My experience is that class generator is actually much more usable in
B3/RC1 than it was in B2. So I guess some usage scenarios work better
than others (e.g. I used it extensively on an ROP project, meaning
standard server and client templates, but rarely with custom
templates). BTW, before retesting any of these could you please reset
the preferences by deleting ~/.cayenne/prefs directory, just in case.
* previous code generation behaviour was:
a) select a class->generate code-> result: only one class preselected
in the code generation dialog
b) select a map->generate code-> result: all classes are preselected
in the code generation dialog
I think I recognize this issue - will work on it.
* the modeler can't seem to remember the output directory for code
generation, that is, it always appends "cayenne" to an otherwise valid
(remembered) directory path
I am not seeing that. Could you open a bug report with some screenshots
* I had a lot of problems working with custom code generation
templates: once modified, they generated completely invalid
(super)classes, sometimes not parsed at all, other times with weird
import statements
- the solution? used beta 2 with the same templates, worked
like a charm
Also works for me - please log the details in Jira
* the template properties dialog doesn't allow modification of
existing templates
Can you elaborate?
* the template properties dialog doesn't remember the last directory
from which the user loaded a template so the user has to navigate the
whole path for every template version he uses if he defined a couple
of templates in advance (as I did, for example)
Will fix this.
* the modeler would be *sooo* much more understandable alltogether if
it had a tooltip (a sentence or two explaining what the user's looking
at) attached to most components, so that users wouldn't have to wonder
what e.g. "generation mode: Once per every Entity" means or have to
look it up on the Internet.
A couple of other examples where tooltips would help are "allow client
entities", "[Entity object] Qualifier", "Import EOModel". I could name
about several dozens more.
It's been asked by other users recently as well. Won't happen in 1.2,
but we may do this in the future releases.
To wash away the aftertaste left behind by such a list of rought
edges, I can only say I look forward to developing cayenne-powered
apps in the future as well as returning to work on my current
cayenne-powered app, now that beta 2 came through where RC1 failed. :)
And you can also recall that 1.1 didn't have most of this
functionality at all. It is new, and we are still testing it, so
thanks for the feedback - we will certainly fix all these wrinkles.
Andrus
