> You're quite right that it doesn't, but neither does anything else > that creates temporary strings in a different encoding ;-)
In my day-or-two-old parrot copy, the only other code that uses the transcoding table only uses it with the second param != null (ie, save into existing string). Other places that call string_make use it to return a new string (ie, it's okay to leave it uncollected.). Although it looks like test_main.c could use a "string_destroy(t);" in there next to the "string_destroy(s);" :) > As we're using garbage collection we shouldn't need to do an explicit > free though surely - in fact I'm not quite sure why string_destroy > even exists... I'm not sure if the GC'ing will apply to strings, or just PMCs. I imagine PMC's will be able to contain strings, and could handle GC strings that way? Anyone care to shed some light on what the plan is? (No, I don't really want to start a GC war here...so if the plans are still up in the air, that's all the answer I'm looking for. :) Mike Lambert
