-
Notifications
You must be signed in to change notification settings - Fork 10.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Require Java 11+ to *build* Guava but continue to support Java 8 at runtime #6549
Comments
Well, not completely: We could use |
It may be reasonable to use I would be inclined to instead make jdk11 the minimum version. Those who are performance sensitive should not be on jdk8 and those users are likely to only need critical bug fixes to be backported. Android annoyingly only supports a subset of jdk11, but likely enough for your needs. You'll certainly have people complain but external are suffering regardless and internal are posing a security risk, so you might not have as much actionable push back if pursued. I recall some tools were not multi-release jar friendly (e.g. if naively creating a fatjar) and since they are not widely used, I'd consider that to likely be a higher risk of unexpectedly breaking someone in a surprising way than adopting jdk11. |
Thanks, I'm not sure whether I ever knew that about the annotations. I'll have to dig up some past threads and leave a link from them to here. I'll also read your link about the field updaters. That's good news for someday getting off Unsafe. As for requiring a new Java version, that is above my pay grade :) Thankfully, even our final straggler internally was moved off Java 8 a while back, but we'll likely need to maintain compatibility as long as enough open-source Google and non-Google libraries do. Sadly, if we try to make only bug fixes to a Java 8 backport, I'd expect the same kind of version conflicts that users are used to from the days when we often removed APIs: If one of your deps updates to Guava 32.4.0, then it might use APIs that aren't available to you. It's good to know that we should look into the fat-jar situation if we try mrjars. I think we had some trouble with that kind of thing internally, if I'm remembering right. Hopefully it's been sorted out both internally and externally, but it's worth remembering that there are risks to everything (just in case the recent CVE-Windows experience didn't remind me enough :)). And at the moment, the benefits of using Java 9+ APIs are looking small (especially after what you said about the annotations), so we will have less motivation to go with a multi-release jar (though it's still a possibility). -- One other note that I should have put in my original post: Also related to toolchains: #5457 That's where I'd previously mentioned google/error-prone#3895 (comment), which is the key to testing Java 8 compatibility even when building with a newer version. |
My guess at the full set of related commits in Error Prone is: |
This is the first use of a Java 9 API in Guava, but we use the API only when it's available, so we maintain compatibility with Java 8. Use of Java 9 APIs is relevant to #6549 and #3990 (and also mojohaus/animal-sniffer#67). I didn't make the same change for `guava-android`, which [will add `java.util.zip.CRC32C` in API Level 34](https://developer.android.com/reference/java/util/zip/CRC32C). I don't know if Android is providing similar performance improvements, so it might not even matter. But even if I wanted to do it, I can't with my current approach, which relies on `MethodHandle`—unless I want to make even the usage of `MethodHandle` conditional on a reflective check :) RELNOTES=`hash`: Enhanced `crc32c()` to use Java's hardware-accelerated implementation where available. PiperOrigin-RevId: 539722059
This is the first use of a Java 9 API in Guava, but we use the API only when it's available, so we maintain compatibility with Java 8. Use of Java 9 APIs is relevant to #6549 and #3990 (and also mojohaus/animal-sniffer#67). I didn't make the same change for `guava-android`, which [will add `java.util.zip.CRC32C` in API Level 34](https://developer.android.com/reference/java/util/zip/CRC32C). I don't know if Android is providing similar performance improvements, so it might not even matter. But even if I wanted to do it, I can't with my current approach, which relies on `MethodHandle`—unless I want to make even the usage of `MethodHandle` conditional on a reflective check :) RELNOTES=`hash`: Enhanced `crc32c()` to use Java's hardware-accelerated implementation where available. PiperOrigin-RevId: 539722059
This is the first use of a Java 9 API in Guava, but we use the API only when it's available, so we maintain compatibility with Java 8. Use of Java 9 APIs is relevant to #6549 and #3990 (and also mojohaus/animal-sniffer#67). I didn't make the same change for `guava-android`, which [will add `java.util.zip.CRC32C` in API Level 34](https://developer.android.com/reference/java/util/zip/CRC32C). I don't know if Android is providing similar performance improvements, so it might not even matter. But even if I wanted to do it, I can't with my current approach, which relies on `MethodHandle`—unless I want to make even the usage of `MethodHandle` conditional on a reflective check :) RELNOTES=`hash`: Enhanced `crc32c()` to use Java's hardware-accelerated implementation where available. PiperOrigin-RevId: 539722059
This is the first use of a Java 9 API in Guava, but we use the API only when it's available, so we maintain compatibility with Java 8. Use of Java 9 APIs is relevant to #6549 and #3990 (and also mojohaus/animal-sniffer#67). I didn't make the same change for `guava-android`, which [will add `java.util.zip.CRC32C` in API Level 34](https://developer.android.com/reference/java/util/zip/CRC32C). I don't know if Android is providing similar performance improvements, so it might not even matter. But even if I wanted to do it, I can't with my current approach, which relies on `MethodHandle`—unless I want to make even the usage of `MethodHandle` conditional on a reflective check :) RELNOTES=`hash`: Enhanced `crc32c()` to use Java's hardware-accelerated implementation where available. PiperOrigin-RevId: 539983316
We might also eventually see new versions of Maven plugins drop support for Java 8, as noted in uber/NullAway#778. In fact, even today, we have special build setup for Error Prone, which requires Java 11+ [edit: for now]. |
If any of our annotation dependencies start being built for Java 9+, then we would start to need to build with a newer javac in all cases. I speculate that the annotations targeting Java 9+ might still be OK to references from a Guava that targets Java 8, but we'd want to verify that—especially since Java 8 has a bug in its handling of type-use annotations that it can't resolve. |
An interesting question from an internal discussion:
I would like to think so, but that sounds like a good thing for us to check if we ever look seriously into publishing a multi-release jar. |
(As I just noted in #3990 (comment), even a build-time requirement could be inconvenient for users if they build Guava themselves: In this case, it could be inconvenient if they have continued to build with an older JDK version. So, for example, if an organization both builds with JDK 8 and runs with JDK 8, then they'd no longer be able to build Guava if its build required JDK 11, even if newer compilers could produce a version of Guava that runs on JDK 8.) |
I am in the process of trying to conditionally use a Java 9+ API again as part of #6634 (comment) :\ |
On the toolchain front: I heard a while back that JDiff breaks at JDK ~13, so we wouldn't necessarily want to build with newer versions, at least for our Javadoc generation. Or we could finally migrate to japicmp, which seems to be the replacement that everyone recommends. [edit: But I didn't see any obvious problems when #7087 changed Javadoc+JDiff to use Java 21. I suppose it's possible that JDiff failed silently. Or maybe what I heard is that JDiff breaks on JDK 13 bytecode or something?] [edit: Or did it fail after all?] |
This would also be convenient (but not strictly necessary) if we want to start generating a |
… to 1.0.2 to prepare for a release (which we'll then need to update Guava to use). I chose an `Automatic-Module-Name` over an actual `module-info`, even for this dependency-free artifact, because [I can do that without requiring JDK 9+ for builds](#6549 (comment)). Granted, there would be relatively little harm in requiring JDK 9+ for `failureaccess` builds, since `failureaccess` isn't part of our normal build process. (Guava's build pulls an already released version of `failureaccess`.) Still, it's possible that someone is building both Guava _and_ `failureaccess` with JDK 8, so it may be nice not to break that workflow. Plus, I'm not sure that a proper module definition buys us much (relative to `Automatic-Module-Name`) when we have no deps? Still, I am a bit tempted, if only to try to shake out remaining issues that `module-info` might cause our users. This CL is progress toward fixing #6776 (or "toward working around a Maven bug," if you prefer). It's also a tiny bit of progress toward modularizing Guava (#2970), since `failureaccess` is one of its existing unmodularized dependencies. RELNOTES=Added an `Automatic-Module-Name` to `failureaccess`, [Guava's one strong runtime dependency](https://github.com/google/guava/wiki/UseGuavaInYourBuild#what-about-guavas-own-dependencies). PiperOrigin-RevId: 573236700
… to 1.0.2 to prepare for a release (which we'll then need to update Guava to use). I chose an `Automatic-Module-Name` over an actual `module-info`, even for this dependency-free artifact, because [I can do that without requiring JDK 9+ for builds](#6549 (comment)). Granted, there would be relatively little harm in requiring JDK 9+ for `failureaccess` builds, since `failureaccess` isn't part of our normal build process. (Guava's build pulls an already released version of `failureaccess`.) Still, it's possible that someone is building both Guava _and_ `failureaccess` with JDK 8, so it may be nice not to break that workflow. Plus, I'm not sure that a proper module definition buys us much (relative to `Automatic-Module-Name`) when we have no deps? Still, I am a bit tempted, if only to try to shake out remaining issues that `module-info` might cause our users. This CL is progress toward fixing #6776 (or "toward working around a Maven bug," if you prefer). It's also a tiny bit of progress toward modularizing Guava (#2970), since `failureaccess` is one of its existing unmodularized dependencies. RELNOTES=Added an `Automatic-Module-Name` to `failureaccess`, [Guava's one strong runtime dependency](https://github.com/google/guava/wiki/UseGuavaInYourBuild#what-about-guavas-own-dependencies). PiperOrigin-RevId: 573930127
I just noticed that we could simplify our build configuration slightly if we could require 17+: Lines 339 to 345 in 280b5d2
That alone is clearly not a reason to do so, but it's a reminder that sometimes upgrades solve problems. And while upgrading to JDK 13 or so would break JDiff, it would also let us use the nicer format for code snippets in our Javadoc. I think that, in particular, we could start lines with " |
More toolchain thoughts:
|
We're hitting a javac8 issue again in #7080:
(https://bugs.openjdk.org/browse/JDK-8144101) Also, I think I saw that new versions of Maven will soon require Java 17. That may require us to finally do the toolchain setup if we want to test under Java 8. |
|
Builds with Java 11 are fine, but if we go to even newer versions (under the theory that we should build with the newest version available), it appears that we'll have problems with Javadoc. Maybe we can continue generating Javadoc snapshots with java 11, even as we perform the main build with a newer version (especially since we target Java 8 and restrict ourselves to Java 8 APIs)? (Or maybe we can, you know, make it work with new versions :) But I fear that the problem is JDiff, so it might not be easy to fix.) We might see problems with that approach if we try to use Java 12+ APIs internally, but I don't know that we'll do that anytime soon. And even if we do, we should be providing Java-8-compatible sources alongside the Java 12+ sources, so I'd hope that we can "just" point Javadoc to those sources. |
Oh, good point, thanks. I occasionally hear of a new such tool, and I dump it into a list I have in an internal issue about JDiff. It looks like the tool that I've seen the most energy around is japicmp (used by AdoptJDK and recommended by Kotlin). |
We hit one interesting quirk when building with JDK 21+ and running tests under JDK 8 this week: jspecify/jspecify#517. That happened specifically with JUnit 5, but it could happen in other code that performs reflection. I haven't tested Guava with the 21-8 mix to see what happens here, so I'm just mentioning the problem now as something that we'd want to remember in case we encounter it when we try toolchains for Guava or our other projects. (We still won't be able to release under JDK 21+ or even 13+, thanks to JDiff.) |
Another note on toolchains: https://openjdk.org/jeps/471 is coming, so eventually we'll find it difficult to compile against the newest JDK and continue to refer to |
In particular: - Use JDK 22 for compilation to [avoid a JDK 11 bug](#7331). - Another way to avoid that bug would be to use JDK 8, which [would also provide a `--release`-like compatibility guarantee](#3990). However, that could complicate [using newer APIs conditionally](#6549). And of course we'd expect JDK 8 to be buggier than JDK 22. (In fact, we still have a workaround or two for JDK 8 bugs (with a brand new one coming in cl/655556207), and we could now remove those—assuming that none of our users use JDK 8 to build Guava outside of our Maven build.) JDK 22 also supports new versions of Error Prone, while JDK 8 does not. - This change also allows us to simplify our Error Prone configuration, which until now needed different profiles in order to support both JDK 8 and JDK 9+. We could now upgrade Error Prone, but I haven't done so yet. - Continue to use JDK 11 for Javadoc, as [we're doing now](https://github.com/google/guava/blob/5041fbe61965a73ea269c7c24ea746d89bd1b1ba/.github/workflows/ci.yml#L89-L99) because of [problems with at least JDK 21](#7109). - What matters might actually be the version used [by _JDiff_](#6549 (comment)), which comes from the version in the linked `ci.yml` file. But since we're using JDK 11 currently for docs in general, I'm sticking with that for now. Still, we should consider [upgrading the version used for Javadoc generation](#6790 (comment)). But this CL is already complicated enough.... - When we hard-code JDK 11, we need to remove the `<source>${java.specification.version}</source>` line: That would otherwise set (for example) `-source 17` when running Maven under JDK 17, and JDK 11 wouldn't recognize it. As I recall, the `java.specification.version` usage was from the days in which we tried to inherit Javadoc from the JDK. Inheritance had [stopped working](#6790), and we ripped it out in cl/614693592. (See also [these notes](#5457 (comment)).) Fixes #7331 RELNOTES=n/a PiperOrigin-RevId: 655592201
In particular: - Use JDK 22 for compilation to [avoid a JDK 11 bug](#7331). - Another way to avoid that bug would be to use JDK 8, which [would also provide a `--release`-like compatibility guarantee](#3990). However, that could complicate [using newer APIs conditionally](#6549). And of course we'd expect JDK 8 to be buggier than JDK 22. (In fact, we still have a workaround or two for JDK 8 bugs (with a brand new one coming in cl/655556207), and we could now remove those—assuming that none of our users use JDK 8 to build Guava outside of our Maven build.) JDK 22 also supports new versions of Error Prone, while JDK 8 does not. - This change also allows us to simplify our Error Prone configuration, which until now needed different profiles in order to support both JDK 8 and JDK 9+. We could now upgrade Error Prone, but I haven't done so yet. - Continue to use JDK 11 for Javadoc, as [we're doing now](https://github.com/google/guava/blob/5041fbe61965a73ea269c7c24ea746d89bd1b1ba/.github/workflows/ci.yml#L89-L99) because of [problems with at least JDK 21](#7109). - What matters might actually be the version used [by _JDiff_](#6549 (comment)), which comes from the version in the linked `ci.yml` file. But since we're using JDK 11 currently for docs in general, I'm sticking with that for now. Still, we should consider [upgrading the version used for Javadoc generation](#6790 (comment)). But this CL is already complicated enough.... - When we hard-code JDK 11, we need to remove the `<source>${java.specification.version}</source>` line: That would otherwise set (for example) `-source 17` when running Maven under JDK 17, and JDK 11 wouldn't recognize it. As I recall, the `java.specification.version` usage was from the days in which we tried to inherit Javadoc from the JDK. Inheritance had [stopped working](#6790), and we ripped it out in cl/614693592. I assume that we'll now get the default from the JDK whose Javadoc binary we're using, which (again) will now be 11. That seems fine. We could consider setting it to 8 to match our normal build (which I thought I had remembered was the `maven-javadoc-plugin` default, but I don't think I'm seeing that, at least not under our current versions), but I don't see much downside to 11—or even to newer versions that we might someday use for Maven Javadoc generation, given that we keep the code compiling under new versions already. Some other thing I'm wondering: - I wonder if we should activate(?) some of the plugins, including the new toolchain plugins, in the `<plugins>` (not just `<pluginManagement>`) section of the parent `pom.xml`. Might that save us from having to do so in each separate `pom.xml`? (We might actually mostly get away with activating(?) them only in the main `guava` build: That _downloads and registers_ the toolchains, and then at least the other projects' _per-plugin_ toolchain configuration probably finds them? But for the more general configuration to work, I think we at least need to activate(?) `maven-toolchains-plugin` in each? I haven't experimented a ton with this.) (See also [these notes](#5457 (comment)).) So Fixes #7331 RELNOTES=n/a PiperOrigin-RevId: 655592201
In particular: - Use JDK 22 for compilation to [avoid a JDK 11 bug](#7331). - Another way to avoid that bug would be to use JDK 8, which [would also provide a `--release`-like compatibility guarantee](#3990). However, that could complicate [using newer APIs conditionally](#6549). And of course we'd expect JDK 8 to be buggier than JDK 22. (In fact, we still have a workaround or two for JDK 8 bugs (with a brand new one coming in cl/655556207), and we could now remove those—assuming that none of our users use JDK 8 to build Guava outside of our Maven build.) JDK 22 also supports new versions of Error Prone, while JDK 8 does not. - This change also allows us to simplify our Error Prone configuration, which until now needed different profiles in order to support both JDK 8 and JDK 9+. We could now upgrade Error Prone, but I haven't done so yet. - Continue to use JDK 11 for Javadoc, as [we're doing now](https://github.com/google/guava/blob/5041fbe61965a73ea269c7c24ea746d89bd1b1ba/.github/workflows/ci.yml#L89-L99) because of [problems with at least JDK 21](#7109). - What matters might actually be the version used [by _JDiff_](#6549 (comment)), which comes from the version in the linked `ci.yml` file. But since we're using JDK 11 currently for docs in general, I'm sticking with that for now. Still, we should consider [upgrading the version used for Javadoc generation](#6790 (comment)). But this CL is already complicated enough.... - When we hard-code JDK 11, we need to remove the `<source>${java.specification.version}</source>` line: That would otherwise set (for example) `-source 17` when running Maven under JDK 17, and JDK 11 wouldn't recognize it. As I recall, the `java.specification.version` usage was from the days in which we tried to inherit Javadoc from the JDK. Inheritance had [stopped working](#6790), and we ripped it out in cl/614693592. I assume that we'll now get the default from the JDK whose Javadoc binary we're using, which (again) will now be 11. That seems fine. We could consider setting it to 8 to match our normal build (which I thought I had remembered was the `maven-javadoc-plugin` default, but I don't think I'm seeing that, at least not under our current versions), but I don't see much downside to 11—or even to newer versions that we might someday use for Maven Javadoc generation, given that we keep the code compiling under new versions already. Some other thing I'm wondering: - I wonder if we should activate(?) some of the plugins, including the new toolchain plugins, in the `<plugins>` (not just `<pluginManagement>`) section of the parent `pom.xml`. Might that save us from having to do so in each separate `pom.xml`? (We might actually mostly get away with activating(?) them only in the main `guava` build: That _downloads and registers_ the toolchains, and then at least the other projects' _per-plugin_ toolchain configuration probably finds them? But for the more general configuration to work, I think we at least need to activate(?) `maven-toolchains-plugin` in each? I haven't experimented a ton with this.) (See also [these notes](#5457 (comment)).) So Fixes #7331 RELNOTES=n/a PiperOrigin-RevId: 655592201
In particular: - Use JDK 22 for compilation to [avoid a JDK 11 bug](#7331). - Another way to avoid that bug would be to use JDK 8, which [would also provide a `--release`-like compatibility guarantee](#3990). However, that could complicate [using newer APIs conditionally](#6549). And of course we'd expect JDK 8 to be buggier than JDK 22. (In fact, we still have a workaround or two for JDK 8 bugs (with a brand new one coming in cl/655556207), and we could now remove those—assuming that none of our users use JDK 8 to build Guava outside of our Maven build.) JDK 22 also supports new versions of Error Prone, while JDK 8 does not. - This change also allows us to simplify our Error Prone configuration, which until now needed different profiles in order to support both JDK 8 and JDK 9+. We could now upgrade Error Prone, but I haven't done so yet. - Continue to use JDK 11 for Javadoc, as [we're doing now](https://github.com/google/guava/blob/5041fbe61965a73ea269c7c24ea746d89bd1b1ba/.github/workflows/ci.yml#L89-L99) because of [problems with at least JDK 21](#7109). - What matters might actually be the version used [by _JDiff_](#6549 (comment)), which comes from the version in the linked `ci.yml` file. But since we're using JDK 11 currently for docs in general, I'm sticking with that for now. Still, we should consider [upgrading the version used for Javadoc generation](#6790 (comment)). But this CL is already complicated enough.... - When we hard-code JDK 11, we need to change the `<source>${java.specification.version}</source>` line: That would otherwise set (for example) `-source 17` when running Maven under JDK 17, and JDK 11 wouldn't recognize it. As I recall, the `java.specification.version` usage was from the days in which we tried to inherit Javadoc from the JDK. Inheritance had [stopped working](#6790), and we ripped it out in cl/614693592. I first tried going with the default from the JDK whose Javadoc binary we're using, which (again) will now be 11. But that led to a problem in `org.codehaus.plexus.languages.java.jpms.CmdModuleNameExtractor`, which apparently tries to look up the module name for the `-source 11` run but uses the Maven run's JDK instead of the Javadoc toolchain or Maven toolchain. So now I've set it to 8 to match what we use for `maven-compiler-plugin`. (I _thought_ I had remembered that `maven-javadoc-plugin` defaulted to matching `maven-compiler-plugin`, even though that's weird. Maybe the two actually just read from the same Maven property or something, or maybe the behavior changed.) Some other thing I'm wondering: - I wonder if we should activate(?) some of the plugins, including the new toolchain plugins, in the `<plugins>` (not just `<pluginManagement>`) section of the parent `pom.xml`. Might that save us from having to do so in each separate `pom.xml`? (We might actually mostly get away with activating(?) them only in the main `guava` build: That _downloads and registers_ the toolchains, and then at least the other projects' _per-plugin_ toolchain configuration probably finds them? But for the more general configuration to work, I think we at least need to activate(?) `maven-toolchains-plugin` in each? I haven't experimented a ton with this.) - I forgot the other thing while I was typing :( (See also [these notes](#5457 (comment)).) So Fixes #7331 RELNOTES=n/a PiperOrigin-RevId: 655592201
In particular: - Use JDK 22 for compilation (also, for any other [affected plugins](https://maven.apache.org/guides/mini/guide-using-toolchains.html#prerequisites)) to [avoid a JDK 11 bug](#7331). - Another way to avoid that bug would be to use JDK 8, which [would also provide a `--release`-like compatibility guarantee](#3990). However, that could complicate [using newer APIs conditionally](#6549). And of course we'd expect JDK 8 to be buggier than JDK 22. (In fact, we still have a workaround or two for JDK 8 bugs (with a brand new one coming in cl/655556207), and we could now remove those—assuming that none of our users use JDK 8 to build Guava outside of our Maven build.) JDK 22 also supports new versions of Error Prone, while JDK 8 does not. - This change also allows us to simplify our Error Prone configuration, which until now needed different profiles in order to support both JDK 8 and JDK 9+. We could now upgrade Error Prone, but I haven't done so yet. There are probably other simplifications that we could perform, as well, such as `maven-javadoc-plugin.additionalJOptions`. - Originally, I'd set up this CL to explicitly set only the toolchain of `maven-compiler-plugin` to 22. I had it using 11 for any other plugins (just Animal Sniffer, maybe?), I think from when I was trying to get toolchains to take effect at all. I've since changed this CL to set the _default_ toolchain to 22 while still including overrides for `maven-javadoc-plugin` and `maven-surefire-plugin`. - Continue to use JDK 11 for Javadoc, as [we're doing now](https://github.com/google/guava/blob/5041fbe61965a73ea269c7c24ea746d89bd1b1ba/.github/workflows/ci.yml#L89-L99) because of [problems with at least JDK 21](#7109). - What matters might actually be the version used [by _JDiff_](#6549 (comment)), which comes from the version in the linked `ci.yml` file. But since we're using JDK 11 currently for docs in general, I'm sticking with that for now. Still, we should consider [upgrading the version used for Javadoc generation](#6790 (comment)). But this CL is already complicated enough.... - When we hard-code JDK 11, we need to change the `<source>${java.specification.version}</source>` line: That would otherwise set (for example) `-source 17` when running Maven under JDK 17, and JDK 11 wouldn't recognize it. As I recall, the `java.specification.version` usage was from the days in which we tried to inherit Javadoc from the JDK. Inheritance had [stopped working](#6790), and we ripped it out in cl/614693592. I first tried going with the default from the JDK whose Javadoc binary we're using, which (again) will now be 11. But that led to a problem in `org.codehaus.plexus.languages.java.jpms.CmdModuleNameExtractor`, which apparently tries to look up the module name for the `-source 11` run but uses the Maven run's JDK instead of the Javadoc toolchain or Maven toolchain. So now I've set it to 8 to match what we use for `maven-compiler-plugin`. (I _thought_ I had remembered that `maven-javadoc-plugin` defaulted to matching `maven-compiler-plugin`, even though that's weird. Maybe the two actually just read from the same Maven property or something, or maybe the behavior changed.) Some other thing I'm wondering: - I wonder if we should activate(?) some of the plugins, including the new toolchain plugins, in the `<plugins>` (not just `<pluginManagement>`) section of the parent `pom.xml`. Might that save us from having to do so in each separate `pom.xml`? (We might actually mostly get away with activating(?) them only in the main `guava` build: That _downloads and registers_ the toolchains, and then at least the other projects' _per-plugin_ toolchain configuration probably finds them? But for the more general configuration to work, I think we at least need to activate(?) `maven-toolchains-plugin` in each? I haven't experimented a ton with this.) - I forgot the other thing while I was typing :( (See also [these notes](#5457 (comment)).) Fixes #7331 RELNOTES=n/a PiperOrigin-RevId: 655592201
In particular: - Use JDK 22 for compilation (also, for any other [affected plugins](https://maven.apache.org/guides/mini/guide-using-toolchains.html#prerequisites)) to [avoid a JDK 11 bug](#7331). - Another way to avoid that bug would be to use JDK 8, which [would also provide a `--release`-like compatibility guarantee](#3990). However, that could complicate [using newer APIs conditionally](#6549). And of course we'd expect JDK 8 to be buggier than JDK 22. (In fact, we still have a workaround or two for JDK 8 bugs (with a brand new one coming in cl/655556207), and we could now remove those—assuming that none of our users use JDK 8 to build Guava outside of our Maven build.) JDK 22 also supports new versions of Error Prone, while JDK 8 does not. - This change also allows us to simplify our Error Prone configuration, which until now needed different profiles in order to support both JDK 8 and JDK 9+. We could now upgrade Error Prone, but I haven't done so yet. There are probably other simplifications that we could perform, as well, such as `maven-javadoc-plugin.additionalJOptions`. - Originally, I'd set up this CL to explicitly set only the toolchain of `maven-compiler-plugin` to 22. I had it using 11 for any other plugins (just Animal Sniffer, maybe?), I think from when I was trying to get toolchains to take effect at all. I've since changed this CL to set the _default_ toolchain to 22 while still including overrides for `maven-javadoc-plugin` and `maven-surefire-plugin`. - Continue to use JDK 11 for Javadoc, as [we're doing now](https://github.com/google/guava/blob/5041fbe61965a73ea269c7c24ea746d89bd1b1ba/.github/workflows/ci.yml#L89-L99) because of [problems with at least JDK 21](#7109). - What matters might actually be the version used [by _JDiff_](#6549 (comment)), which comes from the version in the linked `ci.yml` file. But since we're using JDK 11 currently for docs in general, I'm sticking with that for now. Still, we should consider [upgrading the version used for Javadoc generation](#6790 (comment)). But this CL is already complicated enough.... - When we hard-code JDK 11, we need to change the `<source>${java.specification.version}</source>` line: That would otherwise set (for example) `-source 17` when running Maven under JDK 17, and JDK 11 wouldn't recognize it. As I recall, the `java.specification.version` usage was from the days in which we tried to inherit Javadoc from the JDK. Inheritance had [stopped working](#6790), and we ripped it out in cl/614693592. I first tried going with the default from the JDK whose Javadoc binary we're using, which (again) will now be 11. But that led to a problem in `org.codehaus.plexus.languages.java.jpms.CmdModuleNameExtractor`, which apparently tries to look up the module name for the `-source 11` run but uses the Maven run's JDK instead of the Javadoc toolchain or Maven toolchain. So now I've set it to 8 to match what we use for `maven-compiler-plugin`. (I _thought_ I had remembered that `maven-javadoc-plugin` defaulted to matching `maven-compiler-plugin`, even though that's weird. Maybe the two actually just read from the same Maven property or something, or maybe the behavior changed.) Some other thing I'm wondering: - I wonder if we should activate(?) some of the plugins, including the new toolchain plugins, in the `<plugins>` (not just `<pluginManagement>`) section of the parent `pom.xml`. Might that save us from having to do so in each separate `pom.xml`? (We might actually mostly get away with activating(?) them only in the main `guava` build: That _downloads and registers_ the toolchains, and then at least the other projects' _per-plugin_ toolchain configuration probably finds them? But for the more general configuration to work, I think we at least need to activate(?) `maven-toolchains-plugin` in each? I haven't experimented a ton with this.) - I forgot the other thing while I was typing :( (See also [these notes](#5457 (comment)).) Fixes #7331 RELNOTES=n/a PiperOrigin-RevId: 655647768
I'd mentioned this in #3990 (comment).
It would be nice to be able to refer in source code to newer types (e.g.,
CRC32C
,VarHandle
, and@ReservedStackAccess
)—with, of course, appropriate runtime guards. Of course, that would be incompatible with one of the possible approaches to #3990, which is to use--release
.(Using Java 11 might also still dodge bugs in older versions of javac.)
We'd probably accept a PR for this unless some of our users want to tell us that they build Guava from source themselves with javac 8 (or 9 or 10), in which case the decision would get more complex.
The text was updated successfully, but these errors were encountered: