Another probably well-Ninjaed reply from history (7/Jul/2010, to be precise, page 18, am trying not to post such necroginous replies more than once for every 10 pages I'm catching up on, so slightly worse than I intended. But this one was irresistable.).
The first priority will probably just be dealing with walled-off hallways and stuff like that though -- things that are really easy to do that kill off entire sieges. I guess it's entirely possible that they wouldn't be able to find you if you hid behind 10 meter thick walls hidden off in random places, but then it should be possible for them to live in your upper levels for years while you work away tradeless on mushrooms down below (although there are major obstacles to that that make it non-practical for the dev page). They'd probably just leave after looting all your exposed items and slaughtering anybody left outside (including any entity pop infrastructure you've got out there). Although once they can get through a single wall, they can just dig ambitious tunnels for you at random I suppose. As long as walling yourself in has reasonable results, I'll be happy with however they handle it. Right now it's too much of an exploit (of course, the whole "digging invaders" is enough of a touchy subject that it's explictly stated as optional on the dev page, and how you handle exploits is up to you at that point).
I would propose 2 systems to distinguish normal invaders from game ending HFS. It's realistic that if you hole up behind really really thick walls and never trade, then normal invaders don't have a way to realise there's dwarves hiding so they should leave you alone. At the same time, HFS shouldn't be stopped by walling it off no matter how thick the walls are.
Actually the HFS can have an alternative to digging to get around the current "wall off" exploit: what if they could pass through walls and floors? That is, they swim through rock just like they swim through magma? Then HFS would truly be the game enders and the existing infinite spawning of them would ensure forts die no matter what.
...
What would prevent the clowns from coming straight from the H.F.S. Clown Car to your wagon at embark? If they really want to Entertain (i.e., bring FunTM to) the world's children, and they can swim through rock...
I sort of had the idea that that was what ClownMetal was all about. Something HFS couldn't interact with. Either some form of their own private Kryptonite, or an Unobtanium shell impervious to their effects. (The fact that when the shell is breached they
currently are initially stopped by normal matter, or at least have no difficulty (that I know of) deconstructing constructions made of the self-same material as well as that of normal matter
I would treat as an anomaly that may change in the future.
And on another matter I was previously tempted to reply to. (Probably already addressed.)
If you shift from (a six-limbed creature with a wound to the middle left limb) to a human, I'd expect you actually shift to (a human with a significant wound on the of your lower abdomen/back). Bleeding is bleeding - you can't shift to a human and mosey off to the dang hospital.
My thoughts too. Heavy laceration to a limb that you 'reabsorb' and no longer have would be damage to the skin on the side of the torso that mapped directly (not necessarily precisely in code, just in general abstraction) to the where the limb emerged. A broken bone within the limb might relate to a rib in that location, but that's a dodgier relationship.
OTOH, while the talk (so far as I have read) has been a matter of defining primary and secondary upper limb-pairs, etc, and an assumption that a second set of arms (e.g. as an additional to the 'usual' tetrapedal limbs in a creature having a bipedal stance) would bud from the same part of the body as non-primary wings might (e.g. from the back/spine area of a tetrapod with quadrapedal stance). In some ways, this could be something we could make use of 'inconstencies' in the raws for.
e.g. A polymorphic creature in the form of a griffin or pegasus could have forelegs (left and right), hind legs (ditto) and wings (also paired). Changing into a centaur, are their forelegs now their arms and the wings the forelegs? The wings migrated 'up' the torso and are the arms, with the quadruped stance staying as-was? I'd be tempted for the wings (including any damage) to retreat into the mid-torso while the neck of the griffin/pegasus morph into the torso (sketching over, very briefly, whether the horse/lionish-ish arrangement of internal organs needs redistrubiting or doubling up into the 'new' chest cavity area) and buds its arms. Injuries to the pegasus wings would be reflected in a back injuries of the centaur. Injuries to the arms of the centaur might be neck/upper shoulder injuries in the pegasus form. But add a humaniform into that equation. Human arms == centaur arms? Or pegasus forelegs? Arghh! Unless you are happy for griffon wings to traverse down into pegasus forelimbs (morphing as they go) the only real solution is that both pegasus and centaur forelimbs 'debud' into the pelvis that the entire quadrapedal torso concertinas into while transforming the four lower limbs (and 'main' trunk of the body) into a two-legged lower half.
From a DF POV, this doesn't need much actual thought (at least not until some 3D graphical revamp of the future, if it remains unobstructed by a 'modesty shimmer' or similar effect). Possibly conflate (and average out) damage and injury to a lower limb of a beast form into the combined single lower limb of that side. (Monopods, of however many upper limbs, would combine all previous limb damage into their now sole pedestal.)
Logically, of course, a spider-form might have a number of relationships to a biped. I would tend to go for the three rear pairs of limbs marked as equivalent to lower limbs of a biped, the foremost ones being arm-equivalent manipulators, and thus having that connection. (Mandibles could be considered distortions of mouth-parts) Another argument would make all eight legs "lower body". Although biologically-speaking, a perhaps stronger argument would be that all eight legs were 'arms' (coming from the cephalothorax "head/torso" combo segment) and no legs at all sprouting from the abdominal segment(s). Indeed, despite the foremost pair of legs/arms (chilicerae) being the primary handling choice, all limbs could be deemed to have near identical prehensile abilities when it comes to large-scale work like grabbing and wrapping prey.
So, you (generic Toady-style developer) might be tempted to look at getting something into the raws to make some mapping which need not exactly map between creatures, but could be used as a guide. As an infrequent MODder (getting stuff like War Elephants is the top end of what I'd normally poke away to obtain) I've just looked at the GCS details, but can't actually spot the "there are eight legs" bit, at first glance, so can't see easily how limb damage on that entity might relate upon change to (say) Slugman form. Or Giant Earthworm. But as there's likely to be further raw changes between now and that future point in time where its possible I'm going to hold off suggesting the exact mechanisms in coding (for guesswork) or markup (where explicit relationships can be made between bodyform components) might be used.
Does that make sense? I sense not.