IcedTea 7 Patches for with-project=bsd - A Plan

[Michael Franz]

On Mon, Mar 9, 2009 at 6:40 AM, Andrew John Hughes <
gnu_andrew at member.fsf.org> wrote:

> 2009/3/9 Andrew John Hughes <gnu_andrew at member.fsf.org>:
> > 2009/3/9 Michael Franz <mvfranz at gmail.com>:
> >> Here are the patches against IcedTea7 changeset:   1623:5bda1b46a64f
> they
> >> allow Fedora 10 to build the bsd-port  changeset:   65:72d04adc33f8.
> The
> >> build is not perfect and does not work as well as an older set of
> patches
> >> (the build fails with some genxx.sh script but finishes fine if re-run).
> >>
> >> Since my original attempt the nio2 was added so, this needs to be
> disabled.
> >> I have tested configured as:
> >> ./configure --without-rhino --disable-liveconnect --disable-xrender
> >> --with-hotspot-build --disable-nio2 --with-project=bsd
> >>
> >
> > b50 on includes NIO2 so disabling it is no longer an option.  The BSD
> > tree should be synced to this.
> >
> >> This almost builds on OS X, but I have not figured out my issue with jar
> and
> >> how specify a new include directory and a new lib directory.  These
> patches
> >> do not allow zero to build.  I did have this working in the previous
> >> version.
> >>
> >
> > twisti added some support for additional lib directories IIRC, when
> > adding OpenSolaris support.
> > Check the bit where you added BSD support.
> >
> >> The ecj build requires gjavah as there are patches that assume gjavah is
> >> used instead of a standard javah, it seems gjavah generates code a
> little
> >> differently.
> >>
> >
> > This is intentional.  The ecj bootstrap is designed for GNU
> > Classpath-based solutions and was designed around gcj.  It probably
> > should be split into individual tests for specific bits of missing
> > functionality.
> >
> >> bsd-Makefile.am.patch should indicate how I worked the patches into the
> >> build system.
> >>
>
> I'm not sure what this does.  It seems to introduce a whole heap of
> duplicate patches which would be impossible to maintain.
>

What I did is determine which IcedTea patches fail on the BSD port and moved
the failing parts into two files.  The original failing part is in jdk7
directory which are applied when using IcedTea's normal build. The same
patches are then applied to the bsd port and put into bsd directory.

The failing parts are mostly where there are new conditionals around BSD and
caused the files to be different enough that patch could not figure out how
to apply them.  If the FUZZ level is increased from 0 to 3 more of he
patches apply without modification.


>
> Note that we want most IcedTea patches to go upstream where possible,
> and some of those in your version of IcedTea have already.
>

I am not sure what you mean by this.  The version I started with is not that
old maybe 2 weeks.  The BSD port is constantly merging in changes from up
stream.  Were there many patches in those two weeks?  Is this more of a
timing issue?  Or is this a difference between IcedTea 6 and 7?


> >> Michael
> >>
> >>
> >>
> >
> > --
> > Andrew :-)
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/distro-pkg-dev/attachments/20090309/e3fb879e/attachment.html