[Cfp-interest] WG14 IEEE 754-C binding meeting minutes 2014/07/12

[Rajan Bhakta]

2014/06/12, 12:00 EST:

  Attendees: Rajan, Fred, Marius, David, Jim, Ian, Mike

  New agenda items:
    None.
 
  Old action items:
    David: Part 5: (From last meeting): Complete exception specification 
with the full syntax dealing with scope and sub-exceptions. Include a 
discussion document with reasons choices and alternatives. - Partially 
done (more of an outline. Sent on 2014/05/12). Keep open.
    David: Part 5: For fast_subnormal, change to "allows abrupt underflow" 
-> "allows (but not requires) abrupt underflow". Done.
    David: Part 5: Put in exception category in the sub-exceptions as the 
prefix. Ex. FE_DIV_ZERO -> FE_DIVBYZERO_DIVIDES. Done.
    David: Part 5: Add in FE_INVALID_OTHER and FE_DIVBYZERO_OTHER. Done.
    David: Part 5: OPTFLAG -> MAY_FLAG, NOFLAG -> NO_FLAG, SUBSTITUTEXOR 
-> SUBSTITUTE_XOR. Done first two, last is pending issue resolution.

  Next meeting:
    July 15th (Tuesday), 2014, 12:00 EST, 9:00 PDT
    Same teleconference number.

  New action items:
    David: Syntax.txt: Add in the beginning something that gives the 
purpose of the document. Ex. The CFP group is asking for feedback from 
WG14 for ...
    David: Syntax.txt: Semantics: Change exception1/exception2 to 
exceptions1/exceptions2
    David: Syntax.txt: Add to the end of the document other ideas 
considered.
    David: Syntax.txt: Add a sentence to handle thread and object state 
considerations.
    David: Syntax.txt: Add a sentence about ASAP vs deferred exception 
handling (try-catch vs try-patch).
 
  Discussion:
    Part 1: Reviewing draft as edited by ISO. Jim has sent back comments 
on changes made.

    Part 2: DTS ballot issued.

    Part 3: PDTS ballot issued.

    Part 4: PDTS ballot issued.
 
    Part 5: (Email discussion based) (http://www.validlab.com/cfp/*.txt)
      http://www.validlab.com/cfp/syntax.txt
        *ToDo: David: Add in the beginning something that gives the 
purpose of the document. Ex. The CFP study group is asking for feedback 
from WG14 for ...
        *ToDo: David: Semantics: Change exception1/exception2 to 
exceptions1/exceptions2
        Should we add the other things we considered like:
          1) PL/I, Basic and other languages "ON event DO/GOTO/GOSUB 
handler"
          2) Exception handler callback function registration (like signal 
handlers)
          3) Using SIGFPE signal handling
          4) Testing flags with the existing C floating point environment 
handling functions and adding in new flags like underflow
        at the end of the document? Yes
        *ToDo: David: Add to the end of the document other ideas 
considered.
        Mike: Avoiding extra nesting is advantageous if we don't do 
try/catch.
        Fred: How does what we do work with threads and object state?
          Rajan: Should be written/stored to variables in the scope of the 
exception handling be in an indeterminate state only?
          Rajan: Also side effects are not known in sequencing.
          Yes, we should cover both cases for object state.
          For threads, this (exception handling) should be thread local 
like the existing C Standard floating point environment being thread 
local.
          Fred: Signal handling in a multi-threaded program results in 
undefined behaviour.
        David will make the changes discussed and post it for review 
before sending it out to the wider WG14 group.
      Substitution (pre/xor):
        David: Haven't looked at pre-substitution since it is cheaper to 
do exception/test-and-branch.
          Ex. Instead of "SUBSTITUTE(z = x / y, DIV_ZERO, z = x' / y')" do 
"try { z = x / y; } catch (DIV_ZERO) { z = x' / y' }"
        Easier to get one set of changes made rather than new syntax for 
each part with their attendant issues and potential problems. This means 
the exception handling approach is better.
      Raise-no-flag:
        David: Requires trapping, otherwise flags would be raised.
        Shows that the programmer has explicitly thought about the flags. 
i.e. opt-out vs opt-in
        Can be done with a pragma or try/catch (if catch does not raise a 
flag for the caught exception).
      We can document how to use the exception handling (ex. try-catch) to 
handle a lot of these items (substitution, may-flag, raise-no-flag, etc.).
      Should we have two versions (asap-try-catch vs try-catch)? An ASAP 
and a deferred one?
        Perhaps have different catch's? catch vs deferred_catch/patch or a 
#pragma that identifies the following code as ASAP or not.
        Loops (indeterminate or large limits) may want to do asap while 
smaller ones may want deferred. Should we give the compiler hints or 
directives for this?
        We should do both mechanisms since not doing asap means you can 
use the flag checking instead.
        catch can be implemented in patch, except for exact underflow

Regards,

Rajan Bhakta
z/OS XL C/C++ Compiler Technical Architect
ISO C Standards Representative for Canada
C Compiler Development
Contact: rbhakta at us.ibm.com, Rajan Bhakta/Houston/IBM
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.oakapple.net/pipermail/cfp-interest/attachments/20140612/9aef6913/attachment.html