On Jul 6, 2013, at 10:38 AM, kristian <m.krist...@web.de> wrote:
> hi,
>
>
> On Sat, Jul 6, 2013 at 3:10 PM, Hirotsugu Asari <asari.r...@gmail.com> wrote:
>
> Good to see a lot of progress being made here (especially CI!). Some
> remaining issues:
>
> tzdata stuff is still a bit iffy. The tzdata version that jruby-core uses is
> hard-coded as ${tzdata.version}. To build with the latest version,
> core/pom.xml must be modified, which is a bit awkward.
> currently you can do
> $ mvn -Dtzdata.version=2013c -Ptzdata
> but the 2013c artifact does exists and must exists before.
OK, this works, but it is quite cumbersome, but I dare say it's *reasonable*.
> one why to make it more convenient is to hardcode the 2013c in to
> core/pom.xml for -Ptzdata case that means
> $ mvn -Ptzdata
>
> or just make the command line simpler:
> $ mvn -Dtzdata.version=2013c
>
> but in both cases the respective timezone artifact must exist beforehand.
> so let me know which version you prefer.
Hardcoding it in core/pom.xml would affect all mvn profiles, correct? So that
seems out of the question.
The latter means that the default maven goal is getting whatever tzdata.version
given, right? That's probably more natural. (And I assume that, without
tzdata.version given in this way, it would default to the version distributed
with the joda-time release we are bundling.)
Incidentally, once I generate the artifact, will Constants.java ever be reset
to the joda-time-supplied version? I had expected that 'mvn clean' would do
this, but it doesn't seem to.
> jar-complete target needs to be converted. (Did I miss this?)
>
> still, missing I am about to have a look at this.
Great! Thanks for the work.
>
> - christian
>
