Maybe it's due to the conferences, but this week was a low-traffic one.
Peter Scott asks: if a tie constructor doesn't return a blessed reference, the resulting object is not tied, but perl doesn't produce any warning. Is this behaviour deliberate? Or, in other words, is there useful (and documented) cases where a tie constructor would like to decline the construction of a tied object.
Ton Hospel asks whether
SvPOK should return true on tainted strings. Nicholas Clark answers negatively, arguing that tainted strings have taint magic, meaning that their string value shouldn't be accessed directly. Paul Fenwick further comments on the handling on tainted scalars through C code.
Meanwhile, Tim Bunce suggests that adding a mechanism to taint everything would be a good way to flush out magic-related bugs (more precisely, missing
SvGETMAGIC calls to get string values).
Some regression tests for the newest versions of
I18N::LangInfo are failing for Rafael. Besides some corrections that have been made by Sean Burke to the tests, tests keeps failing, and a system problem is suspected.
Andrew Savige reports a case of random crash when sort() with a named subroutine is invoked from different threads. (Bug #30333.) I imagine that the implementation of sort() calling subroutines must be reviewed for thread-safety.
Paul Johnson wonders whether the deep recursion warning could be issued only for deeper recursions than currently. There is no compelling reason to so this, however.
Colin Watson reports that localising the
$? variable looses the exit value from the
perl command. This shouldn't occur, since
$? is documented to affect the exit status of
perl only within END blocks. (Bug #30296.)
This summary was written by Rafael Garcia-Suarez. Weekly summaries are published on http://use.perl.org/ and posted on a mailing list, which subscription address is email@example.com. Comments and corrections welcome.