RFR: 8015666: test/tools/pack200/TimeStamp.java failing
kumar.x.srinivasan at oracle.com
Thu Jun 27 17:04:25 UTC 2013
I started looking at this, my initial comment, the Unpacker.unpack
does not close its output and we allow multiple pack files to be
I am assuming out.finish() will allow further jar files to be appended ?
or would this cause a problem ?
> The zip time related changes I pushed back last month appears
> to have the compatibility risk of breaking existing code. The main
> idea in that changeset is to use the more accurate and timezone
> insensitive utc time stored in the extra field for the
> if possible. However it turns out the reality is that the code out there
> might have already had some interesting workaround/hack solution
> to workaround the problem that the time stamp stored in the "standard'
> zip entry header is a MS-DOS standard date/time, which is a "local
> date/time" and sensitive to timezone, in which, if the zip is archived
> in time zone A (our implementation converts the "java" time to dos
> time by using the default tz A) and then transferred/un-archived in
> a different zone B (use default tz B to convert back to java time), we
> have a time stamp mess up. The "workaround" from pack200 for this
> issue when pack/unpacking a jar file is to "specify/recommend/suggest"
> in its spec that the "time zone" in a jar file entry is assumed to be
> so the pack/unpack200 implementation set the "default time" to utc
> before the pack/unpack and set it back to the original after that. It
> "perfectly" for a roundtrip pack/unpacking, until the changeset , in
> which ZipEntry.getTime() (packing) returns a real utc time and the
> ZipEntry.setTime() (unpacking), then mess up the MS-DOS date/time
> entry, this is the root cause of this regression.
> Given the facts that
> (1) there are actually two real physical time stamps in a zip file
> one is in the date/time fields, which is MS-DOS-local-date/time-with-2-
> seconds-granularity , one is in the extra data field, which is
> (2) and there are applications over there that have dependency on the
> MS-DOS date/time stamp.
> I'm proposing the following approach to add the functionality of
> the "utc-date/time-with-1-second granularity" and keep the old behavior
> of the get/setTime() of the ZipEntry.
> (1) keep the time/setTime()/getTime() for the MS-DOS standard date/time.
> To set via the old setTime() will only store the time into zip's
> date/time field, which is in MS-DOS date/time. And getTime() only
> the date/time from that field, when read from the zip file/stream.
> (2) add mtime/set/getLastModifiedTime() to work on the UTC time fields,
> and the last modified time set via the new method will also set
> the "time",
> and the getLastModifiedTime() also returns the "time", if the UTC
> stamp fields are not set in the zip file header. The idea is that
> for the new
> application, the recommendation is to use
> for better/correct time stamp, but the existing apps keep the
> same behavior.
> (3) jar and ZipOutputStream are updated to use the
> (4) Pack/unpack continues to use the set/getTime(), so the current
> continues work. I will leave this to Kuma to decide how it should
> be handled
> going forward. (there are two facts need to be considered here,
> a) the
> existing jar file might not have the utc time instored, and b)
> all "extra" data
> are wiped out during the pack/unpacking process)
> (5) additionally add another pair of atime/get/setLastAccessTime and
> (6) The newly added 3 pairs of the m/a/ctime get/set methods use the
> nio FileTime, instead of the "long". This may add some additional
> cost of
> conversion when working with them, but may also help improve the
> performance if the time stamps are directly from nio file system
> get/set XYZTime. Good/bad?
> Comment, option and suggestion are appreciated.
>  http://hg.openjdk.java.net/jdk8/tl/jdk/rev/90df6756406f
More information about the core-libs-dev