Ondrej
The design is as follows. If you hit 2, and points are not the only type enabled, then it will save what is currently enabled, clear everything, and enable only points. If you hit 2, and points are the only type enabled, then it will restore the enabled types it saved earlier and also enable points. So 2 will toggle between only points and points plus something else.
What are you finding unintuitive? The hotkeys to isolate a type for selection mimic how regular selectors work, so is it using the same hotkey to switch back to that type plus whatever was enabled earlier?
Another option is to have 2,3,4 only isolate, and have a separate hotkey to restore your last state that had multiple component types enabled. I just thought it would be faster to hit a hotkey to temporarily limit selection to a particular type, make your selection, and then hit that same hotkey to get back to your previous selection types.
This, all this is unintuitive. I'm not sure how I can convince you of this since to you it seems intuitive. Let's make an experiment, regardless of what suggestions I'm going to make in this post, leave them (if you're considering them) for a future version, say H16. Till then let's count the number of people, especially new comers, that will allude or complain about this. Sure, this the argument from popularity, but since we're “measuring” intuitiveness, I'm not sure what other approach is in order.
Here's how a help section should read in explaining a simple straightforward workflow:
-activate a transform tool to enter edit mode
-to select points press ‘2’
-to select edges press ‘3’
-to select faces press ‘4’
-to enable the tweak mode, press 'x'. In tweak mode, to toggle Off (remove) one component ctrl click its respective icon; to toggle it On (adding), shift click itThat's it.
Here's another one proposed earlier, which doesn't have a separate tweak mode:
-activate a transform tool to enter edit mode
-to select points press ‘2’
-to select edges press ‘3’
-to select faces press ‘4’
-pressing the hotkey to the currently active mode will enable all components
-to toggle Off one component ctrl click it; to toggle it On, shift click it
-when in “all mode” or any combination of two, pressing any comp. hotkey will isolate it (and only that); pressing it again, will go to “all”, not the previous combo (this last one is important because it removes the necessity of having a degree in selections)
The 2nd workflow might seem a little more elegant (which is another way of saying, it solves the same problems by being simpler) than the 1st, but in fact it is not from a practical stand point. Sure, it's one less “mode”, but it requires a little more thought from the user.
Pausing for a second almost every time you want to switch a component is going to put a toll on you. Please model something in 0.294 - all kind of ops, switching objects back and forth, for about half an hour, then multiply that with twenty.
I'm going to go out on a limb here and say that you approached this from a purely theoretical position:
you do that which means that and then that… It's not informing you on what experience the user is going to have when using it on and on without testing it for at least 15-30min. I know devs. don't have time for prolonged tests, but I did it for you in this case.
Regarding the current workflow, I swear that right now I don't know what is a bug and what is normal, that is, a consequence of the implemented workflow.
As a quick example, after I did a few modeling ops on a few objects, deleted them, created a new one, pressed 2 to enter point, pressed T and bam! I'm in edge mode. Is this is bug or a consequence? I have a feeling it's the latter.
I'm sure I could figure it out after experimenting with it which is which, but I'm not going to because I'm never going to use Houdini for modeling if this is it.
I've continued my small rant here, but I've decided that that's… well, “lame” is probably the best way to put it along with the fact that this is not how things get done.edit: reading again your explanation, it seems that you're under the impression that in tweak mode, modelers often use combinations of two (points /w edges, polys / points, etc), but this in fact false from my own experience along with other XSI users I know of. I almost never used a combo. I'm either in point, edge or poly or, in tweak mode where I need access to all three. I invite other XSI users to confirm or deny what I'm saying here.