another ProE annoyance indeed ... and it is still implemented 8-(
The long way around this is: first delete the original pattern, replace all components using layout, then recreate patterns and pick all referring components to pattern them again.
I know some people like pattern tables, I try to respect that. But I hate them. Here's why a.. You know how, with instance patterns, you can pick an instance in the model tree and the part lights up? Not so with table patterns; these are as dumb as the Half-Amateur/TABLE they are built on (don't get me started on this decrepit, bullshit functionality, we'd be all day). b.. No more can you change a pattern increment value and have the instances update automatically; you have to go into each one and add values by hand (sit there with a calculator and add 2x, 3x, 4x, etc.) c.. You have no idea where the zero/zero point of the pattern is, so you start off, with any mods to the pattern, with some detective work (not fun, not productive), just to establish the baseline from which you are modifying (more stupid shit) d.. And Pro/TABLE is just crap, useless garbage, junk to work with; a headache, a grotesque boil; the "Elephant Man" of Pro/e
Since it is possible to redefine the pattern options of the "master" feature to table driven one can save complex patterns :-) and it is good practice to have the related components arranged into a single subassembly (e. g. bolt, nut & washer) to ease pain.
I greatly agree with creating groups of components and patterning the group; group fails, all components fail; group succeeds, all components succeed~not doing the same thing once for the washer, again for the lock, again for the screw.
Why the replacement ob subassemblies can´t work on the whole pattern is beyond me ... whenever I know beforehand that a "replace" task will become necessary I make both feature (e. g. hole) and subassembly (e. g. bolt & nut) after a table driven "master" (e. g. axis, point), and make multiple use of the pattern table.
Here's what I thought of doing, haven't tried it yet. Hide the old component, assemble new without ref to old; take pattern (original comp?) and reroute to new comp refs; delete old comp. Should be clean, maybe a little more work than Replace but, in the long run, less trouble.
Some stuff learned along the way in this somewhat useful excercise in humiliation: a.. Comps in patterns and groups appear, in the model tree status column, to be regenerated, but the features/components WITHIN may all be frozen b.. Beside Modelcheck>Interactive, the most useful tool for checking for "Frozen" and other errant conditions, is the Binoculars (^F), Search for "component", Status, "Frozen" (or "Suppressed", etc), highlight, move to active list, Tree expands/highlights to reveal culprits; RMB actions possible on same c.. No ref pattern works correctly unless you reference the FIRST instance of the original, base pattern d.. You can, however, redefine a second, third, etc. instance independently by picking it and doing 'Ed Def'; often, rest of patterned comps regen correctly e.. Groups stand in the way of redefining member placement, so 'Ungroup' is necessary f.. Start at top of tree; often, when you fix early dependencies, the later dependencies get fixed successively, consequentially, automatically g.. The pattern friendliest, most bullet proof way of creating new parts that Replace smoothly is building them as instances of a family of parts David Janes