Skip to:

  1. Skip to Jira Navigation
  2. Skip to Side Navigation
  3. Skip to Main Content

review date and time handling for Java 8

Description

As part of the v4/Java 8 changes, we need to review the use of date/time APIs.

There is one use of Joda-Time in ComponentInfo, which uses Joda-Time's DateTime as start/end instants. These are based on The ISO chronology and the UTC time zone, presumably so that if you just .toString them, you'll get something sane and consistent. It seems to me likely, however, that these should in fact be {{Instant}}s both in Joda-Time and under Java 8 and JSR-310, with time zone assignment and formatting left to the caller. There are no real uses of these objects within the MDA codebase at present, so this looks low impact.

Joda-Time conversion stuff:

http://blog.joda.org/2014/11/converting-from-joda-time-to-javatime.html

http://blog.joda.org/2009/11/why-jsr-310-isn-joda-time_4941.html

Most other time-related functionality seems to be based around millisecond representations. I guess that's fine with the presence of the @Duration annotation and the DurationToLongConverter in XML configurations. We should review anyway, though.

Environment

None

Activity

Show:

Ian YoungMarch 20, 2019 at 4:22 PM

commit a54e14305561cba4f4103041b41b436ee2ff3822 revises PullUpCacheDurationStage, PullUpValidUntilStage, SetCacheDurationStage, SetValidUntilStage, ValidateValidUntilStage so that their APIs are no longer long-based but Instant or Duration based as appropriate.

Ian YoungMarch 14, 2019 at 12:14 PM

commit 9822edc0814ae77f60db58dae2758d92b589dabc uses the maven-enforcer-plugin to ensure that the MDA no longer has either a direct or an indirect dependency joda-time, and that such dependencies can't sneak back in without our noticing.

Ian YoungMarch 14, 2019 at 10:59 AM

I think the long-as-milliseconds parts of the API really need to go as well, in order to be in line with common sense and the other Shibboleth components. There are quite a few of these in places like SetValidUntilStage.

Ian YoungFebruary 2, 2018 at 11:19 AM

Ran into some issues with the multi-JDK test, because Java 9 gives more sub-second digits when you stringify an Instant. This doesn't seem like a problem in principle, but should be mentioned in the release notes.

Ian YoungJanuary 31, 2018 at 6:08 PM

This has removed the explicit dependency on Joda-Time, but there is a transitive one coming in from java-support. That has ticket JSPT-79. Hold this open until that has at least been reviewed.

Done

Details

Assignee

Reporter

Fix versions

Affects versions

Created August 17, 2017 at 3:31 PM
Updated May 16, 2024 at 12:37 PM
Resolved March 21, 2019 at 2:21 PM