I'm not sure I quite agree that is_deeply doesn't do a deep comparison at all, though I agree that it has a quirk or two. But more to the point, what does Test::Struct offer that Test::Deep doesn't? Not meant rhetorically, I'm just curious, since they both seem to have the same basic goal.

I think I should leave that up to Fergal to answer, however its my understanding that what T::S does is different from what T::D does (I know for sure that the latter is considerably more flexible and attempts to solve problems that T::S doesnt and never will).

Test::Struct was simply my attempt at making an is_deeply() that worked the way I expect it to. To save you reference to some very long and nasty perl-qa threads I'll simply say that as the author of the most accurate Perl Data serialization module currently on CPAN I found every other tool for testing insufficient for my purposes. Test::Struct is/was a prototype I wrote for the kind of testing I wished had been around for Data::Dump::Streamer. Some argue that this level of accuracy is only really useful for testing serialization modules but whatever. Anyway i wrote T::S just before i went on holiday (which im still on) so its in an alpha state right now, somethiong i plan to resolve when i return.

---
$world=~s/war/peace/g

TS is the strictest possible comparison that doesn't involve comparing reference addresses, so if $x=[];$y=[] then TS will say that [$x,$x] is not the same as [$x,$y] because one is an array with 2 refs to 1 array and the other has refs to 2 different arrays. This is a valid difference. TD ignores this difference. TD also lets you do something a little bit like regular expressions for data structures.