Remy Maucherat wrote: > It's possible it would not cause any regression, I don't know for sure > (it's a bit the issue). > > The original claim was to fix 39704, which cannot be fixed (the bug is > really invalid, for the reasons I stated inside the bug, since when > creating the loader the privileged flag value will always be "false"; > other similar bugs may be valid, but this one is not), so I tried to > find something a bit safer for the 5.5 branch. I cleaned up the creation > of the Loader in TC 6, so that it's no longer mandatory to pass the > parent CL when creating the object (and it should fix this mess without > all the creative hacks).
How about porting the changes from TC6 but adding an option to the Loader that defaults to the existing behaviour but can be set to use the TC6 behaviour? This way we have: - no regression risk - no API change - cleaner code Assuming there are no objections, I'll do this later this week. Mark --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
