropriate or not - these references to inherent advantages of the bad W system have predictably proven as most unadept psychologically and that way Innotek hasn't exactly made many friends among OS/2 end-users. However, one should keep clearly in mind the actual target group of such statements - and that is not the end-user with a strong emotional bond to OS/2 for sure.
Others demur potential problems regarding stability, performance, and compatibility in comparison to a native OS/2 port. Stability would suffer, because you wouldn't get the bugs of the Windows version only, but also those of the Odin layer between it and the OS/2 system. The additional layer would impair performance, since a redirection or complete emulation would have to be done for every command. If the emulation is not perfect, this would result in reduced compatibility or a reduced feature set. As a proof of diminished performance a comparison of benchmark results of versions 1.3.1 and 1.4.1_01 has been posted in the international news groups, which seems to confirm that path of argumentation.
So, is there fire behind the smoke? It is impossible to make a statement regarding stability and compatibility after such a short time, especially since we are talking about a beta version. When compared directly to the native Java 1.3.1 JVM for OS/2 in terms of performance, it shows however that the truth lies somewhere in between and is dependent on the profile of usage.
With smaller applications like Smartcache, differences were hardly noticeable on the test machine (with AMD K6-III 400 CPU, 16 MB JFS cache, UW-SCSI hard disks). When operating GUI applications Java 1.4.1 Innotek's package feels better than IBMs Java 1.3.1, which often calls for the image of stirring a pertinacious dough. Depending on the used applications, the differences between the two JVMs can get downright blatant. In comparison to IBMs Java, jEdit 4.0.3 started more than 10 seconds - equaling more than 30% - faster, if Innotek's package was used! A clear gain especially for end-users who work with different Java applications in turn and don't have them running all of the time. Furthermore, the differences concerning the »dough effect« really show with jEdit's dialogs.
Things are different when it comes to applications which execute complex computations though. When rendering images with Art Of Illusion, the differences turned out as blatant as for starting jEdit - but this time in favor of IBM's JVM. The IBM JVM rendered the test scene in 21:26 minutes, while Innotek's package took 34:26 minutes. Depending on which processor types the JVMs were optimized for during compilation, the results may be different if current processors would be used. Moreover, the comparison was done between two different versions of a product. According to reports Innotek's package still offers some room for optimizations. So we remain curious.
So much for the technical data. Now why are two versions of Java 1.4.x developed in parallel, which have even reached about the same status (Golden Code's version is said to be near beta also)? It seems that Golden Code and Innotek intended to work on this project together at first, but then couldn't agree on the native or Odin-based approach. So is it all SNAFU'd (For those unfamiliar with the term "Situation Normal, All Fouled Up" is a more tempered definition)?
Reasons for the Odin-based version were listed above, with the timely availability of new versions and bug fixes, once the integration environment has been established, seeming most important. For a native version the respective fixes could also be expected, but new versions are likely to require more work. So the apprehension that worrying about whether the next Java version is going to be available at all or not may soon begin again is perfectly comprehensible. On the other hand the question arises whether the Odin-based JVM really allows one to realize the complete feature set. As we can see from Innotek's statement regarding MIDI support, for instance, this applies to multimedia, OS/2's traditional weakness. Furthermore, the package can use TrueType fonts only, while installation cannot be done via the respective standard controls of OS/2.
Finally, the decision is also a political one. OS/2's Java implementations have been among the fastest and best for a long time, which was surely due to IBM enforcing its Java strategy. If there is going to be no native JVM any more, some may ask of what value such a system still is. (In the age of out-sourcing, the fact that a new JVM won't be developed by IBM is of secondary importance.) Against that one might argue that Java is actually predestined for an Odin-based implementation, since it draws all elements of the graphical user interface itself and that way the lack of native PM controls wouldn't strike anyway.
So the motivations are many. In the end having a choice and being able to choose the product you personally prefer the most is a good thing. I only hope that this doesn't result in the two implementations choking off each other's air supply and the venture not being worthwhile for both companies alike.
VOICE Newsletter Update: Please note our new Disclaimer page regarding warranty and copyrights, and the respective terms for submitted articles in our Guidelines for submissions to the VOICE Newsletter.
We are always interested in your thoughts and views on subjects related to OS/2, and would like to see opinion/editorial pieces as well as hardware/software reviews and HowTo articles. If you can help by writing an article please contact me at editor@os2voice.org.
Our Newsletter team is in need of backup. A few people have joined after we published a call for help in one of the last issues, but we could do with some more. To be able to help you don't have to be a very good translator or HTML programmer. If you have profound knowledge of English or German spelling and grammar, you can also help with editing the articles. Some hints on translation activities are also available in the FAQ.
Warp Doctor: Have an idea for Warp Doctor? You can send your comments directly to the Warp Doctor web guy Jeremy at rs@fyrelizard.com or better yet attend one of our Warp Doctor Team meetings, weekly on IRC.
Everyone's help is required to keep this project going. Please note the change in schedule again. The team meets every Sunday at 3PM EST (20:00 GMT) now on IRC in the warpdoctor channel on the WEBBnet IRC network. The alternate Sunday schedule was too confusing. For more information on attending online IRC meetings please see the VOICE Meeting Information page - http://www.os2voice.org/meetinginfo.html.
VOICE Online Update: This month the general member meetings are scheduled on March 3 and 17 at 8PM EST (01:00 GMT). Everyone interested in OS/2 or eComStation is invited to attend either or both of these sessions in VOICE on the Webbnet IRC network. For more information on attending online VOICE IRC meetings please see the VOICE Meeting Information page - http://www.os2voice.org/meetinginfo.html.
If you have an idea for a Speakup event, please submit it to liaison@os2voice.org, and we will try to schedule something. As always, please be sure to check out the updated VOICE Future events Calendar in this newsletter or on the VOICE website at http://www.os2voice.org/calendar.html for more details on future VOICE events.
This month Thomas Klein returns with part six of DrDialog, or: How I learned to stop worrying and love REXX. He continues with his overview of development of REXX programs with GUI using the free development environment DrDialog by IBM. This should be interesting for anybody, especially for those who have always wanted to write their own OS/2 applications, but were afraid of the high learning curve of PM programming.
In OS/2 support forums the question for PCI sound cards which are suitable for use with OS/2 arises again and again. Users who are searching for such a card sometimes get to feel like they are in an episode of The X Files. Lest you have to search for the truth out there any further, Robert Fuchs has compiled a considerable amount of information in PCI sound cards for OS/2 and eComStation.
For a long time OS/2 users couldn't play DVD videos on OS/2. This has changed with WarpVision entering the scene. In Video DVDs and eCS / MCP, Philhard Ackermann depicts how to do it and what to take care of.
Need to transfer data from your camera or similar device to your PC? You don't have PCMCIA? There are alternatives. Mark Dodel tells you all about USB Compact Flash readers and OS/2.
Finally we have the VOICE Newsletter OS/2 Tips page and the Letters, Addenda, Errata page. If you have any OS/2 or eCS tips you've uncovered, please send them to tips@os2voice.org. If you have any comments or suggestions about the newsletter or articles in it, please send them to editor@os2voice.org.
That's it for this month. Upcoming articles include an update by Alfredo Fernßndez Díaz of his July 2000 article on the current state of booting OS/2 from a CD and the next articles in the series on DrDialog.
Mark Dodel, Christian Hennecke and Jason R. Stefanovich
VOICE Newsletter editors
[Feature Index]
editor@os2voice.org
[Previous Page] [Newsletter Index] [Next Page]
VOICE Home Page: http://www.os2voice.org