Experimental patchsets
Greg Lewis
glewis at misty.eyesbeyond.com
Thu Jun 5 08:01:15 PDT 2003
On Thu, Jun 05, 2003 at 11:32:01AM +0200, Sheldon Hearn wrote:
> On (2003/06/05 11:44), Alexey Zelkin wrote:
>
> > Can't say anything for jdk13, but it's bad idea for jdk14. At least now.
> >
> > Those who'll be able to build jdk14 from patchset and wanna try it
> > should expect problems and be ready for them.
>
> But those who are able to build jdk14 from a port marked -devel should
> also expect problems and be ready for them. :-)
>
> Seriously, I don't see how increased exposure could hurt.
I'm not saying that its a bad idea, but here are some of the problems:
1. The checksum will change every week when the next patchset is
auto-generated. This one is actually pretty easy to solve though,
I can auto-commit the checksum change when I upload the new patchset.
2. Any given patchset may not compile, may break in other ways, may
require arbitrary changes to the port, etc. Staying with the old
patchset won't be an option since its getting overwritten, so the
maintainer is going to have to do quite a bit of maintenance to
keep the port in sync with the patchset. Also note the havoc
this may cause should another port come to depend on jdk1?-devel.
Note that the second one specifically defeats my reasons for putting
out experimental patchsets - a low (time) cost way (for me) to get
out more recent work for people to test/use without having to do
release testing and the like. That is to say, I'm all for someone
who wants to do the work to maintain this, I'll even do the automated
commit of the checksum changes as noted above, but I won't be creating
such a port.
--
Greg Lewis Email : glewis at eyesbeyond.com
Eyes Beyond Web : http://www.eyesbeyond.com
Information Technology FreeBSD : glewis at FreeBSD.org
More information about the freebsd-java
mailing list