On Tue, 9 Jun 2026 10:55:03 GMT, Jaikiran Pai <[email protected]> wrote:
>> Can I please get a review of this change which proposes to address >> https://bugs.openjdk.org/browse/JDK-8385924? >> >> For reasons explained in that issue, the `java.util.zip.GZIPInputStream` >> class currently behaves differently depending on which version of Java is in >> use. The change in this current PR proposes to introduce a system property >> which allows applications to decide whether they prefer the decades old >> behaviour of `GZIPInputStream` where it uses `InputStream.available()` for >> deciding whether to read a subsequent member in the stream, or whether the >> application prefers the more recent behaviour of `GZIPInputStream` where it >> skips the `available()` call and instead directly does a read operation for >> detecting a subsequent member. >> >> The system property by default isn't set which means that `GZIPInputStream` >> will call `InputStream.available()` and thus matches its decades old >> behaviour. The introduction of this system property is a stopgap measure as >> noted in https://github.com/openjdk/jdk/pull/30925#discussion_r3318466104 : >> >>> For JDK 27, introduce a system property that controls the reliable handling >>> of streams with more than one member. Long standing behavior does not >>> handle the case where the underlying stream doesn't have a useful >>> avaialble() method and it additionally swallows any exception when reading >>> ahead. The system property can be documented as an implNote in the 2-arg >>> constructor or class description. The property will need default to long >>> standing behavior (not the more recent/changed behavior). >> >> As noted in that same comment, the long term solution involves enhancing the >> `GZIPInputStream` class to introduce a constructor which allows applications >> to decide how the `GZIPInputStream` should behave when dealing with multiple >> members: >> >>> For JDK 28 or later, introduce a new constructor to create a >>> GZIPInputStream that works like it should have done in JDK 1.1 when the API >>> was originally introduced. We discussed two possible sets of parameters. We >>> also discussed maybe deprecating the two existing constructors due to the >>> issue of treating available==0 as EOF and the issue of swallowing >>> exceptions. >> >> That new API will need careful thought and time and thus the current >> proposal to introduce this system property in the meantime. >> >> A new jtreg test has been introduced to verify the usage of this system >> property. A CSR will be proposed shortly. Existing tests and this new test >> continue to pass with this change. >> >> ---------... > > Jaikiran Pai has updated the pull request with a new target base due to a > merge or a rebase. The incremental webrev excludes the unrelated changes > brought in by the merge/rebase. The pull request contains five additional > commits since the last revision: > > - adjust the implNote > - replace unset with "not set" > - minor code comment updates > - merge latest from master branch > - 8385924: GZIPInputStream.read() behaves differently on some Java versions src/java.base/share/classes/java/util/zip/GZIPInputStream.java line 71: > 69: * property in environments that process streams with a series of > members. By default, > 70: * the {@code jdk.util.gzip.tryReadAheadAfterTrailer} system property is > not set, > 71: * and {@code InputStream.available()} gets called. This is quite good, just a bit uncomfortable with the "may not always be accurate". What would you think of this: "GZIPInputStream depends on a high quality implementation of available() to reliably read a stream with a series of members. Consequently it may be necessary to set this property in environments that process streams with a series of members." ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/31396#discussion_r3382003596
