Community
Participate
Working Groups
FileReader.hardClose uses Closeable and Flushable interfaces, both are 1.5 ProgressStatistics has references to Long.valueOf(long) Double.valueOf(double) String.format all of which are 1.5 I have temporarily changed the manifest to specify a Bundle-RequiredExecutionEnvironment=1.5 to avoid compilation errors in the integration build.
API tools should catch these problems.
Created attachment 130420 [details] Fix that does not make use of Closeable and Flushable This fix checks if the stream is an Input or Output stream and just calls close. The standard buffered streams will flush on a close. Callers who need more control over stream should not ask FileReader to close the stream for them anyway. The dependency to 1.5 should be fixed by this.
Duh, I missed half of the problem... And also forgot to ask - should the impl be changed to not use any 1.5 API?
Should we be letting people know about this for the IBuild? (that is, are there people who run on a 1.4 VM that will hit strange errors this week)?
We should not use 1.5 API anywhere. We need to change the EE in the manifest to 1.4 and fix all references outside of this set. As for warning the community, I don't think we need to say anything and worry anyone as the 1.5 EE will prevent the plug-ins to resolve and p2 will be disabled.
Created attachment 130441 [details] Fix for all the reported 1.5 classes/methods New patch that fixes all the reported classes/methods. It now uses NumberFormat.
*** Bug 270662 has been marked as a duplicate of this bug. ***
*** Bug 270759 has been marked as a duplicate of this bug. ***
> (that is, are >there people who run on a 1.4 VM that will hit strange errors this week)? They can neither launch Eclipse nor run test using a 1.4 VM.
Released patch into HEAD.
*** Bug 270794 has been marked as a duplicate of this bug. ***
