Hi,
gcc-4.1.1-12, which contains libgcjwebplugin.so, went into Rawhide last night along with java-1.4.2-gcj-compat-plugin which installs an alternatives-managed "libjavaplugin.so" symlink in /usr/lib/mozilla/plugins.
I closed the long-standing "Inclusion of gcjwebplugin" bug:
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=127537
and updated the release notes with a "Handling Java Applets" section:
http://fedoraproject.org/wiki/Docs/Beats/Java
Rawhide users can install gcjwebplugin with "yum install java-1.4.2-gcj-compat-plugin". Please try this out, file bugs and comment on the release note additions.
The GNU Classpath AWT and Swing code was in a state of flux when the backport that included gcjwebplugin was done (because of a large Java2D rewrite to make the Cairo backend the default), so there are some regressions in AWT and Swing applications for FC6test2.
My post-test2 plan for gcjwebplugin is to gage the reaction of Rawhide users, and also to do another GUI backport from GNU Classpath HEAD to Fedora's libgcj, to fix the AWT and Swing regressions. If the reaction turns out to be overly negative, even after the new AWT and Swing backport, we can always make gcjwebplugin not installed by default, for FC6test3. For FC6test2 though, it is installed by default if the Internet group is selected.
Tom
"Tom" == Thomas Fitzsimmons fitzsim@redhat.com writes:
Tom> gcc-4.1.1-12, which contains libgcjwebplugin.so, went into Tom> Rawhide last night along with java-1.4.2-gcj-compat-plugin which Tom> installs an alternatives-managed "libjavaplugin.so" symlink in Tom> /usr/lib/mozilla/plugins.
Congratulations! I know that a bunch of folks worked very hard to make this happen.
Tom> My post-test2 plan for gcjwebplugin is to gage the reaction of Tom> Rawhide users,
FWIW -- I've been using Anthony's earlier jamvm-based plugin on my laptop. I use it almost daily (always with the same applet :-), it has been quite nice.
Tom> and also to do another GUI backport from GNU Classpath HEAD to Tom> Fedora's libgcj, to fix the AWT and Swing regressions.
I wanted to follow up the earlier conversation about FC6 gcj -- the ecj work really was not ready on time, we missed it. It was all we could do to get this big backport in place. That is pretty disappointing... but FC7 will have an enormous upgrade.
Tom
Hi,
On Tue, 2006-07-25 at 15:34 -0400, Thomas Fitzsimmons wrote:
I closed the long-standing "Inclusion of gcjwebplugin" bug:
So cool. And it works nicely. Good work!
The GNU Classpath AWT and Swing code was in a state of flux when the backport that included gcjwebplugin was done (because of a large Java2D rewrite to make the Cairo backend the default), so there are some regressions in AWT and Swing applications for FC6test2.
My post-test2 plan for gcjwebplugin is to gage the reaction of Rawhide users, and also to do another GUI backport from GNU Classpath HEAD to Fedora's libgcj, to fix the AWT and Swing regressions.
My own ideas/plans are as follows:
- GNU Classpath branched for 0.92 last week and a full release is expected early next week. - There is already a first try of a merge of this new version with libgcj head: http://gcc.gnu.org/ml/java/2006-07/msg00057.html (If you want to try this, please read the whole thread, there are some, ehe, subtleties involving svn, merge and creating patches...) - I will try to update that merge to the new classpath-0_92-branch-point today. And hopefully get it committed during the weekend. - When GNU Classpath 0.92 releases get merge updated with last minute release critical patches. - Merge current backport with the refreshed libgcj trunk import.
For that last point it would be really nice to have a svn branch in upstream gcc. Since I am sure we could share some work with other distributions. Matthias (CCed) already tried to sync the Debian gcc packages with the Fedora packages, but we had some trouble getting a working patch out of the srpm. Hopefully we can get the Fedora patch set into svn somehow so we can get svn merge to help us with the update.
Are there any specific deadlines for when any of our plans need to be done?
- gcc 4.2 branch (unknown) - deadline for libgcj/classpath import. - FC6 Test3 Development Freeze (August 23) - deadline for libgcj backport. But probably earlier if we need to do a mass rebuild of all pacakges depending on the new libgcj.
Any parties/persons that we need to keep in the loop/that we need to get approval of for any of the above?
If the reaction turns out to be overly negative, even after the new AWT and Swing backport, we can always make gcjwebplugin not installed by default, for FC6test3. For FC6test2 though, it is installed by default if the Internet group is selected.
Is there a security group for Fedora? If so, maybe they can take a look at the run plugin warning dialogs and give some feedback on it.
Cheers,
Mark
Mark Wielaard writes:
Any parties/persons that we need to keep in the loop/that we need to get approval of for any of the above?
We've got to get the support of the {lib}gcj maintainers and Fedora releng.
According to the Fedora schedule, the FC6 feature freeze has already happened, so we can't add anything new. We can fix bugs, however.
Merging Classpath into Test 3 should be done if we are sure that it fixes important bugs and doesn't cause regressions. We already know that there were some regressions with the Test 2 merge, for example, and we have a chance to fix them. But Test 3 is too late to be causing regressions.
However, if we come up with a list of important bugs that will be fixed by a bulk merge, and we can convince ourselves that there won't be serious regressions, then we could do it.
Andrew.
"Mark" == Mark Wielaard mark@klomp.org writes:
Mark> For that last point it would be really nice to have a svn branch in Mark> upstream gcc. Since I am sure we could share some work with other Mark> distributions. Matthias (CCed) already tried to sync the Debian gcc Mark> packages with the Fedora packages, but we had some trouble getting a Mark> working patch out of the srpm. Hopefully we can get the Fedora patch set Mark> into svn somehow so we can get svn merge to help us with the update.
Jakub had asked for a patch rather than a branch, so that's what we did.
But really we can do both -- I wish I'd thought of this when doing the merge. That is, we can make a branch, do stuff there, and then just diff it against the RH 4.1 branch whenever we need to provide Jakub with an updated patch.
Tom
Hi,
On Mon, 2006-07-31 at 14:34 -0600, Tom Tromey wrote:
"Mark" == Mark Wielaard mark@klomp.org writes:
Mark> For that last point it would be really nice to have a svn branch in Mark> upstream gcc. Since I am sure we could share some work with other Mark> distributions. Matthias (CCed) already tried to sync the Debian gcc Mark> packages with the Fedora packages, but we had some trouble getting a Mark> working patch out of the srpm. Hopefully we can get the Fedora patch set Mark> into svn somehow so we can get svn merge to help us with the update.
Jakub had asked for a patch rather than a branch, so that's what we did.
But really we can do both -- I wish I'd thought of this when doing the merge. That is, we can make a branch, do stuff there, and then just diff it against the RH 4.1 branch whenever we need to provide Jakub with an updated patch.
Note that Jakub added the whole patch to redhat/gcc-4_1-branch svn branch in the upstream gcc repository now. http://gcc.gnu.org/viewcvs?view=rev&revision=115879 So we can base our work on that by doing some svn merge magic :)
Cheers,
Mark
java-devel@lists.fedoraproject.org
-
Andrew Haley -
Mark Wielaard -
Thomas Fitzsimmons -
Tom Tromey