Oh, I can verify that case by having a test case which attempts to make an instance of the loadable class before the lib has been loaded.
If we were still referring to classes by their actual classname, then that wouldn't work, because the Flex compiler would see a reference to " new myclass()" and would link it in with the main app. But because we reference all classes through the 'lz' object, e.g., "new lz.myclass(...)", the flex compiler won't attempt to link that at compile time into the main app. On Tue, Jan 26, 2010 at 2:57 PM, P T Withington <[email protected]>wrote: > On 2010-01-25, at 20:54, Henry Minsky wrote: > > >> 2. Have you a test to ensure that non-anonymous classes that are defined > in > >> an import library and only used in the import library are actually > emitted > >> to the import library rather than the main program? > >> > > > > I checked by inspection, but I'll see if there's some way to add an > > automated test (it's > > hard to think how to do it in LZX, given that the class is going to be > > anonymous and > > have a gensym'ed name) > > No, I am asking, if I have a named class in a snippet, are we sure that > named class is emitted in the snippet? > > This bug showed up because we stopped putting anonymous classes in the `lz` > table. We were emitting the (anonymous) snippet class in the main app, and > it used to be kept by the flex compiler since it was referenced from `lz`, > now it is not, so it was dropped. Hence the bug. > > My concern is: name (NOT anonymous) snippet classes may still have the same > bug, may still be emitted into the main app, rather than the snippet. > Programs will still work, but it seems not in the spirit of the class being > in the snippet... -- Henry Minsky Software Architect [email protected]
_______________________________________________ Laszlo-reviews mailing list [email protected] http://www.openlaszlo.org/mailman/listinfo/laszlo-reviews
