Formalize support for Dynamic Class Referencing
"Dynamic Class Referencing" has been discussed for years as a necessary design pattern. the problem is this: when writing framework rules, and referring to classes in calls for Page-New, Obj-Open, Obj-Browse, etc -- at runtime the generated rule needs to know which classname to substitute. Originally some rules would refer to pxRequestor.pxCurrentWorkPool, and there was a suggestion that one should simply concatenate the subclass name to it.
[I have brought this up in a discussion on Mesh which I no longer have access to, so I am re-posting here.
But it's not official. Most people even refer to it by different names. It may or may not go by that name on the certification quizzes.
There's been a common sense that creating a Declare Page is the best way to do this -- see the Order Management Framework, using the page Declare_AppExtension, for example.
But this is still hacky, since there's no validation tracing this back to the framework class.
Ideally we should have implement a map rule for each ruleset which lists the class nickname and implementing class.
See also a PDN question about Class Nicknames.
"Dynamic Class Referencing" has been discussed for years as a necessary design pattern. the problem is this: when writing framework rules, and referring to classes in calls for Page-New, Obj-Open, Obj-Browse, etc -- at runtime the generated rule needs to know which classname to substitute. Originally some rules would refer to pxRequestor.pxCurrentWorkPool, and there was a suggestion that one should simply concatenate the subclass name to it.
[I have brought this up in a discussion on Mesh which I no longer have access to, so I am re-posting here.
But it's not official. Most people even refer to it by different names. It may or may not go by that name on the certification quizzes.
There's been a common sense that creating a Declare Page is the best way to do this -- see the Order Management Framework, using the page Declare_AppExtension, for example.
But this is still hacky, since there's no validation tracing this back to the framework class.
Ideally we should have implement a map rule for each ruleset which lists the class nickname and implementing class.
See also a PDN question about Class Nicknames.
Additionally, there are rules where you simply most supply a static classname.
This is now FDBK-9978, for those of you with internal access.