Dear all, GFSG discussed our IDL spec, and changed the status to "public comment". The following questions were raised, which should be answered as part of the public review. "Why no reference to SAGA or other standards work?" "IDL mapping is present. What mappings to other languages should be considered (perhaps for other documents)?" The first issue can be easily solved by adding more references to the document. The second issue shows a lack of understanding regarding our language binding documents strategy. We need to be more explicit here. The spec still does not show up on ogf.org. I will monitor this and inform you when the 60 days timer really starts. Regards, Peter.
What mappings to other languages should be considered (perhaps for other documents)?"
Just to clarify my second comment... [BTW see these as early public comments. They are NOT being a blocker to the document going into public comment - we decided that (I thought) on the call on Tuesday.] The IDL document makes it clear you what a mechanism for rendering into specific bindings... and you spend some space in the document showing it for Java. There is no demonstration (to the same level as the Java demonstration) how a binding in other languages would look. Steven
Steven Newhouse wrote:
What mappings to other languages should be considered (perhaps for other documents)?"
Just to clarify my second comment... [BTW see these as early public comments. They are NOT being a blocker to the document going into public comment - we decided that (I thought) on the call on Tuesday.]
The IDL document makes it clear you what a mechanism for rendering into specific bindings... and you spend some space in the document showing it for Java. There is no demonstration (to the same level as the Java demonstration) how a binding in other languages would look.
Thanks for the clarification ! We have a ready-to-publish Java binding, based on the IDL spec. I have the outline of .NET and Python bindings, based on the IDL spec. And we ensured that the C binding also can be derived from the IDL spec. Perl and Ruby are also candidates, because they are somehow comparable to Python. I can give details about the mappings on the public comment page. The fact is that DRMAA group decided to serialize the document submissions. Language bindings should rely on a publicly approved IDL spec, which itself relies on the DRMAA 1.0 grid recommendation. You can therefore expect several language binding submissions when the IDL spec reached 'proposed recommendation' status. Regards, Peter.
Peter, That's great to hear the other bindings are in the works. I'm obviously not expecting these to appear in full in the IDL spec. On reading it the motivation seemed to be 'very' Java oriented. Steven
-----Original Message----- From: Peter Tröger [mailto:peter.troeger@hpi.uni-potsdam.de] Sent: Friday, November 30, 2007 1:32 AM To: Steven Newhouse Cc: drmaa-wg@gridforum.org; Greg Newby Subject: Re: [DRMAA-WG] IDL spec comments from GFSG
Steven Newhouse wrote:
What mappings to other languages should be considered (perhaps for other documents)?"
Just to clarify my second comment... [BTW see these as early public comments. They are NOT being a blocker to the document going into public comment - we decided that (I thought) on the call on Tuesday.]
The IDL document makes it clear you what a mechanism for rendering into specific bindings... and you spend some space in the document showing it for Java. There is no demonstration (to the same level as the Java demonstration) how a binding in other languages would look.
Thanks for the clarification ! We have a ready-to-publish Java binding, based on the IDL spec. I have the outline of .NET and Python bindings, based on the IDL spec. And we ensured that the C binding also can be derived from the IDL spec. Perl and Ruby are also candidates, because they are somehow comparable to Python. I can give details about the mappings on the public comment page.
The fact is that DRMAA group decided to serialize the document submissions. Language bindings should rely on a publicly approved IDL spec, which itself relies on the DRMAA 1.0 grid recommendation. You can therefore expect several language binding submissions when the IDL spec reached 'proposed recommendation' status.
Regards, Peter.
On Nov 30, 2007 10:31 AM, Peter Tröger <peter.troeger@hpi.uni-potsdam.de> wrote:
Thanks for the clarification ! We have a ready-to-publish Java binding, based on the IDL spec.
Will they differ much from current DRMAA 1.0 language bindings (C, Java)? -- Piotr Domagalski
Dear Piotr,
Thanks for the clarification ! We have a ready-to-publish Java binding, based on the IDL spec.
Will they differ much from current DRMAA 1.0 language bindings (C, Java)?
The latest Java binding has one method renaming (without parameter change), in order to apply to the common IDL base. The SGE implementation will mark the old method name as deprecated and simply add the new one. This might be also feasible for you. As you can see in the last meeting minutes, DRMAA group is still discussing about the effect for the version numbers. The current advice is to stick with your implementation(s) until the C / Java language binding reached some OGF document status. Thanks and regards, Peter.
participants (3)
-
Peter Tröger -
Piotr Domagalski -
Steven Newhouse