> >I would have though the 5.005 thread adventure would have driven home
> >that multithreading cannot be made to 'magically' work. The onus
> >(of locking and unlocking) should be on the programmer, not on some
> >mystical hidden logic.
>
> Sorry, I'm not convinced that all the locking should be at the control
> of the programmer.
Where did you pull the "all" from?
Of course data *internal* to the lexer/parses/compiler/optimizer/interpreter,
that is, Perl, should be managed (including proper locking) by Perl.
But the moment someone suggests that access to *user* data should be
"transparently" mutexed or whatever by Perl, I wince and groan.
> Locking needed to avoid deadlocks should be at the control of the
> programmer. Any and all locking needed to avoid coredumps and memory
> corruption should be the sole province of the perl guts. :-)
>
> Sarathy
> [EMAIL PROTECTED]
--
$jhi++; # http://www.iki.fi/jhi/
# There is this special biologist word we use for 'stable'.
# It is 'dead'. -- Jack Cohen
