home *** CD-ROM | disk | FTP | other *** search
- From: <jsh@usenix.org>
-
-
- An Update on UNIX* and C Standards Activities
-
- January 1990
-
- USENIX Standards Watchdog Committee
-
- Jeffrey S. Haemer, Report Editor
-
- IEEE 1003.4: Real-time Extensions Update
-
- Rick Greer <rick@ism.isc.com> reports on the January 8-12, 1990
- meeting in New Orleans, LA:
-
- 1003.4 goes to ballot
-
- The big news in 1003.4 is that some of it is ready for balloting. My
- copy of the dot-4 ballot was waiting for me when I got back from New
- Orleans. The current draft is a 330-page, eclectic collection of
- real-time features. Some (e.g., asynchronous event notification)
- address significant deficiencies in the dot-1 base, but others (e.g.,
- IPC message passing) seem to be of limited value. It remains to be
- seen whether the limited applicability of some of the proposed
- features is enough to shoot down the entire ballot.
-
- One area that may cause trouble is the shared-memory model in the
- Language-Specific Requirements section. While this language-
- independent model addresses a real need--serialization of reads and
- writes in the presence of simultaneous updates to a common store--it
- does so rather formally; people uncomfortable with formal,
- mathematical models may be put off by it. The fact remains, however,
- that both dot 1 and the ANSI C standard failed to address this
- problem, which is critically important in shared-memory multiprocessor
- architectures.
-
- Threads
-
- The threads proposal is only an appendix in the current draft, and
- won't be subject to formal ballot. Though there were too many loose
- ends in the threads proposal to send it to ballot in this round, most
- of them were tied up in New Orleans. We should have a ballotable
- draft ready after the April meeting.
-
- Meanwhile, the active membership in the threads "small group" is
- changing. Representation from the Ada community has grown from two to
- six; almost a quarter of the active membership is now familiar with
-
- __________
-
- * UNIX is a registered trademark of AT&T in the U.S. and other
- countries.
-
- January 1990 Standards Update IEEE 1003.4: Real-time Extensions
-
-
- - 2 -
-
- Ada and its multitasking model. Most threads people, including me,
- are also becoming active in the new multiprocessor study group.
-
- Discussion within the multiprocessor group promises to be quite
- lively, since the threads group's more contentious issues (e.g.,
- signals) were skirted by defining high-level interfaces, leaving
- details of low-level behavior unspecified. The multiprocessor group,
- on the other hand, must deal with the low-level behavior of
- multiprocessor configurations, and many of the old arguments have
- already re-surfaced (e.g., should signal state be maintained per-
- process or per-thread?). Using high-level interface specifications to
- dodge low-level implementation issues does have its problems, though.
- People unaware of more subtle implementation issues tend to view new,
- high-level interfaces as unnecessary complications. It's difficult to
- convince them that, even if consensus could be reached regarding the
- behavior of primitive functions, we would still need high-level
- interfaces (or rigid coding disciplines) to guarantee that
- independently developed routines use primitives consistently when
- addressing common problems. The real sticker here has been how to
- asynchronously terminate a thread and cause it to execute cleanup
- code. Everyone agrees that this is necessary. Some members,
- particularly those from AT&T/USO, feel that the best way to provide
- this facility is by minor enhancements to traditional UNIX signals,
- but most of the group feels that the best way to deal with notorious
- signal races in a uniform, language-independent manner, is to adopt a
- high-level interface, modeled after one used by DEC/SRC.
-
- 1003.4 turns into .4, .4A, .4B, .4C, and .14
-
- There are three other major, on-going efforts in dot 4: language-
- independent specification of the real-time extensions, identification
- and specification of other, important, non-threads, real-time
- extensions that didn't make it into the current ballot, and
- specification of a real-time application profile. The first is
- farthest along, but none is anywhere near completion. Recognizing
- that these efforts were separate from the current proposal, and from
- one another, the working group submitted four new Program Action
- Requests (PARs). The Sponsor Executive Committee (SEC) approved all
- four, and decided that the application-profile effort was so distinct
- that it needed a new number. The working group's five PARs are now
-
- the current ballot 1003.4
- threads 1003.4A
- language independence 1003.4B
- further real-time extensions 1003.4C
- real-time application profile 1003.14
-
- January 1990 Standards Update IEEE 1003.4: Real-time Extensions
-
-
- Volume-Number: Volume 18, Number 24
-
-