On Thu, 7 Aug 2025 17:32:35 GMT, Aleksey Shipilev <sh...@openjdk.org> wrote:
> Also, maybe check in the generation script as well? I think a subfolder here > would be fine: https://github.com/openjdk/jdk/tree/master/src/utils. It would > be nice if the output of that script could be simply piped into > `precompiled.hpp`, so we can use it later without extra work. > > We can, technically, hook it up a similar way SortIncludes.java is currently > done, but I think it is unnecessary at this point. In a perfect world we > would not be needing precompiled headers. In less ideal world, having a jtreg > test that warns us that a new popular header appeared, or that older header > is not as popular anymore, would be handy. Again, this is something out of > scope for this PR. I'll add the generation script, it looks like something that could be useful at some point in the future. I think there should be a human curator though, because the process is not completely automatic. For example, when the threshold is set to 50 I get this compilation error: /jdk/src/hotspot/share/compiler/compilerEvent.cpp:59:36: error: redefinition of 'phase_names' with a different type: 'GrowableArray<const char *> *' vs 'const char *[100]' static GrowableArray<const char*>* phase_names = nullptr; ^ /jdk/src/hotspot/share/opto/phasetype.hpp:147:20: note: previous definition is here static const char* phase_names[] = { It could happen with any threshold, even 130 as the codebase evolves. ------------- PR Comment: https://git.openjdk.org/jdk/pull/26681#issuecomment-3165535039
