On Thu, 19 Nov 2020 20:20:17 GMT, Brian Burkhalter <b...@openjdk.org> wrote:
>> Please review this modification of `java.io.InputStream.skipNBytes(long)` to >> improve its performance when `skip(long)` skips fewer than the requested >> number of bytes. In the current implementation, `skip(long)` is invoked once >> and, if not enough bytes have been skipped, then `read()` is invoked for >> each of the remaining bytes to be skipped. The proposed implementation >> instead repeatedly invokes `skip(long)` until the requested number of bytes >> has been skipped, or an error condition is encountered. For cases where >> `skip(long)` skips fewer bytes than the number requested, the new version >> was measured to be up to more than one thousand times faster than the old >> version. When `skip(long)` actually skips the requested number of bytes, the >> performance difference is insignificant. > > Brian Burkhalter has updated the pull request incrementally with two > additional commits since the last revision: > > - 8246739: InputStream.skipNBytes could be implemented more efficiently > - 8246739: InputStream.skipNBytes could be implemented more efficiently src/java.base/share/classes/java/io/InputStream.java line 605: > 603: } else if (ns == 0) { // no bytes skipped > 604: // read one byte to check for EOS > 605: if (read() < 0) { Wouldn't it be better to compare return value with -1 (as it was before)? Since the documentation of read method says: _Returns: the next byte of data, or **-1 if the end of the stream is reached**._ ------------- PR: https://git.openjdk.java.net/jdk/pull/1329