Talk:IBM i
![]() | This article is rated B-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | ||||||||||
|
![]() Archives (Index) |
This page is archived by ClueBot III.
|
Is there any reason to believe that "IBM i" isn't just a newer name for the OS that started out with the name "OS/400"?
editOperating systems have often had name changes accompanying version changes or platform changes; is there any reason to believe that "IBM i" is a new technology, rather than just a new name for an OS that has had technology added to it in releases that haven't had name changes? Guy Harris (talk) 01:06, 10 May 2020 (UTC)
- In this case: No. But there are cases in operating system history where it would be relevant to have different articles pre/post a name change, or other changes.. like Next/OpenStep vs Mac OS X and the different strains of Windows. -- Henriok (talk) 17:59, 11 May 2020 (UTC)
- Yes, I know, but I was asking specifically about this case, which is more like "Mac OS X"/"OS X"/"macOS", where the names changed for marketing reasons but the code base changes were just release-to-release changes, without the code base changes with name changes not being more significant than the code base changes without name changes, than like NeXTStEP vs. Mac OS X, where the former acted as the basis for the latter but where a lot of changes were made to make the latter from the former, or like "Windows OT" (16-bit and Windows 95/98/Me) vs. "Windows NT" (from NT 3.1 to 10), where two different code bases implemented the same API and ABI (with many additions in the NT case). Guy Harris (talk) 18:05, 11 May 2020 (UTC)
EWLwiki (talk) 00:51, 14 May 2020 (UTC) I thank Guy Harris and others. What are the functions, objects, elements, attributes and characteristics of one operating system which become deprecated, replaced, rewritten, so that a new name is deserving for the newer operating system? Consider, are all computers essentially calculating machines. Is the abacus the forerunner of all? Here are some comments from 1997, 23 years ago. Additional points from Inside the AS/400, Featuring the AS/400e series, second edition, by Frank G. Soltis. Chapter 3, page 49. confusions about where the fine the operating system in the AS/400. OS/400 lacks most of the functions found in other operating systems. the high-level machine interface, called the MI, is a complete set of APIs for writing applications. OS/400 programs do not manage system resources. MI is technology-independent interface that has no knowledge of the memory configuration and characteristics. Thus memory management needs to be handled below the MI. OS/400 resides above the MI. No hardware-dependent components can be in OS/400, but must be beneath the MI to preserve technology independence.. the MI protects application programs and OS/400 from hardware changes. The operating-system software beneath the MI is called the licensed internal code (LIC). Page 50. OS/400 consists of objects and programs above the MI. The LKIC compress the data structures and programs below the MI.. The LIC is the link between the MI and the hardware. So the AS/400 operating system is the combination of OS/400 and the LIC. Kernel is a convenient way to describe the operation-sytem functions packaged under the MI. Work management with schedules jobs, can be predominantly I OS/400 because such functions have few hardware dependencies. Device driver support can be partially in OS/400 and partially in the LIC. Chapter 7 will show that system-wide security is in OS/400m whereas the authorization to system resources is below, in the LIC. Most major OS functions, like security, have some portions above and some portions below the MI. (end page 51) Page 52. decided to call the kernel microcode. Page 53> Internal code for the RISC-based systems is called System Licensed Internal Code (SLIC). Many changes were required to the LIC below the MI when Rochester began working on the RISC processor in 1991. Page 54. Redesigned and rewrote from scratch. Page 55. Selected C++ programming language for the new SLIC development. Page 56. The SLIC project began in June of 1992 produced more than 1 million lines of C++ code and more than 7,000 classes (classes = template that describes all objects that share the same operations and seat elements in Object-Oriented programming). Counting all the ported code, they had more than 3 million lines of trusted operating system code under the MI. SLIC is treated like the kernel of an operating system. Within the kernel, all code is assumed to be trusted. Only the SLIC developers in Rochester were allowed to generated code that runs in SLIC. No one else is allowed to write SLIC functionsRochester has never shared SLIC-level code with anyone outside the SLIC team. The IBM I technology layers have changed dramatically in the 23 years since 1997, and has much more complexity than what Dr. Frank Soltis described in his 1997 book. I will try to post newer, current content as to what IBM considers the IBM I operating system and its layers purport to be. Steve Will, the current chief scientist for IBM I, frequently speaks on the topic that IBM i is not OS/400EWLwiki (talk) 00:51, 14 May 2020 (UTC) EWLwiki (talk) 00:51, 14 May 2020 (UTC). https://www.helpsystems.com/resources/on-demand-webinars/ibm-i-its-not-just-as400-steve-will
https://www.youtube.com/watch?v=DM6BiQZyOqk
https://www.youtube.com/watch?v=f13ke9hY7Wk
https://www.se-radio.net/2011/07/episode-177-ibm-i-os400-operating-system-with-steve-will/
https://all400s.com/assets/Analogy.pdf EWLwiki (talk) 00:51, 14 May 2020 (UTC)
EWLwiki (talk) 03:41, 14 May 2020 (UTC) another discussion of names https://all400s.com/assets/Analogy.pdf. EWLwiki (talk) 03:41, 14 May 2020 (UTC)
- Yes, I read Inside the AS/400 several years ago. The biggest change you mention is the rewrite of the LIC from PL/AS(?) on the IMPI machines to C++ for the RISC-based machines; that change predates the name "IBM i" - the system was still called the "AS/400" and the OS atop the LIC was still called "OS/400" at the time.
- The "IBMs AS400 - A Simple Analogy Explains Everything" mentions a bunch of things, not all of which arrived at the same time as the "IBM i" name. Updates to RPG are about as relevant to the underlying OS as are updates to C in UN*Xes, i.e. not at all; going from C89/C90 to C99 does not make a new operating system. The first big system-level change they mention is the addition of PASE, which dates back to at least OS/400 V5R1, so it's not something that distinguishes "IBM i" from "OS/400". Similarly, http://www-912.ibm.com/as400/v5r2to/pdfs/j02linux.pdfhttp://www-912.ibm.com/as400/v5r2to/pdfs/j02linux.pdf Linux partitions date back to at least V5R2]; it runs in an LPAR. So I see nothing there that distinguishes between "OS/400" and "IBM i".
- I'll listen to the talks by Steve Will to see what changes were made at what time to see whether there's a clear line between "IBM i" and "OS/400" or "i5/OS" (or between "OS/400" and "i5/OS", for that matter). The diagram on page 5 incorporates both PASE and Linux partitions into "IBM Power Systems", although that's misleading as they were both available before IBM decided to stop marketing nearly-identical System i and System p hardware separately. (Amusing question of the day - other than adding tag bits, what other changes to z/Architecture would be needed to make it capable of running IBM i? :-))
- "We added new technology XXX" is not sufficient to make a new OS; plenty of OSes have added new technologies and kept their identity. For example, the OS running on my laptop replaced its init subsystem, and then apparently replaced the old launchd with a re-implementation, and added Metal and a Mach-message-based inter-process communication mechanism, neither of which were synchronized with the two name changes it got, but it's still just the latest version of the NeXTSTEP derivative it started out as in Mac OS X 10.0. The same could be said about Windows NT, Solaris, IBM AIX, MVS/z/OS, HP-UX, etc.. Guy Harris (talk) 03:50, 14 May 2020 (UTC)
- For the sake of posterity, I decided to change the article to treat OS/400, i5/OS and IBM i as the same operating system after re-reading the history section of Frank Soltis' book and observing that most of the major changes to the operating system occurred during the time it was known as OS/400, as per Guy Harris's comments.
- While OS/400 was an evolution of CPF, CPF is sufficiently different to warrant its own article, e.g. CPF had capability addressing, which was abandoned during the development of OS/400 Vt320 (talk) 16:29, 2 March 2021 (UTC)
Since there seems to have been some fresh confusion on this subject recently, it's worth reiterating the following points (references can be found in-line in the relevant parts of the article).
- Most if not all of the major architectural changes and additions to the OS happened when it was called OS/400. Subsequent releases either added new hardware support, or have implemented functionality on top of the same core features which have been around since the 90s.
- When IBM rebranded OS/400 as i5/OS, the announcement letter explicitly called it "A different name for the same operating system"
- When IBM rebranded i5/OS as IBM i, it renamed two versions already released as i5/OS (specifically V5R4 and V6R1) to IBM i. It's difficult to claim that i5/OS and IBM i are somehow distinct operating systems when IBM retroactively renamed two of the three i5/OS releases to IBM i.
Given that even IBM treated and described OS/400->i5/OS->IBM i as simple name changes, I believe the article should continue to do likewise. There is also clear precedent around using the latest name of a product line to refer to earlier releases, and using the release date of the original in the lede and infobox see macOS, iOS, OpenVMS etc. Vt320 (talk) 22:57, 23 November 2021 (UTC)
Merger proposals
edit- The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section. A summary of the conclusions reached follows.
- One approval, and no discussion. Proceeding with merger. Vt320 (talk) 23:20, 9 March 2021 (UTC)
I propose to merge Auxiliary storage pool into IBM i. There is relatively little content in the Auxiliary storage pool article, and I don't believe that much more needs to be said about it in a wiki article. I think that it would fit well as a feature sub-section of the IBM i article Vt320 (talk) 20:08, 2 March 2021 (UTC)
- Support Guy Harris (talk) 20:43, 2 March 2021 (UTC)
- The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section. A summary of the conclusions reached follows.
- Not proceeding with merger. Vt320 (talk) 10:15, 21 March 2021 (UTC)
I propose merging Qshell into IBM i. The Qshell article previously featured a lot of unencyclopaedic content, such as a list of the Unix commands it supported, and instructions on how to use it. When reduced down to a set of noteworthy statements and references, it overlaps with the description of Qshell in this article. Merging the articles will lead to a negligible increase of content in the IBM i article, and keep information about the system grouped together. Vt320 (talk) 15:08, 17 March 2021 (UTC)
- Although I really appreciate your contributions to the IBM i and other articles, I do not support this suggestion. I don't think that removing all the noteworthy information from the Qshell article and then merging the few remaining parts as a side note to the IBM i article is a good idea. I'd rather like to see the Qshell article being much more expanded and improved as it is a unique Unix shell implementation on a non-Unix operating system and therefore worthy of a separate article in my opinion. Ghettoblaster (talk) 18:31, 17 March 2021 (UTC)
- I don't see "To determine if the Qshell Interpreter licensed program is installed on IBM i..." as noteworthy; as Vt320 said, WP:NOTGUIDE. Are you talking about the list of included commands? Guy Harris (talk) 20:07, 17 March 2021 (UTC)
- I agree that the current state of the article was not (yet) great, but at least it was a start that could be worked on. In the past it already contained some more useful information on how the shell differs from other Unix shells, but that got removed at some point as well. I don't think that merging all short articles into bigger ones really does help getting the content organized and expanded. And as for the command list, this is not uncommon (like in BusyBox) and is useful as well since the commands seem to be built-in and not just part of the OS as in other environments. Ghettoblaster (talk) 21:38, 17 March 2021 (UTC)
- From a quick look at the history, the information on how the shell differs may have been copied from IBM documentation and removed because it was a copyright violation. If somebody wants to provide that information in their own words, they should go for it.
- "Built-in" in what sense? The IBM documentation mentioned in the edit that removed the stuff that violated copyright says you can write your own utilities, although I haven't looked for documentation on how that's done. Newer versions of the OS seem to have introduced some notion of private address spaces per process, but I don't know whether the command runs in another process or another address space. Guy Harris (talk) 21:56, 17 March 2021 (UTC)
- And this is the kind of content I'd like to see in such an article. How is this implementation different from a user and from an architectural perspective from any other Unix shell? Unfortunately, I did not get a chance to research that myself yet. Ghettoblaster (talk) 22:11, 17 March 2021 (UTC)
- I agree that the current state of the article was not (yet) great, but at least it was a start that could be worked on. In the past it already contained some more useful information on how the shell differs from other Unix shells, but that got removed at some point as well. I don't think that merging all short articles into bigger ones really does help getting the content organized and expanded. And as for the command list, this is not uncommon (like in BusyBox) and is useful as well since the commands seem to be built-in and not just part of the OS as in other environments. Ghettoblaster (talk) 21:38, 17 March 2021 (UTC)
- I reinstated the command list, I was unaware of the precedent from articles like Busybox of listing commands. I still stand by removing instructional information. I am inclined to leave the Qshell article as a standalone article. I'm not sure if I will get around to contributing to it. Vt320 (talk) 01:14, 21 March 2021 (UTC)
- I don't see "To determine if the Qshell Interpreter licensed program is installed on IBM i..." as noteworthy; as Vt320 said, WP:NOTGUIDE. Are you talking about the list of included commands? Guy Harris (talk) 20:07, 17 March 2021 (UTC)
Recommend Removal from IBM i article material that is duplicative of content found in AS/400 article
editEWLwiki The IBM i article is for IBM i OS, not for extensive discussion of hardware. The IBM Power Systems article is a good location for discussion of hardware for IBM i. I ask Vt320 and Guy_Harris for consensus on removing from the IBM i article the content which is duplicative of content in the IBM_AS/400 article, specifically, discussions of hardware. These IBM i sections have content and verbiage almost identical to content found on the AS/400 page.
o Origin https://en.wikipedia.org/wiki/IBM_i#Origin o The Move to PowerPC. https://en.wikipedia.org/wiki/IBM_i#The_move_to_PowerPC o Rebranding. https://en.wikipedia.org/wiki/IBM_i#Rebranding o The IBM i article sections on History at https://en.wikipedia.org/wiki/IBM_i#History and Architecture also have much content that is repetitive of material in the IBM_AS/400
EWLwiki (talk) 23:59, 11 December 2021 (UTC)
- Every single section you have listed is about the operating system, and not the hardware. It is of course difficult to not discuss hardware somewhat when discussing things like the origin of the OS (given that it was designed around the AS/400) and the move to the PowerPC-AS architecture, but at least some of that is needed for context. Vt320 (talk) 20:52, 12 December 2021 (UTC)
- I agree with Vt320 here. The discussion of hardware here doesn't strike me as extensive; it's providing necessary context, as Vt320 notes. Guy Harris (talk) 22:14, 12 December 2021 (UTC)
OS name changes, fundamental OS changes, platform name changes, and platform changes
editFor the platform formerly known as AS/400 and the OS formerly known as OS/400, the following name changes and platform changes have occurred:
- One (or more) IMPI processor changes - as noted in the AS/400 article, "There were at least two generations of IMPI processors, the second was released in 1991.[1]"; no OS name changes were made, and any OS changes for the new processor(s) were probably just in the vertical "microcode" layer to support the new hardware. There may also have been other non-CPU hardware changes requiring software changes, but no OS name change or platform name change was made.
- Introduction of the PowerAS-based machines (1995). This involved a complete rewrite of the vertical "microcode" layer; they took the opportunity to rewrite most of it in C++ (replacing old PL/MP code). There may also have been changes to the XCPF layer required. However, no OS name change or platform name change was made; the only name change was that the lower OS layer was renamed from "vertical microcode", or whatever they were calling it at that point, to "system licensed internal code".
- PowerAS processor changes (1995, 1997, 1998) - as with the IMPI processor changes, these probably mostly involved changes to the SLIC layer, and did not involve name changes.
- Dropping of IMPI support (1997) - OS/400 version 4 dropped support for IMPI processors.[2] No OS name change was made.
- Platform rename to "eServer iSeries" (2000) - the table in AS/400 § Hardware seems to indicate that the first machines released with the new name used the RS64 III processor, but that's unlikely to have provoked the platform name change, given that the introduction of RS64 II didn't, and the subsequent introduction of RS64 IV didn't. The name change appears to have been the bright idea of somebody in IBM's marketing department, to try to give their z/Architecture, PowerPC, and x86 server lines similar names (eServer {z,i,p,x}Series). The OS was not renamed at that point.
- Platform change to use POWER4 processors (2000) - this meant that the same processors were used for all new iSeries and pSeries machines (although some previous pSeries machines used RS64 processors rather than POWERn processors). No platform name change or OS name change ensued; the OS required changes to support the new hardware.
- Platform change to use POWER5 processors, platform name change to eServer i5, OS name change to i5/OS (2004) - not sure why IBM marketing decided that sticking the ordinal number of that particular generation of POWERn processors was a Good Idea, but expecting rationality from marketing departments is usually a mistake. This presumably involved the usual OS changes to support the new processor (and changes in AIX for the same reason), but, as IBM themselves said, the name change was just to "a different name for the same operating system".[3] It continued to run on pre-POWER5 machines; the previous reference says "It runs on IBM iSeries™ servers and AS/400® Models 720, 730, 740, 170, SB2, SB3, 250, 270, 800, 810, 825, 870, 890, and IBM i5 Models 520 and 570.", which goes back to RS64 II.
- Rebranding to "System i5" (2005) - no OS name change.
- Rebranding to "System i" (2006) - no OS name change.
- Platform change to use POWER6 processors (2007) - no OS name change, even though the "5" in "i5/OS" no longer applied to all machines on which the OS ran, and no platform name change, as the "5" in "System i5" had already been dropped.
- Replacement of of System p and System i platforms with unified "IBM Power Systems", OS rename to "IBM i" (2008) - new platform, but not a major platform change, given that both platforms were using POWERn processors and at least some common hardware already.
The one non-incremental OS change was the addition of support for PowerAS processors. The platform change to PowerAS was a non-incremental platform change. The other OS changes, and most platform changes, appear to be incremental (new features added to the OS, new processors added to the platform); maybe the merger of System p and System i wasn't incremental, although, as far as I know, the difference between System p and System i hardware had been decreasing over time already, so perhaps it was just an acknowledgement of reality.
The biggest change, from the IMPI CISC processors to the PowerAS RISC processors, was not accompanied by a platform or OS name change, so it's not as if an OS name change signals a major OS or platform change.
Yes, the latest version of IBM i won't run on any of the machines branded "AS/400", but the oldest OS release rebranded as "IBM i" was 5.4, and the OS continued to support older pre-POWER5, pre "eServer i5" machines.[4]
So it's irrelevant to the question of whether IBM i is a separate OS from OS/400 or i5/OS (or whether OS/400 is a separate OS from i5/OS) whether the current version of IBM i runs on any machines that ran OSes branded "OS/400" or "i5/OS" - sometimes newer versions of OSes drop support for older hardware and those changes are usually not accompanied by name changes, and an OS name change doesn't necessarily accompany the dropping of support for older hardware, and even if the first version of the OS with the new name drops support for some older hardware, it's rarely if ever the case that there is no overlap between the platforms supported by the last OS version with the old name and the first OS version with the new name.
(For another example of platform changes and OS name changes, consider Apple's current UNIX-for-Macs, originally named Mac OS X and now named macOS. The major platform changes for Macs were the changes from PowerPC to x86 and from x86 to ARM, with the addition of 64-bit PowerPC and 64-bit x86 being minor changes. Those required less OS change than the switch from IMPI to PowerAS, because macOS programs are compiled into directly machine code when the software is built rather than being done by low-level OS code (VMC/SLIC) when the program is first started; the bulk of the OS code is platform-independent. In no case did the platform change involve an OS name change - the "Mac" in "Mac OS X" was partially dropped in Lion and fully dropped in Mountain Lion, and "OS X" became "macOS" in Sierra, but the first OS version to support x86 was Tiger and the first OS release to support ARM was Big Sur. (And. np, Apple changing from calling OS major versions "10.x", with "x" increasing by 1 from major release to major release, to calling them just "y", with "y" increasing by 1 from major release to major release, had no special significance, even though it occurred when ARM support was introduced. It was just Apple ditching the last bit of "pretend this is just at the next version after Mac OS 9", the first bit having been dropped when the name was changed to "macOS"; the original name emphasized continuity from the old, completely-different classic Mac OS, while the new name emphasized that the OS for the Macs was part of the same family of OSes as the OSes for iPhones, iPads, Apple TVs, and Apple Watches.) Guy Harris (talk) 10:41, 15 July 2022 (UTC)
References
- ^ Schmierer, Q.G.; Wottreng, A.H. (1991). "IBM AS/400 processor architecture and design methodology". [1991 Proceedings] IEEE International Conference on Computer Design: VLSI in Computers and Processors. IEEE International Conference on Computer Design: VLSI in Computers and Processors. pp. 440–443. doi:10.1109/ICCD.1991.139942. ISBN 0-8186-2270-9.
- ^ "AS/400e System Handbook" (PDF). IBM. 1999-08-03. p. 295. Retrieved 2021-03-21.
Version 4 of OS/400 does not run on earlier models of AS/400 based on IMPI processors.
- ^ "IBM i5/OS V5R3 — the next generation of OS/400". ibm.com. 2004-05-04. Retrieved 2021-03-14.
- ^ IBM i 6.1 Technical Overview (PDF). IBM. December 2009. p. 2.
IBM i 6.1 runs on all POWER5TM and POWER6® technology systems, as well as previous technology 9406 models 800, 810, 825, 870, and 890. ... IBM i 5.4 runs on all POWER5 and POWER6 technology systems, as well as previous technology 9406 models 800, 810, 820, 825, 830, 840, 870, 890, SB2 and SB3.
- To reiterate a point from a previous discussion - when i5/OS became IBM i, IBM renamed the two most recent i5/OS releases at that time - V5R4 and V6R1 to IBM i 5.4 and IBM i 6.1. There was no change in hardware support, just the name changed (see citations inline in article). The first release of i5/OS also ran on certain AS/400 systems.[1] The idea that the name changes indicate a dropping of support for previous hardware is incorrect.
References