Obsolete reference to deprecated function MPI_Attr_get

[mpiforumbot]

Originally by rsthakur on 2009-04-05 12:11:40 -0500

---

On 260:8-9, when referring to predefined attributes, it says "The value of these attributes can be inquired by using the function MPI_ATTR_GET described in Chapter 6."

However, MPI_ATTR_GET is no longer in Chapter 6 (has been moved to Chapter 15), and in any case we don't want to encourage use of the deprecated function. Therefore, I propose the simple change of replacing MPI_ATTR_GET with MPI_COMM_GET_ATTR, which really is defined in Chapter 6.

History

Occurs as a result of changes in 2.1

Proposed Solution

pg 260, ln 8-9: change the sentence "The value of these attributes can be inquired by using the function MPI_ATTR_GET described in Chapter 6."

to

"The value of these attributes can be inquired by using the function MPI_COMM_GET_ATTR described in Chapter 6."

Impact on Implementations

None

Impact on Applications / Users

None

Alternative Solutions

None

Entry for the Change Log

None

[mpiforumbot]

Originally by rsthakur on 2009-04-05 12:12:25 -0500

---

Please review this simple fix.

[mpiforumbot]

Originally by gropp on 2009-04-05 12:41:00 -0500

---

Reviewed and fine (as long as it remains consistent with the rest of the attribute discussion; these are certainly fine in C/C++ in any event).

[mpiforumbot]

Originally by balaji on 2009-04-05 13:55:58 -0500

---

Reviewed. Looks good.

[mpiforumbot]

Originally by jsquyres on 2009-04-06 05:45:59 -0500

---

Looks good.

[mpiforumbot]

Originally by RolfRabenseifner on 2009-04-06 17:31:21 -0500

---

Review: Okay.

[mpiforumbot]

Originally by rsthakur on 2009-04-06 17:36:24 -0500

---

Marking this as reviewed.

[mpiforumbot]

Originally by traff on 2009-07-21 08:01:59 -0500

---

There's also a usage of MPI_Attr_get in the example on p.220-221 that should be updated.
Bear with me if a ticket has already been filed. Also, I'm not sure about the text at the end of p.468

Jesper

[mpiforumbot]

Originally by RolfRabenseifner on 2009-07-21 12:31:30 -0500

---

Example on p220-221 --> Ticket #12

End of page 468 --> which text and what should be wrong?
There is nothing related to this ticket - or?

Or do you mean page 486? See ticket #55, p486:44 reads ...

[mpiforumbot]

Originally by traff on 2009-07-22 03:00:55 -0500

---

Replying to RolfRabenseifner:

Example on p220-221 --> Ticket #12

well spotted!! thanks

End of page 468 --> which text and what should be wrong?
There is nothing related to this ticket - or?

The text p.486, line 44-48 explains the deprecated functions; partly this is superseded by p.487, line 4ff. Wouldn't it be more correct to at least explain the deprecated functions after the recommended ones? Ticket #55 partly addresses this, but the suggested addition on p.487 seems to introduce a lot of redundancy - something can surely be deleted? I'll try to have a closer look (with Jeff) at #55

Or do you mean page 486? See ticket #55, p486:44 reads ...

thanks again for the pointer, excellent!

jesper

[mpiforumbot]

Originally by RolfRabenseifner on 2009-07-26 14:46:01 -0500

---

Currently, the proposed text of this ticket contains only a reference to Chapter 6.

The handling of predefined attributes is done in Section 6.7 on page 221 and Section 16.3.7 on page 486, see especially p487:46-p488:3. And, ticket #55 enhances Section 16.3.7 and, e.g., clarifies usage of these predefined keys.

Therefore, it is not possible to change MPI_ATTR_GET into MPI_COMM_GET_ATTR without mentioning Section 16.3.7 on page 486.

-This is a serious bug of this ticket.*

This problem can be solved very easily by substituting "in Chapter 6" by "in Section 6.7 on page 221 and in Section 16.3.7 on page 486".

Because bugs should be solved before 2nd vote, I edited the proposed solution directly, see diff. I hope, this was okay.

-Please, owner and all reviewers of this ticket should comment it at least at the July meeting before 2nd vote is done.*

[mpiforumbot]

Originally by gropp on 2009-07-27 09:16:10 -0500

---

The routine MPI_COMM_GET_ATTR is defined in Chapter 6. 16.3.7 is about interlanguage access to attributes and while mentioning that at this point of the text is helpful, it is by no means required. Thus there was no bug in this ticket; the standard is to be read as a whole and it is not necessary for every cross reference to be completely inclusive.

[mpiforumbot]

Originally by gropp on 2009-07-27 17:52:00 -0500

---

Reverted to prevote text

[mpiforumbot]

Originally by bosilca on 2009-08-14 17:47:01 -0500

---

Attachment added: inquiry-ticket149.pdf (119.0 KiB)

[mpiforumbot]

Originally by bosilca on 2009-08-14 17:49:59 -0500

---

Revision r368 fixes this ticket.

[mpiforumbot]

Originally by rsthakur on 2009-08-14 20:07:32 -0500

---

George, it should be MPI_COMM_GET_ATTR not MPI_COMM_ATTR_GET.

[mpiforumbot]

Originally by RolfRabenseifner on 2009-08-20 08:59:16 -0500

---

Attachment added: ticket149.pdf (315.9 KiB)

[mpiforumbot]

Originally by RolfRabenseifner on 2009-08-20 09:00:18 -0500

---

I updated the pdf.

PDF review: now okay.

[mpiforumbot]

Originally by jsquyres on 2010-09-18 05:08:17 -0500

---

This ticket is (long-since) complete; marking it resolved/text committed.