Em ter., 21 de mai. de 2024 às 09:25, Peter Eisentraut <pe...@eisentraut.org> escreveu:
> On 20.05.24 15:59, Tom Lane wrote: > > Peter Eisentraut <pe...@eisentraut.org> writes: > >> This patch converts the compile-time settings > >> COPY_PARSE_PLAN_TREES > >> WRITE_READ_PARSE_PLAN_TREES > >> RAW_EXPRESSION_COVERAGE_TEST > > > >> into run-time parameters > > > >> debug_copy_parse_plan_trees > >> debug_write_read_parse_plan_trees > >> debug_raw_expression_coverage_test > > > > I'm kind of down on this. It seems like forcing a bunch of > > useless-in-production debug support into the standard build. > > What of this would be of any use to any non-developer? > > We have a bunch of other debug_* settings that are available in > production builds, such as > > debug_print_parse > debug_print_rewritten > debug_print_plan > debug_pretty_print > debug_discard_caches > debug_io_direct > debug_parallel_query > debug_logical_replication_streaming > If some of this is useful for non-developer users, it shouldn't be called debug, or in this category. > Maybe we could hide all of them behind some #ifdef DEBUG_OPTIONS, but in > any case, I don't think the ones being proposed here are substantially > different from those existing ones that they would require a separate > treatment. > > My goal is to make these facilities easier to use, avoiding hand-editing > pg_config_manual.h and having to recompile. > Although there are some developer users. I believe that anything that is not useful for common users and is not used for production should not be compiled at runtime. best regards, Ranier Vilela
