Closed mattys101 closed 6 years ago
This may well be true.
You have the concept correctly, that there is a table if imports on a per class bases so MyClass can have imports and these can be very different to YourClass. Also, the idea is to allow you to import a class into your namespace and alias it.
Please can you tell me more about now the current structure is not finding classes?
If we have:
"some/package/MyClass.st"
Smalltalk import: 'some.other.package.SomeClass'.
SomeClass subclass: MyClass.
The import leads to:
some.package ->
SomeClass ->
SomeClass ->
some.other.package.SomeClass
So when SomeClass
is resolved in the context of MyClass
the import cannot be found. Because it doesn't actually have any imports.
Does that help?
That does help. Thank you for such a detailed description of the problem and especially for the suggested fix. I have pushed a fix to master. Please let me know if this solves the issue. I did some testing but not as involved as you have.
Looks like that fixed it :-) Thanks.
Yay. Again thank you for such a detailed problem description and suggested fix.
Sent from my Commodore 64
On 24 Nov 2017, at 12:08 pm, Matt notifications@github.com wrote:
Looks like that fixed it :-) Thanks.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.
The import code in
RedlineSmalltalk
appears to use the wrong classname. If I am reading it correctly the import structure is as follows:To properly support this I think the
addImport
method needs to have a 4 arg signature like:rather than the current:
Which leads to
and hence the resolution of the import cannot find the imported class.