[m-rev.] Trivial diff: fix hlc.par segfaults

[Peter Ross]

Julien Fischer wrote:

>On Fri, 5 Nov 2004, Ralph Becket wrote:
>
>  
>
>>Simon Taylor, Friday,  5 November 2004:
>>    
>>
>>>That should be in BUGS, not NEWS.  It would also be helpful to describe
>>>the "low-level concurrency bugs", and say whether .par grades are actually
>>>usable.
>>>      
>>>
>>Good point.  The message is in NEWS not because anything has changed or
>>broken in the compiler/library, but because we no longer have any group
>>members who fully understand the .par code.  The .par grades are quite
>>usable.
>>
>>    
>>
>hlc.par.gc is quite usable; the others having varying degrees of problems.
>
>  
>
The advantage of the hlc.par.gc grade is its simplicity.  It simply 
starts a call in a new pthread.  The only other thing that had to change 
was that the exception handler pointer in the exception library used to 
be in a global variable, it is now in thread local storage.  That's all 
there is to the hlc.par.gc grade.

The lowlevel version of the grades are a lot more complicated, as Tom 
did a lot of work to allow parallel programs to work in the 
single-threaded grades.  Thus there is a lot more code to maintain, and 
I certainly don't understand it all.

So for me, I think it would be reasonable just to support the hlc.par.gc 
grade.

--------------------------------------------------------------------------
mercury-reviews mailing list
post:  mercury-reviews at cs.mu.oz.au
administrative address: owner-mercury-reviews at cs.mu.oz.au
unsubscribe: Address: mercury-reviews-request at cs.mu.oz.au Message: unsubscribe
subscribe:   Address: mercury-reviews-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------