RFR of 8129632: (fs) Files.probeContentType returns null on Mac OS X

Brian Burkhalter brian.burkhalter at oracle.com
Tue Jun 30 01:07:04 UTC 2015

On Jun 29, 2015, at 8:35 AM, Brian Burkhalter <brian.burkhalter at oracle.com> wrote:

>> In NioLibrary.gmk then you probably should split the LDFLAGS_SUFFIX_macosx line to avoid the long line.


>> The protected static method in sun.nio.fs.AbstractFileTypeDetector is a bit inconsistent. Can you change this to be final and change the comment so that it's consistent with the other final method.


>> Are you confident that the update test and the image types will pass everywhere? We've had to dial-down the tests in this area because there is no guarantee that all test machines will the mappings.
> I’ve run these through one of the regression test queues several times as well as several VMs and this set is what it boiled down to. I can try one or two other queues before pushing.

I ran the test on a different queue and saw some failures on Linux and Solaris variants so I dialed this back to test only Mac OS X for now which is what is of concern here anyway.

>> Also in the delete then you can use Files.delete(file).
> OK


>> For Windows then startingWith("Windows") is commonly used.
> In the various tests what is checked for in “os.name” seems to be all over the map vis-a-vis upper and lower case, startsWith() and contains().

No longer pertinent.

>> The comments in UTIFileTypeDetector.c need to be cleaned up before this is pushed. I don't think you need to reference MacOSXPreferencesFile.m as it's only a tiny amount of duplication. Bets to keep these /* */ style to be consistent with the other code.
> OK.

Done. Also added class doc to UTIFileTypeDetector.java.

An updated patch is viewable here:



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/nio-dev/attachments/20150629/cb820675/attachment-0001.html>

More information about the nio-dev mailing list