attached patch has 2 new tests - 1st appears to show that the -V:<regex> construct is safe from abuses like -V:(?{system 'rm -rf /'}) the 2nd exhibits a regex parse problem I cant fathom. ok 24 - regex protected against cmdline DOS # /usr/local/bin/perl "-I../lib" '-V:abuse.*(?{print qq{Danger Will Robinson!}})' Sequence (?{...}) not terminated or not {}-balanced in regex; marked by <-- HERE in m/^abuse.*(?{ <-- HERE print=/ at ../lib/Config.pm line 1244. Attempt to free unreferenced scalar. res: not ok 25 - paranoid taint # TODO borked due to quoting errors, or something else ?? # Failed at t/run/switches.t.new line 246 # got '' # expected /(?-xism:)/ I saw various regex Sequence tests, but they dont help me see why mine is failing. ext/re/re_comp.c: vFAIL("Sequence (?{...}) not terminated or not {}-balanc the patch also contains a commented out test - for an embedding-friendly flavor of -V:ccflags: that I sent some weeks ago (9/24). While Yitzchak liked it, Robin also offered a nearby patch, that AMS didnt like. This patch debunks my notions of re-proposing that previous patch under the 'security' rubrik. thx.Thread Next