Community
Participate
Working Groups
3.2 RC5 Those Java core options use values ENABLED and DISABLED. In contrast with compiler options, which in case of a DISABLED value tolerate an unchanged input and simply do not exhibit a supplementary function associated to the said option (see COMPILER_PB_SUPPRESS_WARNINGS for example), the two options toggle in fact between the two folliwing logical states: - the function is enabled; - the function is not enable *and* the other parameters associated to the said function must be removed. I believe that an ENABLE / FORBIDDEN values pair would underline the real behavior and clarify the contrast with the general 'don't use extraneous parameters when disabled' policy. Of course, this would need a gradual deprecation of DISABLED in this case. (Alternatively, the option to simply ignore secondary parameters might be considered. I assume though that the design was elaborated with a precise end in mind, and that the check on the absence of such parameters is needed by some clients of the API.)
This would be a breaking API change. Philippe, do we still want to do it or simply improve the doc to reflect Maxime's statement?
No action planned. We will leave with the current story.
Verified for 3.5M3 using I20081026-2000 build.