Mark, On 4/2/15 3:52 PM, Mark Thomas wrote: > On 02/04/2015 15:19, Christopher Schultz wrote: >> Mark, >> >> On 4/1/15 3:29 AM, Mark Thomas wrote: >>> On 31/03/2015 22:48, Konstantin Kolinko wrote: >>>> 2015-04-01 0:09 GMT+03:00 Konstantin Kolinko <knst.koli...@gmail.com>: >>>>> 2015-03-31 22:15 GMT+03:00 <ma...@apache.org>: >>>>>> Author: markt >>>>>> Date: Tue Mar 31 19:15:48 2015 >>>>>> New Revision: 1670437 >>>>>> >>>>>> URL: http://svn.apache.org/r1670437 >>>>>> Log: >>>>>> Fix https://bz.apache.org/bugzilla/show_bug.cgi?id=57765 >>>>>> File.lastModified() has a resolution of 1s (1000ms). The last modified >>>>>> time has to be more than 1000ms ago to ensure that modifications that >>>>>> take place in the same second are not missed. >>>>> >>>>> IIRC, file modification time on FAT file system has resolution of 2s, >>>>> >>>>> https://msdn.microsoft.com/en-us/library/windows/desktop/ms724290%28v=vs.85%29.aspx >>> >>> Given the limitations of FAT32 on Windows (max partition 32GB, files no >>> bigger than 4GB) I'd be surprised if more than a handful of people were >>> running Tomcat on FAT in production. >>> >>>> The following tests are failing with Tomcat 7, >>>> TEST-org.apache.catalina.startup.TestHostConfigAutomaticDeployment.BIO.txt >>>> TEST-org.apache.catalina.startup.TestHostConfigAutomaticDeployment.NIO.txt >>>> >>>> and are also failing in trunk (NIO, NIO2 there). >>> >>> Looks like the change broke that test everywhere. I'll take a look at >>> why now. My guess is the test needs to allow for the offset. If that is >>> the case, I'll make it a protected constant so we can use it in >>> HostConfig and in the test. >> >> Could you make that constant default to e.g. 1000ms but pick-up an >> overridable value from a >> org.apache.catalina.startup.HostConfig.warFileMinAgetoTriggerReload >> system property of something? I can see us quickly finding the one >> person running Tomcat on FAT32 :) > > I could, but ignoring the fact I loath system properties, I'm not sure > it is worth the effort. The one person hasn't complained about BZ 57765 > in the last 10+ years so in the unlikely even they do now we can point > them to the workarounds. > > If we do end up implementing a config option, it really needs to be a > property on the Host.
Understood. -chris
signature.asc
Description: OpenPGP digital signature
