On Fri, 8 Aug 2025 18:55:07 GMT, Daniel D. Daugherty <dcu...@openjdk.org> wrote:
>> Thanks for the clarification, I think that it is fine, then, although a >> comment explaining in the code would be great. > > I suspect that the multiple includes reported in the .d files are accurate. > The .d file appears > to report the number of times we `#include` a file, however, because if the > include-guard > constructs we use, the content is pulled in only once. The number of times a header is (directly or indirectly) included by a given source file doesn't affect the number of times that header appears in the .d file associated with that source file. The compiler uniquifies that set (though links might confuse things, but that's not an issue here). The .d files consist of a set of targets, and for each target, the set of dependencies. For the <sourcefile>.d files, there's one target, <sourcefile>.o, with its dependencies (once each). For BUILD_LIBJVM.d, there's a target for each .o file. And for each of those targets, it's dependencies (once each). Essentially BUILD_LIBJVM.d includes a concatenation of all the <sourcefile>.d files. And since there are duplicates among the dependencies of different targets, there are many(!) duplicates in this file. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/26681#discussion_r2265529851
