[OpenJDK 2D-Dev] hg: jdk7/2d/jdk: 6897844: Fix broken build on newer versions of X11 (libXext = 1.1.0)
Changeset: 90bdc961b3cb Author:andrew Date: 2009-11-03 23:23 + URL: http://hg.openjdk.java.net/jdk7/2d/jdk/rev/90bdc961b3cb 6897844: Fix broken build on newer versions of X11 (libXext = 1.1.0) Summary: Recent changes to X11's header structure break the build Reviewed-by: prr, flar ! src/solaris/native/sun/awt/awt_GraphicsEnv.h
Re: [OpenJDK 2D-Dev] AWT Dev [PATCH FOR APPROVAL]: Fix broken build on newer versions of X11 (libXext = 1.1.0)
awt_Graphics and XShm is more for 2D than AWT, but I'm not sure how much it matters for this small change. Attach the patch to a bugzilla report .. someone will need to generate a sun bug id too. Can you post a zip of the webvrev somewhere? And is there an X11 reference you can cite to this apparent source incompatible change there? -phil. Andrew John Hughes wrote: With the new version of X11 (specifically libXext = 1.1), the XShm.h header has been refactored. As a result, the build fails on awt_GraphicsEnv.c. This simple patch: http://cr.openjdk.java.net/~andrew/xshm/webrev.01 fixes the issue, without affecting older versions. It's trivial, but very important; this new X11 is already in Gentoo, it'll be in F12 (where we first discovered this issue), and it's no doubt heading to an Ubuntu near you soon. The patch was contributed by Diego Pettenò flamee...@gmail.com, who I'm informed has signed the SCA. Does this look ok? If so, can I have a bug ID to push this to the awt-gate (or wherever is appropriate)? Thanks,
Re: [OpenJDK 2D-Dev] AWT Dev [PATCH FOR APPROVAL]: Fix broken build on newer versions of X11 (libXext = 1.1.0)
2009/11/3 Phil Race phil.r...@sun.com: awt_Graphics and XShm is more for 2D than AWT, but I'm not sure how much it matters for this small change. It's called awt_Graphics hence the AWT list. I doubt the distinction between 2d and awt classes is clear to anyone outside Sun. Attach the patch to a bugzilla report .. someone will need to generate a sun bug id too. Can you post a zip of the webvrev somewhere? I'm aware we need a Sun bug ID; that's why I asked for one to be allocated in the e-mail. I have commit rights so I don't need mentoring; I just need a review and a bug ID so I can push the fix. I don't see why you need all this other superfluous stuff, as it wasn't needed for any of my other pushes to various repos. Is the patch ok? If so, could you please allocate it a bug ID. And is there an X11 reference you can cite to this apparent source incompatible change there? There's http://lists.x.org/archives/xorg-devel/2009-June/001242.html but I avoided posting this in the original mail because it seems to have changed again between that commit and the final release, presumably due to compatibility issues (XShm.h is back and it's now shmproto.h as seen in the patch). I've built the repo with this patch here with the old version, and others have built it with the new version; it does work for both. The same patch is already in Gentoo's ebuild and IcedTea, and a similar patch has been used for the Fedora rawhide RPMs for some time. It would be good to get it upstream as well. -phil. Andrew John Hughes wrote: With the new version of X11 (specifically libXext = 1.1), the XShm.h header has been refactored. As a result, the build fails on awt_GraphicsEnv.c. This simple patch: http://cr.openjdk.java.net/~andrew/xshm/webrev.01 fixes the issue, without affecting older versions. It's trivial, but very important; this new X11 is already in Gentoo, it'll be in F12 (where we first discovered this issue), and it's no doubt heading to an Ubuntu near you soon. The patch was contributed by Diego Pettenò flamee...@gmail.com, who I'm informed has signed the SCA. Does this look ok? If so, can I have a bug ID to push this to the awt-gate (or wherever is appropriate)? Thanks, -- Andrew :-) Free Java Software Engineer Red Hat, Inc. (http://www.redhat.com) Support Free Java! Contribute to GNU Classpath and the OpenJDK http://www.gnu.org/software/classpath http://openjdk.java.net PGP Key: 94EFD9D8 (http://subkeys.pgp.net) Fingerprint: F8EF F1EA 401E 2E60 15FA 7927 142C 2591 94EF D9D8
Re: [OpenJDK 2D-Dev] AWT Dev [PATCH FOR APPROVAL]: Fix broken build on newer versions of X11 (libXext = 1.1.0)
The sun bug ID is 6897844. Jennifer Phil Race wrote: PS Is the patch ok? yes. -phil.
Re: [OpenJDK 2D-Dev] AWT Dev [PATCH FOR APPROVAL]: Fix broken build on newer versions of X11 (libXext = 1.1.0)
Andrew John Hughes wrote: There's http://lists.x.org/archives/xorg-devel/2009-June/001242.html but I avoided posting this in the original mail because it seems to have changed again between that commit and the final release, presumably due to compatibility issues (XShm.h is back and it's now shmproto.h as seen in the patch). I've built the repo with this patch here with the old version, and others have built it with the new version; it does work for both. The same patch is already in Gentoo's ebuild and IcedTea, and a similar patch has been used for the Fedora rawhide RPMs for some time. It would be good to get it upstream as well. At first I was going to ask how the existing #include succeeds when the link says that Xshm.h is going away, but now I see that you said they brought it back. What is it now? Just an empty include to prevent #include failures? (I don't see how that works since the build will break anyway as soon as a missing constant is referenced...?) (It seems odd that they bring it back to [not really] avoid build breakages, but then don't just have it include the new split files to finish the backwards compatibility story...?) ...jim
Re: [OpenJDK 2D-Dev] AWT Dev [PATCH FOR APPROVAL]: Fix broken build on newer versions of X11 (libXext = 1.1.0)
2009/11/3 Jim Graham jim.a.gra...@sun.com: Andrew John Hughes wrote: There's http://lists.x.org/archives/xorg-devel/2009-June/001242.html but I avoided posting this in the original mail because it seems to have changed again between that commit and the final release, presumably due to compatibility issues (XShm.h is back and it's now shmproto.h as seen in the patch). I've built the repo with this patch here with the old version, and others have built it with the new version; it does work for both. The same patch is already in Gentoo's ebuild and IcedTea, and a similar patch has been used for the Fedora rawhide RPMs for some time. It would be good to get it upstream as well. At first I was going to ask how the existing #include succeeds when the link says that Xshm.h is going away, but now I see that you said they brought it back. What is it now? Just an empty include to prevent #include failures? (I don't see how that works since the build will break anyway as soon as a missing constant is referenced...?) (It seems odd that they bring it back to [not really] avoid build breakages, but then don't just have it include the new split files to finish the backwards compatibility story...?) ...jim It's quite convoluted, that's why I was just going to avoid posting the link, as it makes things even more confusing. I believe the reinstated XShm.h does have content that was still needed. The initial version I linked to did remove XShm.h, so the original fix for Fedora 12 removed XShm.h, added the two additional headers and defined some other stuff which I believe was in XShm.h originally. It was a pretty nasty patch, hence why it wasn't committed to IcedTea or OpenJDK. I gather now that XShm.h is back and has the additional material in it. I don't have a copy locally to check, but several people have said this fix works and Fedora RPMs have been built with the original fix. More importantly, I have confirmed myself that it doesn't break earlier versions, which are still used on the majority of systems. It's now several months on from our initial discovery of the problem and more and more people are asking about this in e-mail and on IRC, so a general fix is needed and this fits the bill. Hope that makes some sense! Thanks, -- Andrew :-) Free Java Software Engineer Red Hat, Inc. (http://www.redhat.com) Support Free Java! Contribute to GNU Classpath and the OpenJDK http://www.gnu.org/software/classpath http://openjdk.java.net PGP Key: 94EFD9D8 (http://subkeys.pgp.net) Fingerprint: F8EF F1EA 401E 2E60 15FA 7927 142C 2591 94EF D9D8
Re: [OpenJDK 2D-Dev] AWT Dev [PATCH FOR APPROVAL]: Fix broken build on newer versions of X11 (libXext = 1.1.0)
2009/11/3 Phil Race phil.r...@sun.com: Andrew John Hughes wrote: My superfluous comment actually referred to the additional request for an OpenJDK bugzilla entry. I fail to see the point of this, given a Sun bug ID is still needed to commit. Most of the bugs there just seem to be in danger of bitrotting, and I'd prefer to avoid adding one that's just going to be closed fairly swiftly anyway. It would be nice if we could use OpenJDK bugzilla IDs for commits, and thus didn't have to hassle Sun employees for Sun bug IDs. But that still doesn't seem to have been implemented. Ah .. yes .. well you may be right you don't need that if you can push it directly. I keep having to look up that part of the process myself. But IIIRC theory its supposed to be used to submit patches, not report bugs (sans patch), and you had a patch, which is why I suggested it. I guess I'm as confused as you are regarding it, so I've just tended to go with what I've found to work. The impression I got from the announcement was that, at the moment, it's just for posting patches that need a sponsor/mentor to get them into the repository (i.e. the situations that lead to a 'Contributed by' tag). It was supposed to be being developed into something that would replace the Sun bug ID system altogether for external contributors, but things seem to have gone no further since the launch (no doubt in part due to the acquisition and various other things taking precedence). As such, it's currently a bit pointless for those with commit access as a Sun bug ID is still needed, regardless. Commits were supposed to support using OpenJDK IDs, but this has never happened. The gory details are at: http://openjdk.java.net/groups/web/bugzilla.html Stages 2 and 3 have not come to fruition, and even the 'one-line change' to jcheck hasn't happened. Is the patch ok? If so, could you please allocate it a bug ID. I overlooked that in your email. But I already asked Jennifer to allocate one. Thanks. I'll push once it's allocated. Jennifer says she's doing it now. -phil. -- Andrew :-) Free Java Software Engineer Red Hat, Inc. (http://www.redhat.com) Support Free Java! Contribute to GNU Classpath and the OpenJDK http://www.gnu.org/software/classpath http://openjdk.java.net PGP Key: 94EFD9D8 (http://subkeys.pgp.net) Fingerprint: F8EF F1EA 401E 2E60 15FA 7927 142C 2591 94EF D9D8
Re: [OpenJDK 2D-Dev] AWT Dev [PATCH FOR APPROVAL]: Fix broken build on newer versions of X11 (libXext = 1.1.0)
Andrew John Hughes wrote: 2009/11/3 Phil Race phil.r...@sun.com: awt_Graphics and XShm is more for 2D than AWT, but I'm not sure how much it matters for this small change. It's called awt_Graphics hence the AWT list. I doubt the distinction between 2d and awt classes is clear to anyone outside Sun. But Graphics is I'd hope obviously 2D, and lots of things have AWT in the name as hangovers from JDk 1.0, 1.1, where there was no 2D. Attach the patch to a bugzilla report .. someone will need to generate a sun bug id too. Can you post a zip of the webvrev somewhere? I'm aware we need a Sun bug ID; that's why I asked for one to be allocated in the e-mail. I have commit rights so I don't need mentoring; I just need a review and a bug ID so I can push the fix. I don't see why you need all this other superfluous stuff, as it wasn't needed for any of my other pushes to various repos. The superfluous stuff is the copy of the webrev? We archive them. Not all groups do that. Swing, AWT and 2D do. Occasionally someone may fail to get one from a contribution but its still the theoretical process to have it. Is the patch ok? If so, could you please allocate it a bug ID. I overlooked that in your email. But I already asked Jennifer to allocate one. And is there an X11 reference you can cite to this apparent source incompatible change there? There's http://lists.x.org/archives/xorg-devel/2009-June/001242.html but I avoided posting this in the original mail because it seems to have changed again between that commit and the final release, presumably due to compatibility issues (XShm.h is back and it's now shmproto.h as seen in the patch). I've built the repo with this patch here with the old version, and others have built it with the new version; it does work for both. The same patch is already in Gentoo's ebuild and IcedTea, and a similar patch has been used for the Fedora rawhide RPMs for some time. It would be good to get it upstream as well. OK .. although I was looking for something where they pointed out this was likely to cause build failures but was justified because ... -phil. -phil. Andrew John Hughes wrote: With the new version of X11 (specifically libXext = 1.1), the XShm.h header has been refactored. As a result, the build fails on awt_GraphicsEnv.c. This simple patch: http://cr.openjdk.java.net/~andrew/xshm/webrev.01 fixes the issue, without affecting older versions. It's trivial, but very important; this new X11 is already in Gentoo, it'll be in F12 (where we first discovered this issue), and it's no doubt heading to an Ubuntu near you soon. The patch was contributed by Diego Pettenò flamee...@gmail.com, who I'm informed has signed the SCA. Does this look ok? If so, can I have a bug ID to push this to the awt-gate (or wherever is appropriate)? Thanks,
Re: [OpenJDK 2D-Dev] AWT Dev [PATCH FOR APPROVAL]: Fix broken build on newer versions of X11 (libXext = 1.1.0)
Yes, indeed, that all makes sense for your fix. I wasn't intending to register an objection with the fix, I was just curious about the changes they made which, as you say, seem quite convoluted... ...jim Andrew John Hughes wrote: 2009/11/3 Jim Graham jim.a.gra...@sun.com: Andrew John Hughes wrote: There's http://lists.x.org/archives/xorg-devel/2009-June/001242.html but I avoided posting this in the original mail because it seems to have changed again between that commit and the final release, presumably due to compatibility issues (XShm.h is back and it's now shmproto.h as seen in the patch). I've built the repo with this patch here with the old version, and others have built it with the new version; it does work for both. The same patch is already in Gentoo's ebuild and IcedTea, and a similar patch has been used for the Fedora rawhide RPMs for some time. It would be good to get it upstream as well. At first I was going to ask how the existing #include succeeds when the link says that Xshm.h is going away, but now I see that you said they brought it back. What is it now? Just an empty include to prevent #include failures? (I don't see how that works since the build will break anyway as soon as a missing constant is referenced...?) (It seems odd that they bring it back to [not really] avoid build breakages, but then don't just have it include the new split files to finish the backwards compatibility story...?) ...jim It's quite convoluted, that's why I was just going to avoid posting the link, as it makes things even more confusing. I believe the reinstated XShm.h does have content that was still needed. The initial version I linked to did remove XShm.h, so the original fix for Fedora 12 removed XShm.h, added the two additional headers and defined some other stuff which I believe was in XShm.h originally. It was a pretty nasty patch, hence why it wasn't committed to IcedTea or OpenJDK. I gather now that XShm.h is back and has the additional material in it. I don't have a copy locally to check, but several people have said this fix works and Fedora RPMs have been built with the original fix. More importantly, I have confirmed myself that it doesn't break earlier versions, which are still used on the majority of systems. It's now several months on from our initial discovery of the problem and more and more people are asking about this in e-mail and on IRC, so a general fix is needed and this fits the bill. Hope that makes some sense! Thanks,
Re: [OpenJDK 2D-Dev] AWT Dev [PATCH FOR APPROVAL]: Fix broken build on newer versions of X11 (libXext = 1.1.0)
2009/11/3 Jennifer Godinez jennifer.godi...@sun.com: The sun bug ID is 6897844. Jennifer Phil Race wrote: PS Is the patch ok? yes. -phil. Pushed: http://hg.openjdk.java.net/jdk7/2d/jdk/rev/90bdc961b3cb Thanks everyone, -- Andrew :-) Free Java Software Engineer Red Hat, Inc. (http://www.redhat.com) Support Free Java! Contribute to GNU Classpath and the OpenJDK http://www.gnu.org/software/classpath http://openjdk.java.net PGP Key: 94EFD9D8 (http://subkeys.pgp.net) Fingerprint: F8EF F1EA 401E 2E60 15FA 7927 142C 2591 94EF D9D8
Re: [OpenJDK 2D-Dev] AWT Dev [PATCH FOR APPROVAL]: Fix broken build on newer versions of X11 (libXext = 1.1.0)
PS Is the patch ok? yes. -phil.