I have discovered a couple of errors in the 0.7.1 spec, mostly related to exceptions. I also added a separate table for correlating IDL exceptions to Java exceptions. Hopefully I have now also completely removed all uses of the old (pre-0.4) naming from the spec. There is now one open issue that I will need to resolve with Peter. I believe that we agreed to add a NoMoreElementsException to the IDL spec to be thrown from the cursor functions instead of InvalidArgumentException when the iterator is exhausted. I do not, however, see that error code listed in the currect IDL spec, and I can't get to the tracker site at the moment to confirm my recollection. For now, the Java spec references this missing error code. If it turns out that I am misremembering the decision regarding this error code, I will remove the reference before I make the Java spec final. Daniel
Hi,
I have discovered a couple of errors in the 0.7.1 spec, mostly related to exceptions. I also added a separate table for correlating IDL exceptions to Java exceptions. Hopefully I have now also completely removed all uses of the old (pre-0.4) naming from the spec.
Great. I also got some feedback for IDL spec from HPI people, but mostly regarding formulations. I will release the final document after christmas, if there are no more issues found by somebody else on the list.
There is now one open issue that I will need to resolve with Peter. I believe that we agreed to add a NoMoreElementsException to the IDL spec to be thrown from the cursor functions instead of InvalidArgumentException when the iterator is exhausted. I do not, however, see that error code listed in the currect IDL spec, and I can't get to the tracker site at the moment to confirm my recollection. For now, the Java spec references this missing error code. If it turns out that I am misremembering the decision regarding this error code, I will remove the reference before I make the Java spec final.
I am somehow confused. I thought the NO_MORE_ELEMENTS error is only needed for the string vector helper functions in the C binding. Java and friends have native vector types, so there is simply no need for the helper functions, and therefore also no need for this error. Right ?!? Peter.
Peter, You are not wrong about the DRMAA_ERRNO_NO_MORE_ELEMENTS being only for non-object-oriented language-without-native-lists bindings. The IDL spec should, however, include enough detail to make it possible to generate the various other language bindings, including the non-OO bindings. (The reference to the error in the Java spec it to say that is has no mapping.) OK. Now I'm confused. I just looked through the latest IDL spec, and I no longer see the references to what to do when your language doesn't have exceptions or what to do when your language is not introspective. Has the IDL spec become the OO, introspective IDL spec, i.e the C#/Java spec? Daniel Peter Troeger wrote:
Hi,
I have discovered a couple of errors in the 0.7.1 spec, mostly related to exceptions. I also added a separate table for correlating IDL exceptions to Java exceptions. Hopefully I have now also completely removed all uses of the old (pre-0.4) naming from the spec.
Great. I also got some feedback for IDL spec from HPI people, but mostly regarding formulations. I will release the final document after christmas, if there are no more issues found by somebody else on the list.
There is now one open issue that I will need to resolve with Peter. I believe that we agreed to add a NoMoreElementsException to the IDL spec to be thrown from the cursor functions instead of InvalidArgumentException when the iterator is exhausted. I do not, however, see that error code listed in the currect IDL spec, and I can't get to the tracker site at the moment to confirm my recollection. For now, the Java spec references this missing error code. If it turns out that I am misremembering the decision regarding this error code, I will remove the reference before I make the Java spec final.
I am somehow confused. I thought the NO_MORE_ELEMENTS error is only needed for the string vector helper functions in the C binding. Java and friends have native vector types, so there is simply no need for the helper functions, and therefore also no need for this error. Right ?!?
Peter. -- drmaa-wg mailing list drmaa-wg@ogf.org http://www.ogf.org/mailman/listinfo/drmaa-wg
Peter, I thought that we had also agreed that there should be some text explicitly discussing what happens (or is not guaranteed to happen) when a control(SESSION_ALL) call fails. I don't see that in the control() method description. Daniel Daniel Templeton wrote:
Peter,
You are not wrong about the DRMAA_ERRNO_NO_MORE_ELEMENTS being only for non-object-oriented language-without-native-lists bindings. The IDL spec should, however, include enough detail to make it possible to generate the various other language bindings, including the non-OO bindings. (The reference to the error in the Java spec it to say that is has no mapping.)
OK. Now I'm confused. I just looked through the latest IDL spec, and I no longer see the references to what to do when your language doesn't have exceptions or what to do when your language is not introspective. Has the IDL spec become the OO, introspective IDL spec, i.e the C#/Java spec?
Daniel
Peter Troeger wrote:
Hi,
I have discovered a couple of errors in the 0.7.1 spec, mostly related to exceptions. I also added a separate table for correlating IDL exceptions to Java exceptions. Hopefully I have now also completely removed all uses of the old (pre-0.4) naming from the spec.
Great. I also got some feedback for IDL spec from HPI people, but mostly regarding formulations. I will release the final document after christmas, if there are no more issues found by somebody else on the list.
There is now one open issue that I will need to resolve with Peter. I believe that we agreed to add a NoMoreElementsException to the IDL spec to be thrown from the cursor functions instead of InvalidArgumentException when the iterator is exhausted. I do not, however, see that error code listed in the currect IDL spec, and I can't get to the tracker site at the moment to confirm my recollection. For now, the Java spec references this missing error code. If it turns out that I am misremembering the decision regarding this error code, I will remove the reference before I make the Java spec final.
I am somehow confused. I thought the NO_MORE_ELEMENTS error is only needed for the string vector helper functions in the C binding. Java and friends have native vector types, so there is simply no need for the helper functions, and therefore also no need for this error. Right ?!?
Peter. -- drmaa-wg mailing list drmaa-wg@ogf.org http://www.ogf.org/mailman/listinfo/drmaa-wg
-- drmaa-wg mailing list drmaa-wg@ogf.org http://www.ogf.org/mailman/listinfo/drmaa-wg
Peter, Sorry. I leaped before looking. The text that I was expecting to find is under the SESSION_ALL description. I think it would be useful to replicate that text under the control() method as well, or perhaps more it there completely. Daniel Daniel Templeton wrote:
Peter,
I thought that we had also agreed that there should be some text explicitly discussing what happens (or is not guaranteed to happen) when a control(SESSION_ALL) call fails. I don't see that in the control() method description.
Daniel
Daniel Templeton wrote:
Peter,
You are not wrong about the DRMAA_ERRNO_NO_MORE_ELEMENTS being only for non-object-oriented language-without-native-lists bindings. The IDL spec should, however, include enough detail to make it possible to generate the various other language bindings, including the non-OO bindings. (The reference to the error in the Java spec it to say that is has no mapping.)
OK. Now I'm confused. I just looked through the latest IDL spec, and I no longer see the references to what to do when your language doesn't have exceptions or what to do when your language is not introspective. Has the IDL spec become the OO, introspective IDL spec, i.e the C#/Java spec?
Daniel
Peter Troeger wrote:
Hi,
I have discovered a couple of errors in the 0.7.1 spec, mostly related to exceptions. I also added a separate table for correlating IDL exceptions to Java exceptions. Hopefully I have now also completely removed all uses of the old (pre-0.4) naming from the spec.
Great. I also got some feedback for IDL spec from HPI people, but mostly regarding formulations. I will release the final document after christmas, if there are no more issues found by somebody else on the list.
There is now one open issue that I will need to resolve with Peter. I believe that we agreed to add a NoMoreElementsException to the IDL spec to be thrown from the cursor functions instead of InvalidArgumentException when the iterator is exhausted. I do not, however, see that error code listed in the currect IDL spec, and I can't get to the tracker site at the moment to confirm my recollection. For now, the Java spec references this missing error code. If it turns out that I am misremembering the decision regarding this error code, I will remove the reference before I make the Java spec final.
I am somehow confused. I thought the NO_MORE_ELEMENTS error is only needed for the string vector helper functions in the C binding. Java and friends have native vector types, so there is simply no need for the helper functions, and therefore also no need for this error. Right ?!?
Peter. -- drmaa-wg mailing list drmaa-wg@ogf.org http://www.ogf.org/mailman/listinfo/drmaa-wg
-- drmaa-wg mailing list drmaa-wg@ogf.org http://www.ogf.org/mailman/listinfo/drmaa-wg
Peter, One more comment. The text in the wait() description that says that after a successful call to wait(), all subsequent calls to wait will fail, should probably say that all subsequent calls to wait(), synchronize(), control(), and getJobProgramStatus() will fail. Daniel Daniel Templeton wrote:
Peter,
Sorry. I leaped before looking. The text that I was expecting to find is under the SESSION_ALL description. I think it would be useful to replicate that text under the control() method as well, or perhaps more it there completely.
Daniel
Daniel Templeton wrote:
Peter,
I thought that we had also agreed that there should be some text explicitly discussing what happens (or is not guaranteed to happen) when a control(SESSION_ALL) call fails. I don't see that in the control() method description.
Daniel
Daniel Templeton wrote:
Peter,
You are not wrong about the DRMAA_ERRNO_NO_MORE_ELEMENTS being only for non-object-oriented language-without-native-lists bindings. The IDL spec should, however, include enough detail to make it possible to generate the various other language bindings, including the non-OO bindings. (The reference to the error in the Java spec it to say that is has no mapping.)
OK. Now I'm confused. I just looked through the latest IDL spec, and I no longer see the references to what to do when your language doesn't have exceptions or what to do when your language is not introspective. Has the IDL spec become the OO, introspective IDL spec, i.e the C#/Java spec?
Daniel
Peter Troeger wrote:
Hi,
I have discovered a couple of errors in the 0.7.1 spec, mostly related to exceptions. I also added a separate table for correlating IDL exceptions to Java exceptions. Hopefully I have now also completely removed all uses of the old (pre-0.4) naming from the spec.
Great. I also got some feedback for IDL spec from HPI people, but mostly regarding formulations. I will release the final document after christmas, if there are no more issues found by somebody else on the list.
There is now one open issue that I will need to resolve with Peter. I believe that we agreed to add a NoMoreElementsException to the IDL spec to be thrown from the cursor functions instead of InvalidArgumentException when the iterator is exhausted. I do not, however, see that error code listed in the currect IDL spec, and I can't get to the tracker site at the moment to confirm my recollection. For now, the Java spec references this missing error code. If it turns out that I am misremembering the decision regarding this error code, I will remove the reference before I make the Java spec final.
I am somehow confused. I thought the NO_MORE_ELEMENTS error is only needed for the string vector helper functions in the C binding. Java and friends have native vector types, so there is simply no need for the helper functions, and therefore also no need for this error. Right ?!?
Peter. -- drmaa-wg mailing list drmaa-wg@ogf.org http://www.ogf.org/mailman/listinfo/drmaa-wg
-- drmaa-wg mailing list drmaa-wg@ogf.org http://www.ogf.org/mailman/listinfo/drmaa-wg
-- drmaa-wg mailing list drmaa-wg@ogf.org http://www.ogf.org/mailman/listinfo/drmaa-wg
I added an explicit reference to the SESSION_ALL description in the control() and synchronize() text blocks. This avoids triplication of the error condition description. Peter. Am 23.12.2006 um 00:20 schrieb Daniel Templeton:
Peter,
Sorry. I leaped before looking. The text that I was expecting to find is under the SESSION_ALL description. I think it would be useful to replicate that text under the control() method as well, or perhaps more it there completely.
Daniel
Daniel Templeton wrote:
Peter,
I thought that we had also agreed that there should be some text explicitly discussing what happens (or is not guaranteed to happen) when a control(SESSION_ALL) call fails. I don't see that in the control() method description.
Daniel
Daniel Templeton wrote:
Peter,
You are not wrong about the DRMAA_ERRNO_NO_MORE_ELEMENTS being only for non-object-oriented language-without-native-lists bindings. The IDL spec should, however, include enough detail to make it possible to generate the various other language bindings, including the non-OO bindings. (The reference to the error in the Java spec it to say that is has no mapping.)
OK. Now I'm confused. I just looked through the latest IDL spec, and I no longer see the references to what to do when your language doesn't have exceptions or what to do when your language is not introspective. Has the IDL spec become the OO, introspective IDL spec, i.e the C#/Java spec?
Daniel
Peter Troeger wrote:
Hi,
I have discovered a couple of errors in the 0.7.1 spec, mostly related to exceptions. I also added a separate table for correlating IDL exceptions to Java exceptions. Hopefully I have now also completely removed all uses of the old (pre-0.4) naming from the spec.
Great. I also got some feedback for IDL spec from HPI people, but mostly regarding formulations. I will release the final document after christmas, if there are no more issues found by somebody else on the list.
There is now one open issue that I will need to resolve with Peter. I believe that we agreed to add a NoMoreElementsException to the IDL spec to be thrown from the cursor functions instead of InvalidArgumentException when the iterator is exhausted. I do not, however, see that error code listed in the currect IDL spec, and I can't get to the tracker site at the moment to confirm my recollection. For now, the Java spec references this missing error code. If it turns out that I am misremembering the decision regarding this error code, I will remove the reference before I make the Java spec final.
I am somehow confused. I thought the NO_MORE_ELEMENTS error is only needed for the string vector helper functions in the C binding. Java and friends have native vector types, so there is simply no need for the helper functions, and therefore also no need for this error. Right ?!?
Peter. -- drmaa-wg mailing list drmaa-wg@ogf.org http://www.ogf.org/mailman/listinfo/drmaa-wg
-- drmaa-wg mailing list drmaa-wg@ogf.org http://www.ogf.org/mailman/listinfo/drmaa-wg
Peter, Good idea. :) Daniel Peter Troeger wrote:
I added an explicit reference to the SESSION_ALL description in the control() and synchronize() text blocks. This avoids triplication of the error condition description.
Peter.
Am 23.12.2006 um 00:20 schrieb Daniel Templeton:
Peter,
Sorry. I leaped before looking. The text that I was expecting to find is under the SESSION_ALL description. I think it would be useful to replicate that text under the control() method as well, or perhaps more it there completely.
Daniel
Daniel Templeton wrote:
Peter,
I thought that we had also agreed that there should be some text explicitly discussing what happens (or is not guaranteed to happen) when a control(SESSION_ALL) call fails. I don't see that in the control() method description.
Daniel
Daniel Templeton wrote:
Peter,
You are not wrong about the DRMAA_ERRNO_NO_MORE_ELEMENTS being only for non-object-oriented language-without-native-lists bindings. The IDL spec should, however, include enough detail to make it possible to generate the various other language bindings, including the non-OO bindings. (The reference to the error in the Java spec it to say that is has no mapping.)
OK. Now I'm confused. I just looked through the latest IDL spec, and I no longer see the references to what to do when your language doesn't have exceptions or what to do when your language is not introspective. Has the IDL spec become the OO, introspective IDL spec, i.e the C#/Java spec?
Daniel
Peter Troeger wrote:
Hi,
I have discovered a couple of errors in the 0.7.1 spec, mostly related to exceptions. I also added a separate table for correlating IDL exceptions to Java exceptions. Hopefully I have now also completely removed all uses of the old (pre-0.4) naming from the spec.
Great. I also got some feedback for IDL spec from HPI people, but mostly regarding formulations. I will release the final document after christmas, if there are no more issues found by somebody else on the list.
There is now one open issue that I will need to resolve with Peter. I believe that we agreed to add a NoMoreElementsException to the IDL spec to be thrown from the cursor functions instead of InvalidArgumentException when the iterator is exhausted. I do not, however, see that error code listed in the currect IDL spec, and I can't get to the tracker site at the moment to confirm my recollection. For now, the Java spec references this missing error code. If it turns out that I am misremembering the decision regarding this error code, I will remove the reference before I make the Java spec final.
I am somehow confused. I thought the NO_MORE_ELEMENTS error is only needed for the string vector helper functions in the C binding. Java and friends have native vector types, so there is simply no need for the helper functions, and therefore also no need for this error. Right ?!?
Peter. -- drmaa-wg mailing list drmaa-wg@ogf.org http://www.ogf.org/mailman/listinfo/drmaa-wg
-- drmaa-wg mailing list drmaa-wg@ogf.org http://www.ogf.org/mailman/listinfo/drmaa-wg
OK. Now I'm confused. I just looked through the latest IDL spec, and I no longer see the references to what to do when your language doesn't have exceptions or what to do when your language is not introspective. Has the IDL spec become the OO, introspective IDL spec, i.e the C#/Java spec?
Ok, I scanned the IDL / OO-binding discussions of the last 12 months. At some point in time, we decided (informally) to concentrate the IDL spec on languages with support for 'OO' features. This was reasoned by two facts: - The C binding already relies on the language-independent spec. - The language-independent 1.0 spec and the IDL 1.0 spec are equivalent and exist side-by-side. Therefore, all C-like languages (BTW, which else do we have?) can rely on the language-independent spec. The other solution might be to add a special section for C-like languages in the IDL spec. This section could introduce the necessary additional constructs (like ERROR_NO_MORE_ELEMENTS) and rules (mapping of exception names to error code names). Regards, Peter.
Uh... I was supposed to send out a proposal on wording for some addition to the IDL spec, per the con call today, but I am no longer certain what it was. Was it the optional drmaa_job_ps states? If so, here it is: A DRMAA implementation is not required to be able to return all of the described jobs states. If a given state has no representation in the underlying DRMS, the DRMAA implementation is free to ignore that state. Daniel Peter Troeger wrote:
OK. Now I'm confused. I just looked through the latest IDL spec, and I no longer see the references to what to do when your language doesn't have exceptions or what to do when your language is not introspective. Has the IDL spec become the OO, introspective IDL spec, i.e the C#/Java spec?
Ok, I scanned the IDL / OO-binding discussions of the last 12 months. At some point in time, we decided (informally) to concentrate the IDL spec on languages with support for 'OO' features. This was reasoned by two facts:
- The C binding already relies on the language-independent spec. - The language-independent 1.0 spec and the IDL 1.0 spec are equivalent and exist side-by-side.
Therefore, all C-like languages (BTW, which else do we have?) can rely on the language-independent spec.
The other solution might be to add a special section for C-like languages in the IDL spec. This section could introduce the necessary additional constructs (like ERROR_NO_MORE_ELEMENTS) and rules (mapping of exception names to error code names).
Regards, Peter. -- drmaa-wg mailing list drmaa-wg@ogf.org http://www.ogf.org/mailman/listinfo/drmaa-wg
Hmmm... Let me rephrase that: A DRMAA implementation is not required to be able to return all of the job state values in the JobProgramState enumeration. If a given job state has no representation in the underlying DRMS, the DRMAA implementation MAY ignore that job state value. All DRMAA implementations MUST, however, define the JobProgramState enumeration, and the definition MUST include *all* job state values, including those for unused job states. Daniel Daniel Templeton wrote:
Uh... I was supposed to send out a proposal on wording for some addition to the IDL spec, per the con call today, but I am no longer certain what it was. Was it the optional drmaa_job_ps states? If so, here it is:
A DRMAA implementation is not required to be able to return all of the described jobs states. If a given state has no representation in the underlying DRMS, the DRMAA implementation is free to ignore that state.
Daniel
Peter Troeger wrote:
OK. Now I'm confused. I just looked through the latest IDL spec, and I no longer see the references to what to do when your language doesn't have exceptions or what to do when your language is not introspective. Has the IDL spec become the OO, introspective IDL spec, i.e the C#/Java spec?
Ok, I scanned the IDL / OO-binding discussions of the last 12 months. At some point in time, we decided (informally) to concentrate the IDL spec on languages with support for 'OO' features. This was reasoned by two facts:
- The C binding already relies on the language-independent spec. - The language-independent 1.0 spec and the IDL 1.0 spec are equivalent and exist side-by-side.
Therefore, all C-like languages (BTW, which else do we have?) can rely on the language-independent spec.
The other solution might be to add a special section for C-like languages in the IDL spec. This section could introduce the necessary additional constructs (like ERROR_NO_MORE_ELEMENTS) and rules (mapping of exception names to error code names).
Regards, Peter. -- drmaa-wg mailing list drmaa-wg@ogf.org http://www.ogf.org/mailman/listinfo/drmaa-wg
-- drmaa-wg mailing list drmaa-wg@ogf.org http://www.ogf.org/mailman/listinfo/drmaa-wg
participants (2)
-
Daniel Templeton -
Peter Troeger