From linuxhippy at gmail.com Wed May 20 20:15:51 2009 From: linuxhippy at gmail.com (Clemens Eisserer) Date: Wed, 20 May 2009 23:15:51 -0400 Subject: xrender status in jdk7 In-Reply-To: <200905201359.10001.charlyghislain@gmail.com> References: <200905201359.10001.charlyghislain@gmail.com> Message-ID: <194f62550905202015j71e68582ra3fcdec459351db6@mail.gmail.com> Hi Charly, > I see youre working hard on the implementation of Xrender pipeline. > Being using a intel card now, it seems i can't use the opengl pipeline > anymore. This one is thus very welcomed. > > I saw it's on the feature plan for sun's jdk 7. This jdk is availible as > developpement snapshot, but i could'nt find any new pipeline. I guess your > code is kept on the openjdk repository. There are currently two versions of the pipeline: - The "old" one developed as part of the OpenJDK challenge - The "rewrite" currently developed The old version has been integrated into IcedTea, I guess thats the easiest way to try it out. I am however afraid you won't be very happy because: - The latest versions of the intel-driver is in bad shape - OpenJDK uses the pisces rasterizer, which is a lot slowe than the closed-source rasterizer used in Sun's proprietary JDK. The rewrite version is currently developed in private. Not because I don't want to open it, but because I struggle with mercurial - and I doubt somebody would have a look at it anyway ;) I'll upload it to repository as soon as I've some spare time. > Is there a simple way to test it with the sun's jdk 7 (or event 6) ? The > website show instruction for "jxrender" 0.1, with files compiled for 32 bits > systems. Compiling openjdk seems to be unnececessary hard (I need a > bootstrap?) to get a 64 bit version of those files. > > The mailing list seems dead, but if I can test the pipeline I'll try to > report any bug/problems I can find. (To openJDK bugtracker?) The best place to report bugs for now is the mailing list. As far as I know the OpenJDK bug tracker is currently only accepting patches. Thanks a lot, Clemens From mark at klomp.org Fri May 22 07:47:04 2009 From: mark at klomp.org (Mark Wielaard) Date: Fri, 22 May 2009 16:47:04 +0200 Subject: xrender status in jdk7 In-Reply-To: <194f62550905202015j71e68582ra3fcdec459351db6@mail.gmail.com> References: <200905201359.10001.charlyghislain@gmail.com> <194f62550905202015j71e68582ra3fcdec459351db6@mail.gmail.com> Message-ID: <1243003624.3655.97.camel@hermans.wildebeest.org> Hi, On Wed, 2009-05-20 at 23:15 -0400, Clemens Eisserer wrote: > There are currently two versions of the pipeline: > - The "old" one developed as part of the OpenJDK challenge > - The "rewrite" currently developed > > The old version has been integrated into IcedTea, I guess thats the > easiest way to try it out. If you have IcedTea(6) as shipped by most GNU/Linux distros these days then just add -Dsun.java2d.xrender=True to you java command to try it out. > The rewrite version is currently developed in private. Not because I > don't want to open it, but because I struggle with mercurial - and I > doubt somebody would have a look at it anyway ;) > I'll upload it to repository as soon as I've some spare time. Please do! Of course showing the code won't guarantee anybody looks at it, reviews it and/or supplies patches. But not showing the code definitely guarantees nobody will. And of course it makes it a tad bit harder to integrate in newer IcedTea versions :) Cheers, Mark From mark at klomp.org Sun May 24 15:48:04 2009 From: mark at klomp.org (Mark Wielaard) Date: Mon, 25 May 2009 00:48:04 +0200 Subject: OpenJDK7 and the JavaOne preview (M3) release In-Reply-To: <1243063124.3579.19.camel@hermans.wildebeest.org> References: <17c6771e0905080454i51afcf0bn2d731ab657d58c9d@mail.gmail.com> <17c6771e0905190253u8981eddr10fc5aeb9dbe6174@mail.gmail.com> <17c6771e0905201334p795641bn8e677c7f66014aec@mail.gmail.com> <1243004611.3655.105.camel@hermans.wildebeest.org> <17c6771e0905220828q73969929rd35eb388b8a1b9b5@mail.gmail.com> <1243007226.3655.122.camel@hermans.wildebeest.org> <17c6771e0905221035y824ad5fla5a140831c4f0b92@mail.gmail.com> <1243015508.3655.123.camel@hermans.wildebeest.org> <1243063124.3579.19.camel@hermans.wildebeest.org> Message-ID: <1243205284.30320.15.camel@hermans.wildebeest.org> On Sat, 2009-05-23 at 09:18 +0200, Mark Wielaard wrote: > On Fri, 2009-05-22 at 20:05 +0200, Mark Wielaard wrote: > > On Fri, 2009-05-22 at 18:35 +0100, Andrew John Hughes wrote: > > > Are there any negative side-effects to having this code applied all the time? > > > > No, the patches are designed so that without the property set the old > > pipeline is used. > > That said, if you do build with --disable-xrender then the font issue > doesn't occur. So something else is broken with the xrender support :( > I propose that we disable xrender support by default for now while we > figure out what else is borked. Found it. The X11 pipeline initialization is a bit tricky to follow. The xrender patches tried to clear that up at one point by making the initialization of the X11SurfaceData explicit (just like the XRSurfaceData). Although that code is nicer and cleaner, the new font code depends on the implicit static initializer calling order. That is a bit yucky, but lets play along and make sure that no explicit initialization is done: 2009-05-24 Mark Wielaard * patches/xrender/icedtea-001.patch: Initialize X11SurfaceData implicitly, not explicitly. Now both the standard X11 pipeline (default) and the XRender pipeline (-Dsun.java2d.xrender=True) work again against IcedTea7. Cheers, Mark -------------- next part -------------- A non-text attachment was scrubbed... Name: xrender-implicit-init.patch Type: text/x-patch Size: 3940 bytes Desc: not available Url : http://mail.openjdk.java.net/pipermail/xrender-dev/attachments/20090525/daab2d5f/attachment.bin From mark at klomp.org Tue May 26 02:16:20 2009 From: mark at klomp.org (Mark Wielaard) Date: Tue, 26 May 2009 11:16:20 +0200 Subject: OpenJDK7 and the JavaOne preview (M3) release In-Reply-To: <17c6771e0905260137m15adc51j8c59f4c9494374d1@mail.gmail.com> References: <17c6771e0905080454i51afcf0bn2d731ab657d58c9d@mail.gmail.com> <17c6771e0905201334p795641bn8e677c7f66014aec@mail.gmail.com> <1243004611.3655.105.camel@hermans.wildebeest.org> <17c6771e0905220828q73969929rd35eb388b8a1b9b5@mail.gmail.com> <1243007226.3655.122.camel@hermans.wildebeest.org> <17c6771e0905221035y824ad5fla5a140831c4f0b92@mail.gmail.com> <1243015508.3655.123.camel@hermans.wildebeest.org> <1243063124.3579.19.camel@hermans.wildebeest.org> <1243205284.30320.15.camel@hermans.wildebeest.org> <17c6771e0905260135t59013f4dwc7ad25dea9520d28@mail.gmail.com> <17c6771e0905260137m15adc51j8c59f4c9494374d1@mail.gmail.com> Message-ID: <1243329380.20462.14.camel@hermans.wildebeest.org> On Tue, 2009-05-26 at 09:37 +0100, Andrew John Hughes wrote: > > Thanks for finding and fixing this. I narrowed it down to > > X11SurfaceData on Friday (a diff between the non-xrender and xrender > > versions showed this was the only place with a change that would make > > any effect), but it was too late to find a fix. > > BTW, this demonstrates an issue with importing additional code where > the upstream isn't tracking JDK7 updates... That is precisely what these patches do, make sure the code tracks jdk7 so it can be included in IcedTea7. Cheers, Mark From gnu_andrew at member.fsf.org Tue May 26 03:08:32 2009 From: gnu_andrew at member.fsf.org (Andrew John Hughes) Date: Tue, 26 May 2009 11:08:32 +0100 Subject: OpenJDK7 and the JavaOne preview (M3) release In-Reply-To: <1243329380.20462.14.camel@hermans.wildebeest.org> References: <17c6771e0905080454i51afcf0bn2d731ab657d58c9d@mail.gmail.com> <17c6771e0905220828q73969929rd35eb388b8a1b9b5@mail.gmail.com> <1243007226.3655.122.camel@hermans.wildebeest.org> <17c6771e0905221035y824ad5fla5a140831c4f0b92@mail.gmail.com> <1243015508.3655.123.camel@hermans.wildebeest.org> <1243063124.3579.19.camel@hermans.wildebeest.org> <1243205284.30320.15.camel@hermans.wildebeest.org> <17c6771e0905260135t59013f4dwc7ad25dea9520d28@mail.gmail.com> <17c6771e0905260137m15adc51j8c59f4c9494374d1@mail.gmail.com> <1243329380.20462.14.camel@hermans.wildebeest.org> Message-ID: <17c6771e0905260308s14a81f8es18592db5294a6cda@mail.gmail.com> 2009/5/26 Mark Wielaard : > On Tue, 2009-05-26 at 09:37 +0100, Andrew John Hughes wrote: >> > Thanks for finding and fixing this. ?I narrowed it down to >> > X11SurfaceData on Friday (a diff between the non-xrender and xrender >> > versions showed this was the only place with a change that would make >> > any effect), but it was too late to find a fix. >> >> BTW, this demonstrates an issue with importing additional code where >> the upstream isn't tracking JDK7 updates... > > That is precisely what these patches do, make sure the code tracks jdk7 > so it can be included in IcedTea7. > Yes, but my point was we wouldn't have run into this issues if upstream was keeping pace with jdk7 development. We have to be careful which projects we import if we're going to end up maintaining them over successive jdk7 releases, which happen on a weekly basis. Clemens, can you look into incorporating Mark's changes upstream? > Cheers, > > Mark > > 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 From gnu_andrew at member.fsf.org Tue May 26 01:35:58 2009 From: gnu_andrew at member.fsf.org (Andrew John Hughes) Date: Tue, 26 May 2009 09:35:58 +0100 Subject: OpenJDK7 and the JavaOne preview (M3) release In-Reply-To: <1243205284.30320.15.camel@hermans.wildebeest.org> References: <17c6771e0905080454i51afcf0bn2d731ab657d58c9d@mail.gmail.com> <17c6771e0905190253u8981eddr10fc5aeb9dbe6174@mail.gmail.com> <17c6771e0905201334p795641bn8e677c7f66014aec@mail.gmail.com> <1243004611.3655.105.camel@hermans.wildebeest.org> <17c6771e0905220828q73969929rd35eb388b8a1b9b5@mail.gmail.com> <1243007226.3655.122.camel@hermans.wildebeest.org> <17c6771e0905221035y824ad5fla5a140831c4f0b92@mail.gmail.com> <1243015508.3655.123.camel@hermans.wildebeest.org> <1243063124.3579.19.camel@hermans.wildebeest.org> <1243205284.30320.15.camel@hermans.wildebeest.org> Message-ID: <17c6771e0905260135t59013f4dwc7ad25dea9520d28@mail.gmail.com> 2009/5/24 Mark Wielaard : > On Sat, 2009-05-23 at 09:18 +0200, Mark Wielaard wrote: >> On Fri, 2009-05-22 at 20:05 +0200, Mark Wielaard wrote: >> > On Fri, 2009-05-22 at 18:35 +0100, Andrew John Hughes wrote: >> > > Are there any negative side-effects to having this code applied all the time? >> > >> > No, the patches are designed so that without the property set the old >> > pipeline is used. >> >> That said, if you do build with --disable-xrender then the font issue >> doesn't occur. So something else is broken with the xrender support :( >> I propose that we disable xrender support by default for now while we >> figure out what else is borked. > > Found it. The X11 pipeline initialization is a bit tricky to follow. The > xrender patches tried to clear that up at one point by making the > initialization of the X11SurfaceData explicit (just like the > XRSurfaceData). Although that code is nicer and cleaner, the new font > code depends on the implicit static initializer calling order. That is a > bit yucky, but lets play along and make sure that no explicit > initialization is done: > > 2009-05-24 ?Mark Wielaard ? > > ? ? ? ?* patches/xrender/icedtea-001.patch: Initialize X11SurfaceData > ? ? ? ?implicitly, not explicitly. > > Now both the standard X11 pipeline (default) and the XRender pipeline > (-Dsun.java2d.xrender=True) work again against IcedTea7. > > Cheers, > > Mark > Thanks for finding and fixing this. I narrowed it down to X11SurfaceData on Friday (a diff between the non-xrender and xrender versions showed this was the only place with a change that would make any effect), but it was too late to find a fix. -- 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 From gnu_andrew at member.fsf.org Tue May 26 01:37:19 2009 From: gnu_andrew at member.fsf.org (Andrew John Hughes) Date: Tue, 26 May 2009 09:37:19 +0100 Subject: OpenJDK7 and the JavaOne preview (M3) release In-Reply-To: <17c6771e0905260135t59013f4dwc7ad25dea9520d28@mail.gmail.com> References: <17c6771e0905080454i51afcf0bn2d731ab657d58c9d@mail.gmail.com> <17c6771e0905201334p795641bn8e677c7f66014aec@mail.gmail.com> <1243004611.3655.105.camel@hermans.wildebeest.org> <17c6771e0905220828q73969929rd35eb388b8a1b9b5@mail.gmail.com> <1243007226.3655.122.camel@hermans.wildebeest.org> <17c6771e0905221035y824ad5fla5a140831c4f0b92@mail.gmail.com> <1243015508.3655.123.camel@hermans.wildebeest.org> <1243063124.3579.19.camel@hermans.wildebeest.org> <1243205284.30320.15.camel@hermans.wildebeest.org> <17c6771e0905260135t59013f4dwc7ad25dea9520d28@mail.gmail.com> Message-ID: <17c6771e0905260137m15adc51j8c59f4c9494374d1@mail.gmail.com> 2009/5/26 Andrew John Hughes : > 2009/5/24 Mark Wielaard : >> On Sat, 2009-05-23 at 09:18 +0200, Mark Wielaard wrote: >>> On Fri, 2009-05-22 at 20:05 +0200, Mark Wielaard wrote: >>> > On Fri, 2009-05-22 at 18:35 +0100, Andrew John Hughes wrote: >>> > > Are there any negative side-effects to having this code applied all the time? >>> > >>> > No, the patches are designed so that without the property set the old >>> > pipeline is used. >>> >>> That said, if you do build with --disable-xrender then the font issue >>> doesn't occur. So something else is broken with the xrender support :( >>> I propose that we disable xrender support by default for now while we >>> figure out what else is borked. >> >> Found it. The X11 pipeline initialization is a bit tricky to follow. The >> xrender patches tried to clear that up at one point by making the >> initialization of the X11SurfaceData explicit (just like the >> XRSurfaceData). Although that code is nicer and cleaner, the new font >> code depends on the implicit static initializer calling order. That is a >> bit yucky, but lets play along and make sure that no explicit >> initialization is done: >> >> 2009-05-24 ?Mark Wielaard ? >> >> ? ? ? ?* patches/xrender/icedtea-001.patch: Initialize X11SurfaceData >> ? ? ? ?implicitly, not explicitly. >> >> Now both the standard X11 pipeline (default) and the XRender pipeline >> (-Dsun.java2d.xrender=True) work again against IcedTea7. >> >> Cheers, >> >> Mark >> > > Thanks for finding and fixing this. ?I narrowed it down to > X11SurfaceData on Friday (a diff between the non-xrender and xrender > versions showed this was the only place with a change that would make > any effect), but it was too late to find a fix. > -- > 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 > BTW, this demonstrates an issue with importing additional code where the upstream isn't tracking JDK7 updates... -- 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 From linuxhippy at gmail.com Thu May 28 04:44:16 2009 From: linuxhippy at gmail.com (Clemens Eisserer) Date: Thu, 28 May 2009 07:44:16 -0400 Subject: OpenJDK7 and the JavaOne preview (M3) release In-Reply-To: <17c6771e0905260308s14a81f8es18592db5294a6cda@mail.gmail.com> References: <17c6771e0905080454i51afcf0bn2d731ab657d58c9d@mail.gmail.com> <1243007226.3655.122.camel@hermans.wildebeest.org> <17c6771e0905221035y824ad5fla5a140831c4f0b92@mail.gmail.com> <1243015508.3655.123.camel@hermans.wildebeest.org> <1243063124.3579.19.camel@hermans.wildebeest.org> <1243205284.30320.15.camel@hermans.wildebeest.org> <17c6771e0905260135t59013f4dwc7ad25dea9520d28@mail.gmail.com> <17c6771e0905260137m15adc51j8c59f4c9494374d1@mail.gmail.com> <1243329380.20462.14.camel@hermans.wildebeest.org> <17c6771e0905260308s14a81f8es18592db5294a6cda@mail.gmail.com> Message-ID: <194f62550905280444i1bc28539j57d0f0800b70ee8b@mail.gmail.com> Hi, Sorry for causing problems and additional work, and thanks for fixing the issues. > Yes, but my point was we wouldn't have run into this issues if > upstream was keeping pace with jdk7 development. ?We have to be > careful which projects we import if we're going to end up maintaining > them over successive jdk7 releases, which happen on a weekly basis. > > Clemens, can you look into incorporating Mark's changes upstream? Yes, I'll do that. Soon I'll replace the current code with the rewrite version, hope it won't be a lot of work integrating it. Thanks, Clemens From gnu_andrew at member.fsf.org Thu May 28 11:04:57 2009 From: gnu_andrew at member.fsf.org (Andrew John Hughes) Date: Thu, 28 May 2009 19:04:57 +0100 Subject: OpenJDK7 and the JavaOne preview (M3) release In-Reply-To: <194f62550905280444i1bc28539j57d0f0800b70ee8b@mail.gmail.com> References: <17c6771e0905080454i51afcf0bn2d731ab657d58c9d@mail.gmail.com> <17c6771e0905221035y824ad5fla5a140831c4f0b92@mail.gmail.com> <1243015508.3655.123.camel@hermans.wildebeest.org> <1243063124.3579.19.camel@hermans.wildebeest.org> <1243205284.30320.15.camel@hermans.wildebeest.org> <17c6771e0905260135t59013f4dwc7ad25dea9520d28@mail.gmail.com> <17c6771e0905260137m15adc51j8c59f4c9494374d1@mail.gmail.com> <1243329380.20462.14.camel@hermans.wildebeest.org> <17c6771e0905260308s14a81f8es18592db5294a6cda@mail.gmail.com> <194f62550905280444i1bc28539j57d0f0800b70ee8b@mail.gmail.com> Message-ID: <17c6771e0905281104o138ad277j25d0f8184d8a2e14@mail.gmail.com> 2009/5/28 Clemens Eisserer : > Hi, > > Sorry for causing problems and additional work, and thanks for fixing > the issues. > >> Yes, but my point was we wouldn't have run into this issues if >> upstream was keeping pace with jdk7 development. ?We have to be >> careful which projects we import if we're going to end up maintaining >> them over successive jdk7 releases, which happen on a weekly basis. >> >> Clemens, can you look into incorporating Mark's changes upstream? > Yes, I'll do that. > > Soon I'll replace the current code with the rewrite version, hope it > won't be a lot of work integrating it. > > Thanks, Clemens > Thanks. Just holler if we can be of any help with hg and stuff. I think it'd be much easier for us working on IcedTea7 to just pull your changesets straight into our IcedTea forest than maintain a crazy bunch of patches. -- 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