Use of the information contained in this unapproved document is at your own risk
.Last update: 10 April,2001
1003.1-96 #110
_____________________________________________________________________________
Interpretation Number: XXXX
Topic: raise vs asynchronous signal
Relevant Sections: 2.2.2.10 , p14 l 120
PASC Interpretation Request: (Defect Report)
----------------------------
Date: 2000 June 16
Ref: XBD 25
------------------------------------------------------------------------
7 Defect Report concerning (number and title of International Standard
or DIS final text, if applicable):
IEEE Std 1003.1-1996 (incorporates 1003.1-1990, 1003.1b-1993, 1003.1c-1995,
1003.1i-1995) (ISO 9945-1:1996)
------------------------------------------------------------------------
8 Qualifier (e.g. error, omission, clarification required):
3
Error=1 , Omission=2, Clarification=3
------------------------------------------------------------------------
9 References in document (e.g. page, clause, figure, and/or table
numbers):
P 14, Line 120
------------------------------------------------------------------------
10 Nature of defect (complete, concise explanation of the perceived
problem):
Is the result of raise () (or kill() of self) a synchronous
or asynchronous signal?
------------------------------------------------------------------------
11 Solution proposed by the submitter (optional):
This is to bring work in scope for the Austin Group revision.
The problem is that synchronous and asynchronous signals are not
quite well defined, particularly in the case of raise().
The current wording is "a signal that is not attributable to a specific
thread". Since all signals are generated by SOME thread, they are all
attributable to a specific thread (just not neccessarily the one
handling the signal). I believe the thrust of it is
"attributable to the thread handling the signal".
Right now, "synchronously generated signal" is being treated as
a synonym for "execution fault signal", which it isn't, at least
according to the definition of the word "synchronously", because
raise() would also be synchronously generated.
Suggested alternative wording:
Asynchronously generated signal:
"A signal that is not attributable to the actions of the thread
handling the signal."
and
Synchronously generated signal:
"A signal that is attributable to the thread which is
handling it, and which is handled immediately upon its generation;
that is, is not blocked or ignored.
Note: timer and I/O completion signals are considered attributable
to the timer or I/O system."
Note... there are two intertwined issues here; please address both
of them:
1) "attributable to a process" is trivially true; it's "the same
thread" that's critical to fixing this.
2) "handled immediately". This certainly is the intent, but it
could be circumvented. Doing a raise() of a signal known to be
blocked is not an unreasonable programming technique, but it
wouldn't be a synchronous signal. On the other hand, doing
a raise() of an unblocked signal is synchronous with the program.
(Or... rename the concept to "error signal" or something like that.)
------------------------------------------------------------------------
Interpretation response
------------------------
The answer to the question in the problem statement is "synchronous".
The problem raised in the "solution" section is being referred
to the sponsor.
Rationale
-------------
New definitions are required, those proposed will not work, this is
referred to the sponsor.
Forwarded to Interpretations group: 19 June 2000
Proposed resolution: 25 July 2000
Finalised Interpretation: 29 August 2000