On Fri, 14 Apr 2023 13:28:40 GMT, Alan Bateman <al...@openjdk.org> wrote:
>> Can I please get a review of this change which removes the `java.compiler` >> system property? This addresses https://bugs.openjdk.org/browse/JDK-8041676. >> >> A CSR has been filed for this change and is available at >> https://bugs.openjdk.org/browse/JDK-8305998. The CSR has more details about >> this proposed change, but to summarize, the `java.compiler` system property >> wasn't used in any practical way. With the recent removal of >> `java.lang.Compiler` interface from the JDK, the presence of this system >> property wasn't of any practical importance. >> >> The commit in this PR, removes its reference from the >> `System.getProperties()` documentation and also removes the specific >> implementation from hotspot where it treated the values `NONE` and empty >> string in the absence of `-Xdebug` in a specific manner and considered it to >> be an instruction to run the application in interpreter-only mode. `-Xint` >> option has been around for this purpose, so removal of this `java.compiler` >> system property support from hotspot implementation wouldn't remove any >> usable feature for the applications. >> >> Additionally, the hotspot implementation now logs a warning message when it >> detects the presence of `java.compiler` system property when `java` is >> launched. This warning message will be removed after a few releases. >> >> The current JDK source has reference to the `java.compiler` system property >> in numerous NetBeans project build files. The usage of this system property >> in such files isn't necessary. However, this PR doesn't intend to cleanup >> those references, since it isn't clear which of those NetBeans projects are >> still relevant. I think we can use a separate PR to do that cleanup. >> >> tier1, tier2, tier3 testing has been done with these changes and those tests >> have passed. > > src/hotspot/share/runtime/arguments.cpp line 1313: > >> 1311: " use -Xint if you want to run the application in >> interpreted-only mode."); >> 1312: } else { >> 1313: warning("The java.compiler system property is obsolete and no >> longer supported."); > > This looks okay although the warning for the empty value and "NONE" case will > probably wrap as it's very long. Today, -Djava.compiler=foo is ignored, no > warning, seems good to emit a warning about that too. Hello Alan, > This looks okay although the warning for the empty value and "NONE" case will > probably wrap as it's very long. Agreed, it's a bit long. I can't think of a simpler message though. If anyone has a suggestion, I'll update accordingly. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/13475#discussion_r1166902101
