Hi, Andre, Thanks for your comments.
Hi Yusuke,
some comments:
3.7.1 Notes for Results The error code, GRPC_NOT_INITIALIZED, is not defined in the error reporting functions, in Section 4.7 of GFD-R.52. When the functions are called before grpc_initialize() is called, however, some GridRPC implementations might return GRPC_NOT_INITIALIZED.
-> is that something which should be fixed in the spec then? I.e are the implementations which return GRPC_NOT_INITIALIZED here otherwise non-compliant?
I talked to Hidemoto, a co-chair of GridRPC-WG and we agreed with you. This is a small glitch and should be fixed in the specification document. For the interoperability test document, I would like to add two test cases which check if the two error reporting functions of each implementation appropriately returns GRPC_NOT_INITIALIZED or not. I will send the revised test suites to this mailing list later.
"The differences are not beyond of this kind of learning but understandable."
-> I am not sure if that sentence makes sense.
How about writing the following? --- There are several minor differences among Ninf-G, GridSolve and DIET. The differences come from the basic design of each implementation. When users start to use one of implementations or move from another implementation to the other, they usually learn features of each GridRPC system. The differences are straightforwardly understood during the learning process. --- I think the largest difference is when GRPC_FUNCTION_NOT_FOUND and GRPC_SERVER_NOT_FOUND are possibly retruned (Pleaes see details in Section 3.2 and 3.3). This really depends on the system architecture.
"In addition, three behaviors are not defined in the specification."
-> I think we discussed about taht in Chapel Hill. I'd vote for writing:
"In addition, three behaviors are intentionally left unspecified in the specification."
That should make clear that this does not point to a failure in the interoperab. testing, but is the expected result.
Right. I will add the sentense to the conclusion.
A minor thing about layout: I alsways think that block justified documents look much prettier than left justified ones. That may be my personal preference, and come from the fact that block justified is the default in latex, whereas left justified is the default in Word. Anyway, wanted to mention it... :-)
Thanks for pointing this out. I prefer block justified, too. If no objections, I'll fix this.
Otherwise the doc looks good to me.
Cheers, Andre.
Thanks a lot.
Quoting [Yusuke Tanimura] (Feb 18 2007):
From: Yusuke Tanimura <yusuke.tanimura@aist.go.jp> To: gridrpc-wg@ogf.org Subject: [GRIDRPC-WG] Final call for comments
Dear all,
I modified several parts of the GridRPC interoperability test document according to ESTI comments and discussions at OGF19. Please read through the attachment and send any comments to the mailing list within a week.
If there is no further comments, the document will be resubmitted to the editor.
Thanks for your cooperation.
----------------------------------------------------- Yusuke Tanimura <yusuke.tanimura@aist.go.jp> Grid Technology Research Center, National Institute of AIST 1-1-1 Umezono, Tsukuba Central 2 Tsukuba City 305-8568, Japan TEL: +81-29-862-6703 / FAX: +81-29-862-6601 -- "So much time, so little to do..." -- Garfield
----------------------------------------------------- Yusuke Tanimura <yusuke.tanimura@aist.go.jp> Grid Technology Research Center, National Institute of AIST 1-1-1 Umezono, Tsukuba Central 2 Tsukuba City 305-8568, Japan TEL: +81-29-862-6703 / FAX: +81-29-862-6601