Meeting minutes
Prague [Regrets for next week]
Steven: I will be at Prague, XForms tutorial, and ixml talk
Alain: I will be remote
ACTION Steven MIPs on controls
https://
https://
Steven: Are MIPs allowed on containers as well as core controls
Erik: I think in general yes.
[Alain nods]
Erik: <group ref="..." readonly="true">
… inheritance might work differently
Steven: So I'll add it, and we can see what ramifications it has
ACTION: Steven spec text for 'initial' MIP
https://
Erik: I agree that it is like @calculate.
… It has to happen very early.
Steven: initial and calculate bite each other.
… I assume initial happens before the first update
Erik: We do it in our implementation just before the first calculate
… right after whitespace processing.
Steven: So initial has an ordering dependency like recalculate
Erik: Initial values are done only once; the code is in the update, with evaluation of initial only done on the first update.
Steven: So you treat it like a calculate, but you do it only once.
Steven: And if you have an initial and a calculate, you do the initial before the calculate?
Erik: Yes.
Steven: Does it work if you have no <instance/> but <binds/>?
… If not, then initial will never be available. If so then we have to work out if initial happens
Erik: We don't say that binds are disallowed...
Steven: Does that mean that initial happens on the next update?
Erik: the spec may be a bit underspecified on this point.
Steven: It sounds like we need an extra update after UI initialisation
Erik: We don't implement this bit actually
Alain: I do, but I haven't tested it with binds
Steven: Why don't we just drop the model construction from the UI?
Erik: I would support deprecating or deleting
Alain: I agree
RESOLUTION: Remove the automatic creation of instances during UI construct from XForms
Steven: Last question about initial on <insert/>
Erik: It can be useful.
ACTION: Steven to remove auto generation of instances from spec
AOB
[None]
[Adjourn]