Issue 2676: Minimum CORBA and POA (mincorba-rtf) Source: (, ) Nature: Uncategorized Issue Severity: Summary: Summary: The Minimum CORBA submission describes exactly what should be present in minimum CORBA (basically CORBA 2.2 including the POA) in IDL/PIDL. However, the Java language mapping in CORBA 2.2 does not include the POA -> just the APIs for registering transient objects. One cannot even take recourse to CORBA 2.3 to get the language mapping, since much stuff (OBV, Java to IDL etc.) was added in the intervening time. There does not seem to be any existing document which documents a Java language mapping of CORBA 2.2 including POA without lots of other stuff. Resolution: Revised Text: Actions taken: May 31, 1999: received issue Discussion: End of Annotations:===== Sender: Chris.Smith@uab.ericsson.se Message-ID: <37523F1C.A806219C@uab.ericsson.se> Date: Mon, 31 May 1999 09:49:48 +0200 From: Chris Smith Organization: Ericsson Utvecklings AB X-Mailer: Mozilla 4.51C-CCK-MCD [en] (X11; I; SunOS 5.5.1 sun4u) X-Accept-Language: sv,en-US MIME-Version: 1.0 To: issues@omg.org, drs@nortel.com Subject: Issue - Minimum CORBA & Java Content-Type: multipart/mixed; boundary="------------B9045F9E280B96760B68D2C6" X-UIDL: 746a4650c2630fe0f3ed6f39d1e62422 For Minimum CORBA RTF.... The Minimum CORBA submission describes exactly what should be present in minimum CORBA (basically CORBA 2.2 including the POA) in IDL/PIDL. However, the Java language mapping in CORBA 2.2 does not include the POA -> just the APIs for registering transient objects. One cannot even take recourse to CORBA 2.3 to get the language mapping, since much stuff (OBV, Java to IDL etc.) was added in the intervening time. There does not seem to be any existing document which documents a Java language mapping of CORBA 2.2 including POA without lots of other stuff. There is a similar issue (but perhaps more easily solved) with minimum CORBA and the JDK. Since the JDK now contains org.omg.CORBA and org.omg.CORBA.portable packages which also contain some CORBA 2.3 parts, perhaps we should attempt to clarify which parts of these mapped APIs should be implemented by a minimum ORB, and how we should we deal with the APIs not included in minimum CORBA. I suspect we should live with the fact that these classes are now defined and present in the JDK, and so not create alternative versions, but we could define what system exceptions should be returned by non-minimum CORBA APIs..... Its probably reasonably important to get this sorted, since it is not unreasonable to imagine minimum ORBs being used in a target "embedded Java" environment. [] Chris.Smith.vcf From: Jeffrey Mischkinsky Message-Id: <199906020302.UAA24268@wheel.dcn.davis.ca.us> Subject: Re: issue 2676 -- Minimum CORBA RTF Issue To: mincorba-rtf@omg.org Date: Tue, 1 Jun 1999 20:02:29 -0700 (PDT) In-Reply-To: <4.1.19990601135641.00a86be0@emerald.omg.org> from "Juergen Boldt" at Jun 1, 99 01:57:24 pm X-Mailer: ELM [version 2.4 PL25] Content-Type: text X-UIDL: 08391d0270e6e9e973b0d3870c4c1a63 'Juergen Boldt' writes: > > This is issue # 2676 > > Minimum CORBA and POA > > The Minimum CORBA submission describes exactly what should > be present in minimum CORBA (basically CORBA 2.2 including > the POA) in IDL/PIDL. > > However, the Java language mapping in CORBA 2.2 > does not include the POA -> just the APIs for registering > transient objects. Which are now deprecated. > > One cannot even take recourse to CORBA 2.3 to get the > language mapping, since much stuff (OBV, Java to IDL etc.) > was added in the intervening time. There does not seem to be > any existing document which documents a Java language mapping > of CORBA 2.2 including POA without lots of other stuff. No, there is none. The unfortunate situation is that many would argue that CORBA 2.2 should never have been published as "a separate document" because it does not contain a consistent view of adopted specs. A lot of people have seent a great deal of time and energy to ensure that the document that goes by the name of CORBA2.3 contains a consistent view of adopted specs and RTF reports. I would suggest that we issue anRFP for "CORBA 2.3" as the route to updating the minimum corba spec. In particular it will have to take into account the fairly significant additions that were made for GIOP 1.2. This will also provide a way to deal with a consistent view of the Java language mapping. E.g. the version of the Java language mapping "published" in 2.2 did not contain the outputs from the Java 1.1 RTF. As far as i know, java orb vendors never shippped product based upon corba 2.2, but upon adopted specs that included outputs of the 1.1 and 1.2 (i think) RTFs. jeff > > -- Jeff Mischkinsky jmischki@dcn.davis.ca.us +1 530-758-9850 jeffm@inprise.com +1 650-358-3049 Sender: Chris.Smith@uab.ericsson.se Message-ID: <3754E055.E37CFE05@uab.ericsson.se> Date: Wed, 02 Jun 1999 09:42:13 +0200 From: Chris Smith Organization: Ericsson Utvecklings AB X-Mailer: Mozilla 4.51C-CCK-MCD [en] (X11; I; SunOS 5.5.1 sun4u) X-Accept-Language: sv,en-US MIME-Version: 1.0 To: Jeffrey Mischkinsky CC: mincorba-rtf@omg.org Subject: Re: issue 2676 -- Minimum CORBA RTF Issue References: <199906020302.UAA24268@wheel.dcn.davis.ca.us> Content-Type: multipart/mixed; boundary="------------F710101DF045B32FF74FBC8F" X-UIDL: d03c30b4467da521d97103f403b6d21d Jeffrey Mischkinsky wrote: > > > > One cannot even take recourse to CORBA 2.3 to get the > > language mapping, since much stuff (OBV, Java to IDL etc.) > > was added in the intervening time. There does not seem to be > > any existing document which documents a Java language mapping > > of CORBA 2.2 including POA without lots of other stuff. > > No, there is none. > > The unfortunate situation is that many would argue that CORBA 2.2 > should > never have been published as "a separate document" because it does > not > contain a consistent view of adopted specs. A lot of people have > seent > a great deal of time and energy to ensure that the document that > goes by > the name of CORBA2.3 contains a consistent view of adopted specs and > RTF > reports. > > I would suggest that we issue anRFP for "CORBA 2.3" as the route to > updating the minimum corba spec. In particular it will have to take > into > account the fairly significant additions that were made for GIOP > 1.2. > > This will also provide a way to deal with a consistent view of the > Java > language mapping. E.g. the version of the Java language mapping > "published" > in 2.2 did not contain the outputs from the Java 1.1 RTF. As far as > i know, > java orb vendors never shippped product based upon corba 2.2, but > upon adopted > specs that included outputs of the 1.1 and 1.2 (i think) RTFs. > I dont think things are quite as drastic as that Jeff. Minimum CORBA defines a subset of CORBA which is appropriate and highly suitable for a large number of environments which CORBA is not appropriate for. What we need to do (which is why I submitted this issue), is to establish a Java language mapping for the parts of CORBA present in minimum CORBA. For example, it seems (as a first shot) that CORBA 2.2, plus the conclusion of the POA part, plus perhaps the RTFs you describe could work. Obviously the recommendations and help of the Java language mapping heavyweights such as yourself, Colm etc will be necessary/useful. Note I am not necessarily suggesting alternative versions of any of the JDK classes, just a clarification in the minimum CORBA spec of what APIs are relevant, and which should be "NO_IMPLEMENT"ed. [] Chris.Smith.vcf Sender: Chris.Smith@uab.ericsson.se Message-ID: <37AABEC7.A01BB82C@uab.ericsson.se> Date: Fri, 06 Aug 1999 12:53:59 +0200 From: Chris Smith Organization: Ericsson Utvecklings AB X-Mailer: Mozilla 4.51C-CCK-MCD [en] (X11; I; SunOS 5.6 sun4u) X-Accept-Language: sv,en-US MIME-Version: 1.0 To: Jeffrey Mischkinsky CC: Dave Stringer , mincorba-rtf@omg.org Subject: Re: issue 2676: mincorba: POA & Java References: <199907282338.QAA25616@wheel.dcn.davis.ca.us> Content-Type: multipart/mixed; boundary="------------E609C9595907A9D403D30926" X-UIDL: f8c4e9ea6202c01e677144045bdf7b75 Sorry I didnt reply earlier on this -> I've been on vacation. Much as I would like there to be some other way to do this, I think Jeff is probably right. The thing to do is to start with the 2.3 IDL/Java mapping and subset away bits based on what has been adopted as Minimum CORBA (e.g. take away valueTypes etc.), and perhaps state which JDK APIs can just throw NO_IMPLEMENT (because they are not minimum CORBA). This seems a bit messy, but its a fact of life that we have to do something slightly unusual for minimumCORBA in a Java environment both because of the inconsistency in what was in CORBA 2.2, and because a minimum ORB in Java still has to work against the org.omg.CORBA classes defined in the JDK... Given theres no time left in this RTF, I agree that we should leave this unresolved for now, but it does need to be addressed in the next RTF. It shouldnt be SO hard. Cheers Chris > > The RTF report date approaches, so as we have only one issue > > I've left it until now so that we can have the customary > > panic - and maybe get some people off of the voting list ;-) > > > > 2676 is Chris' issue that a profile of CORBA 2.2 doesn't make > > sense for Java because the Java mapping published in 2.2 > > doesn't contain a Java POA etc. > > > > I could mischieviously note that the puvblished CORBA 2.2 > > doesn't contain Ada POAs nor Smalltalk POAs and so minimum > > CORBA isn't possible with these languages - Java is then > > just another of those languages of historical significance > > only - at least in the 2.2 context ! > > > > but, guessing that such an interpretation isn't palatable to > > everyone, I've thought a bit harder. Chris claims that there > > is no document that describes the Java POA (etc.) without > > bringing in a lot of stuff that is post CORBA 2.2. Trying to > > retrospectively untangles a full 2.2 Java out of that may be > > beyond the scope of an RTF (its debatable). > It's not so much that its out of the scope of an RTF (which it > probably is), but that even if you tried to do the work, it > wouldn't correspond to anything anyone is shipping. > > > > Bur I've found orbos/98-03-10 which would appear to describe > > about the right amount of Java and is phrased like a final > > submission - so I'm guessing it was actually adopted by vote > > (does anyone know?). > > It was, but then highly modified by the subsequent RTFs. I think > the problem you will find is that the actual products that are > being shipped do not correspond to the 2.2 Java chapter. It should > never have been published, as it didn't even include the outputs > from the Java RTFs that had already completed!! > > The Java products that people are shipping are the original lang > mapping plus some number of Java RTF outputs, depending upon what > was approved or likely to be approved at the time they cut their > release. > > Our next release, VBJ 4.0, now in beta, is based on CORBA 2.3. > We never had a product that corresponded to the inconsistent stuff > that was put into CORBA 2.2 > > So as a practical matter I think there really is no realistic choice > but > to start with the 2.3 IDL/java version and subset it via IDL > features > if necessary, e.g. valuetype support. > > Look at it this way: this is kind of an existence proof that FTF > process > can work: the IDL/Java + Java/POA + (RTFs==FTF) = IDL/Java in 2.3. > > jeff > > > > > So my proposed resolution is include this spec in the baseline > > from which the minimum CORBA profile is made. > > > > Comments please ... > > > > Dave > > > > -- > Jeff Mischkinsky > jmischki@dcn.davis.ca.us +1 530-758-9850 > jeffm@inprise.com +1 650-358-3049 [] Chris.Smith2.vcf