Re: x11 resource claim of graph applet in performance toolbar
I've closed every window in NB until the problem disappeared and the huge resource claim is caused by any of the treebased windows. The project group contains 43 POMs. Smaller project groups, cause much less resource claims. Other L like FlatLaf don't have this issue at all. I've looked around in de NB sources and could not find any Metal specific trickery. That leads me to believe the Metal L is causing this. So no bugs to fix in NetBeans. Thanks for the engagement, and thinking with me. Gr. Simon On 9/6/23 01:43, Laszlo Kishalmi wrote: Oh, NetBeans is the java-lang-Thread. It was taken on Ubuntu 23.04, on AMD GPU, using Wayland, and Java 17 as runtime. -- create firmware without programming https://www.bendyroad.com - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
Oh, NetBeans is the java-lang-Thread. It was taken on Ubuntu 23.04, on AMD GPU, using Wayland, and Java 17 as runtime. On 9/5/23 16:40, Laszlo Kishalmi wrote: Just adding my data to here: xrestop - Display: localhost Monitoring 20 clients. XErrors: 0 Pixmaps: 145296K total, Other: 40K total, All: 145337K total res-base Wins GCs Fnts Pxms Misc Pxm mem Other Total PID Identifier 160 44 7 1 13 87 40783K 4K 40787K 49409 Thunderbird 080 11 2 0 0 68 25615K 1K 25617K 5904 NoMachine - punchh, Ubuntu 22.04.3 LTS 200 1 1 0 5 13 24841K 360B 24841K ? 120 14 2 1 228 266 22774K 7K 22782K 32551 NoMachine - punchh, Ubuntu 22.04.3 LTS 260 6 106 1 101 128 9832K 6K 9839K ? java-lang-Thread On 9/5/23 14:50, Simon IJskes - QCG wrote: xrestop - Display: localhost Monitoring 47 clients. XErrors: 0 Pixmaps: 1407157K total, Other: 491K total, All: 1407648K total res-base Wins GCs Fnts Pxms Misc Pxm mem Other Total PID Identifier 180 9 3 0 27 84 594329K 2K 594332K 4070 hhh 360 208 21 1 34 431 485415K 16K 485431K 6593 3c0 9 6600 1 6560 6594 209589K 310K 209899K 64 h - Apache NetBeans IDE 18 (hhh used to anonymize) The GC increments with 30/sec. The Pxms increments with 10/sec. To reproduce: * Enable Metal L * Reset Toolbars When i add "My toolbar" with "Select in favorites" to the end of the toolbar, the growth of the X11 GC drops to 3 per second. When i force a garbage collection, the number of X11 GC and X11 Pxms drops. The large heap is indeed subjective. When i run with 3G heap, netbeans will render without artifacts, on my system. (artifacts like partially rendered texts in tree control) But this will depend on other programs active under X11 so it seems. Gr. Simon On 9/5/23 21:16, Antonio wrote: Hi, Please specify steps to reproduce this. "Large heap" is subjective. "When looking at xrestop you can see 'it' growing every second" is also ambiguous: what's exactly growing every second? I installed xrestop and switched to the Metal Look and Feel, and the xrestop line remains unchanged for NetBeans even if I resize the window. Cheers, Antonio On 5/9/23 19:54, Simon IJskes - QCG wrote: I cant be absolute, but the majority of the graphic contexts is returned with a garbage collection. The problem is, with a large heap, the X11 server resources are exhausted (so it seems) before a garbage collection occurs automatically. Also the Xserver memory consumption can easily go over 4Gb. Other L do not have this problem. It is only reproducable with Metal L - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
Just adding my data to here: xrestop - Display: localhost Monitoring 20 clients. XErrors: 0 Pixmaps: 145296K total, Other: 40K total, All: 145337K total res-base Wins GCs Fnts Pxms Misc Pxm mem Other Total PID Identifier 160 44 7 1 13 87 40783K 4K 40787K 49409 Thunderbird 080 11 2 0 0 68 25615K 1K 25617K 5904 NoMachine - punchh, Ubuntu 22.04.3 LTS 200 1 1 0 5 13 24841K 360B 24841K ? 120 14 2 1 228 266 22774K 7K 22782K 32551 NoMachine - punchh, Ubuntu 22.04.3 LTS 260 6 106 1 101 128 9832K 6K 9839K ? java-lang-Thread On 9/5/23 14:50, Simon IJskes - QCG wrote: xrestop - Display: localhost Monitoring 47 clients. XErrors: 0 Pixmaps: 1407157K total, Other: 491K total, All: 1407648K total res-base Wins GCs Fnts Pxms Misc Pxm mem Other Total PID Identifier 180 9 3 0 27 84 594329K 2K 594332K 4070 hhh 360 208 21 1 34 431 485415K 16K 485431K 6593 3c0 9 6600 1 6560 6594 209589K 310K 209899K 64 h - Apache NetBeans IDE 18 (hhh used to anonymize) The GC increments with 30/sec. The Pxms increments with 10/sec. To reproduce: * Enable Metal L * Reset Toolbars When i add "My toolbar" with "Select in favorites" to the end of the toolbar, the growth of the X11 GC drops to 3 per second. When i force a garbage collection, the number of X11 GC and X11 Pxms drops. The large heap is indeed subjective. When i run with 3G heap, netbeans will render without artifacts, on my system. (artifacts like partially rendered texts in tree control) But this will depend on other programs active under X11 so it seems. Gr. Simon On 9/5/23 21:16, Antonio wrote: Hi, Please specify steps to reproduce this. "Large heap" is subjective. "When looking at xrestop you can see 'it' growing every second" is also ambiguous: what's exactly growing every second? I installed xrestop and switched to the Metal Look and Feel, and the xrestop line remains unchanged for NetBeans even if I resize the window. Cheers, Antonio On 5/9/23 19:54, Simon IJskes - QCG wrote: I cant be absolute, but the majority of the graphic contexts is returned with a garbage collection. The problem is, with a large heap, the X11 server resources are exhausted (so it seems) before a garbage collection occurs automatically. Also the Xserver memory consumption can easily go over 4Gb. Other L do not have this problem. It is only reproducable with Metal L - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
A JFrame with a button on it creates 18 X11 graphics contexts according to xrestop. (tested on JDK 21 build34) Resizing it or pressing the button will very quickly produce more and will overtake the NetBeans instance which I have running in the background within a few seconds. After reading a bit about X11 contexts this doesn't surprise me, I am also not convinced that a high number is a problem. I had a GPU in a lab which did produce artifacts too, it wasn't a big problem since it was fairly rare, at some point the power supply got swapped out which fixed the problem entirely (and I almost sent the GPU back). Graphics artifacts can have many non-obvious hardware causes. I am not saying that this is the case here, but I am skeptical that it has something to do with x11 graphics contexts, at least until there is more evidence :) best regards, michael On 05.09.23 23:50, Simon IJskes - QCG wrote: xrestop - Display: localhost Monitoring 47 clients. XErrors: 0 Pixmaps: 1407157K total, Other: 491K total, All: 1407648K total res-base Wins GCs Fnts Pxms Misc Pxm mem Other Total PID Identifier 180 9 3 0 27 84 594329K 2K 594332K 4070 hhh 360 208 21 1 34 431 485415K 16K 485431K 6593 3c0 9 6600 1 6560 6594 209589K 310K 209899K 64 h - Apache NetBeans IDE 18 (hhh used to anonymize) The GC increments with 30/sec. The Pxms increments with 10/sec. To reproduce: * Enable Metal L * Reset Toolbars When i add "My toolbar" with "Select in favorites" to the end of the toolbar, the growth of the X11 GC drops to 3 per second. When i force a garbage collection, the number of X11 GC and X11 Pxms drops. The large heap is indeed subjective. When i run with 3G heap, netbeans will render without artifacts, on my system. (artifacts like partially rendered texts in tree control) But this will depend on other programs active under X11 so it seems. Gr. Simon On 9/5/23 21:16, Antonio wrote: Hi, Please specify steps to reproduce this. "Large heap" is subjective. "When looking at xrestop you can see 'it' growing every second" is also ambiguous: what's exactly growing every second? I installed xrestop and switched to the Metal Look and Feel, and the xrestop line remains unchanged for NetBeans even if I resize the window. Cheers, Antonio On 5/9/23 19:54, Simon IJskes - QCG wrote: I cant be absolute, but the majority of the graphic contexts is returned with a garbage collection. The problem is, with a large heap, the X11 server resources are exhausted (so it seems) before a garbage collection occurs automatically. Also the Xserver memory consumption can easily go over 4Gb. Other L do not have this problem. It is only reproducable with Metal L - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
i've written in a previous email 9/5/23 23:50, how to reproduce in netbeans, but i think you are looking for specifics of my X11 environment i guess? Gr. Simon On 9/5/23 23:39, Tamás Cservenák wrote: Am using (somewhat limited, but still) Netbeans on Linux + Wayland (Intel XE graphics), no problems so far. This is on laptops with Intel XE. My desktop has an NVidia GeForce GFX so I use X11 as Wayland on nVidia is meh. So, I can run NB on Wayland and X11. Linux on both are Fedora 38... Can you give me some pointers on how to check these? Thanks T -- create firmware without programming https://www.bendyroad.com - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
I'm unable to run with glx extensions. There i run so fast out of memory that my whole X11 crashes (in about 6 hours or so). See: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7626 Gr. Simon On 9/5/23 23:39, Tamás Cservenák wrote: Am using (somewhat limited, but still) Netbeans on Linux + Wayland (Intel XE graphics), no problems so far. This is on laptops with Intel XE. My desktop has an NVidia GeForce GFX so I use X11 as Wayland on nVidia is meh. So, I can run NB on Wayland and X11. Linux on both are Fedora 38... Can you give me some pointers on how to check these? Thanks T -- create firmware without programming https://www.bendyroad.com - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
xrestop - Display: localhost Monitoring 47 clients. XErrors: 0 Pixmaps: 1407157K total, Other: 491K total, All: 1407648K total res-base Wins GCs Fnts Pxms Misc Pxm mem Other Total PID Identifier 180 930 27 84 594329K 2K 594332K 4070 hhh 360 208 211 34 431 485415K 16K 485431K 6593 3c0 9 66001 6560 6594 209589K310K 209899K 64 h - Apache NetBeans IDE 18 (hhh used to anonymize) The GC increments with 30/sec. The Pxms increments with 10/sec. To reproduce: * Enable Metal L * Reset Toolbars When i add "My toolbar" with "Select in favorites" to the end of the toolbar, the growth of the X11 GC drops to 3 per second. When i force a garbage collection, the number of X11 GC and X11 Pxms drops. The large heap is indeed subjective. When i run with 3G heap, netbeans will render without artifacts, on my system. (artifacts like partially rendered texts in tree control) But this will depend on other programs active under X11 so it seems. Gr. Simon On 9/5/23 21:16, Antonio wrote: Hi, Please specify steps to reproduce this. "Large heap" is subjective. "When looking at xrestop you can see 'it' growing every second" is also ambiguous: what's exactly growing every second? I installed xrestop and switched to the Metal Look and Feel, and the xrestop line remains unchanged for NetBeans even if I resize the window. Cheers, Antonio On 5/9/23 19:54, Simon IJskes - QCG wrote: I cant be absolute, but the majority of the graphic contexts is returned with a garbage collection. The problem is, with a large heap, the X11 server resources are exhausted (so it seems) before a garbage collection occurs automatically. Also the Xserver memory consumption can easily go over 4Gb. Other L do not have this problem. It is only reproducable with Metal L - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists -- create firmware without programming https://www.bendyroad.com - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
On 05.09.23 23:29, Simon IJskes - QCG wrote: If i disable the memory graph, the resource claim growth is immediately reduced. The GC i was talking about, are X11 Graphic Contexts. ah I see :) I actually almost asked what you meant by "GC" since the numbers didn't make sense to me, but since clicking the memory graph literally triggers a garbage collection I thought you referred to those, since GC is a commonly used shortcut in java land. I don't know much about X11 graphics contexts so I am not sure if this is a problem. rendering artifacts could also indicate hardware issues. -mbien Gr. Simon - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
Am using (somewhat limited, but still) Netbeans on Linux + Wayland (Intel XE graphics), no problems so far. This is on laptops with Intel XE. My desktop has an NVidia GeForce GFX so I use X11 as Wayland on nVidia is meh. So, I can run NB on Wayland and X11. Linux on both are Fedora 38... Can you give me some pointers on how to check these? Thanks T On Tue, Sep 5, 2023 at 11:31 PM Simon IJskes - QCG wrote: > If i disable the memory graph, the resource claim growth is immediately > reduced. > > The GC i was talking about, are X11 Graphic Contexts. > > Gr. Simon > > On 9/5/23 23:16, Michael Bien wrote: > > hi Simon, > > > > > > On 04.09.23 11:00, Simon IJskes - QCG wrote: > > > >> It does take long before netbeans to rise to the top, and some time > >> later rendering artifacts start to occur. These are solved (so it > >> looks) by clicking on the graph forcing GC. > > > > This would indicate that UI resources are freed during object > > finalization (or equivalent). This wouldn't be a good strategy since it > > correlates two independent resources with each other (although I have > > seen this done in past in other projects). Its like only refueling your > > car if the water bottle is empty. > > > > I can't reproduce this right now and never saw something like this > > happening while using NetBeans, but if this is the case and we can > > figure out what it is, we should take a look at it. > > > > > >> > >> The memory graph in the performance toolbar claims a huge amount of > >> X11 resources. > >> > >> When looking at xrestop you can see it growing every second. > > > > why do you think that the memory graph is causing this? Maybe I am > > missing something but I read the other mails too and don't see the > > connection. You can turn it off, see below. > > > > > >> > >> During the search for this behaviour, i also noticed the applet cannot > >> be removed from the toolbar. Drag operation does not work, which it > >> does on the other buttons in the performance toolbar. > > > > right click on empty tool bar space -> uncheck "Performance" > > > > > >> The rate of GC increments depends on if the performance toolbar is the > >> last in the toolbars, when i add a custom toolbar at the end, it is > >> down to 3 per second. When not about 30+ per second. > > > > How did you measure this? 3 GCs per second in the young generation would > > be completely normal, 3 in the old generation would be very suspicious. > > > > Finalization does also only happen in certain GC phases and is > > implementation dependent (one reason why its deprecated for removal). > > Clicking the memory graph should run it if G1 is used. > > > > > >> I'm starting to think it is a broader problem. Maybe more a LAF thing, > >> if i 'hide' the window... > > > > NB does far more than just rendering text and buttons. It could scan > > your projects for changed files or run other periodic tasks. Those tasks > > are usually suspended when you minimize the main window - you might be > > measuring something here you don't intend to. > > > > > > make sure you test with the latest JDK update release to exclude runtime > > bugs (maybe its even a X11 bug), > > > > > > best regards, > > > > michael > > > > > > - > > To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org > > For additional commands, e-mail: dev-h...@netbeans.apache.org > > > > For further information about the NetBeans mailing lists, visit: > > https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists > > > > > > > > -- > create firmware without programming https://www.bendyroad.com > > > - > To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org > For additional commands, e-mail: dev-h...@netbeans.apache.org > > For further information about the NetBeans mailing lists, visit: > https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists > > > >
Re: x11 resource claim of graph applet in performance toolbar
If i disable the memory graph, the resource claim growth is immediately reduced. The GC i was talking about, are X11 Graphic Contexts. Gr. Simon On 9/5/23 23:16, Michael Bien wrote: hi Simon, On 04.09.23 11:00, Simon IJskes - QCG wrote: It does take long before netbeans to rise to the top, and some time later rendering artifacts start to occur. These are solved (so it looks) by clicking on the graph forcing GC. This would indicate that UI resources are freed during object finalization (or equivalent). This wouldn't be a good strategy since it correlates two independent resources with each other (although I have seen this done in past in other projects). Its like only refueling your car if the water bottle is empty. I can't reproduce this right now and never saw something like this happening while using NetBeans, but if this is the case and we can figure out what it is, we should take a look at it. The memory graph in the performance toolbar claims a huge amount of X11 resources. When looking at xrestop you can see it growing every second. why do you think that the memory graph is causing this? Maybe I am missing something but I read the other mails too and don't see the connection. You can turn it off, see below. During the search for this behaviour, i also noticed the applet cannot be removed from the toolbar. Drag operation does not work, which it does on the other buttons in the performance toolbar. right click on empty tool bar space -> uncheck "Performance" The rate of GC increments depends on if the performance toolbar is the last in the toolbars, when i add a custom toolbar at the end, it is down to 3 per second. When not about 30+ per second. How did you measure this? 3 GCs per second in the young generation would be completely normal, 3 in the old generation would be very suspicious. Finalization does also only happen in certain GC phases and is implementation dependent (one reason why its deprecated for removal). Clicking the memory graph should run it if G1 is used. I'm starting to think it is a broader problem. Maybe more a LAF thing, if i 'hide' the window... NB does far more than just rendering text and buttons. It could scan your projects for changed files or run other periodic tasks. Those tasks are usually suspended when you minimize the main window - you might be measuring something here you don't intend to. make sure you test with the latest JDK update release to exclude runtime bugs (maybe its even a X11 bug), best regards, michael - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists -- create firmware without programming https://www.bendyroad.com - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
hi Simon, On 04.09.23 11:00, Simon IJskes - QCG wrote: It does take long before netbeans to rise to the top, and some time later rendering artifacts start to occur. These are solved (so it looks) by clicking on the graph forcing GC. This would indicate that UI resources are freed during object finalization (or equivalent). This wouldn't be a good strategy since it correlates two independent resources with each other (although I have seen this done in past in other projects). Its like only refueling your car if the water bottle is empty. I can't reproduce this right now and never saw something like this happening while using NetBeans, but if this is the case and we can figure out what it is, we should take a look at it. The memory graph in the performance toolbar claims a huge amount of X11 resources. When looking at xrestop you can see it growing every second. why do you think that the memory graph is causing this? Maybe I am missing something but I read the other mails too and don't see the connection. You can turn it off, see below. During the search for this behaviour, i also noticed the applet cannot be removed from the toolbar. Drag operation does not work, which it does on the other buttons in the performance toolbar. right click on empty tool bar space -> uncheck "Performance" The rate of GC increments depends on if the performance toolbar is the last in the toolbars, when i add a custom toolbar at the end, it is down to 3 per second. When not about 30+ per second. How did you measure this? 3 GCs per second in the young generation would be completely normal, 3 in the old generation would be very suspicious. Finalization does also only happen in certain GC phases and is implementation dependent (one reason why its deprecated for removal). Clicking the memory graph should run it if G1 is used. I'm starting to think it is a broader problem. Maybe more a LAF thing, if i 'hide' the window... NB does far more than just rendering text and buttons. It could scan your projects for changed files or run other periodic tasks. Those tasks are usually suspended when you minimize the main window - you might be measuring something here you don't intend to. make sure you test with the latest JDK update release to exclude runtime bugs (maybe its even a X11 bug), best regards, michael - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
Hi, Please specify steps to reproduce this. "Large heap" is subjective. "When looking at xrestop you can see 'it' growing every second" is also ambiguous: what's exactly growing every second? I installed xrestop and switched to the Metal Look and Feel, and the xrestop line remains unchanged for NetBeans even if I resize the window. Cheers, Antonio On 5/9/23 19:54, Simon IJskes - QCG wrote: I cant be absolute, but the majority of the graphic contexts is returned with a garbage collection. The problem is, with a large heap, the X11 server resources are exhausted (so it seems) before a garbage collection occurs automatically. Also the Xserver memory consumption can easily go over 4Gb. Other L do not have this problem. It is only reproducable with Metal L - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
I cant be absolute, but the majority of the graphic contexts is returned with a garbage collection. The problem is, with a large heap, the X11 server resources are exhausted (so it seems) before a garbage collection occurs automatically. Also the Xserver memory consumption can easily go over 4Gb. Other L do not have this problem. It is only reproducable with Metal L Gr. Simon On 05-09-2023 12:26, Peter Hull wrote: It would be good if someone could try and reproduce this (I don't use netbeans on Linux) but it seems like it might not be Netbeans to blame? Also can you just clarify - are these X resources returned after a GC or is there some amount which is permanently leaked? On Mon, 4 Sept 2023 at 12:40, Simon IJskes - QCG wrote: The rate of GC increments depends on if the performance toolbar is the last in the toolbars, when i add a custom toolbar at the end, it is down to 3 per second. When not about 30+ per second. Using Metal Laf. Gr Simon On 9/4/23 13:19, Simon IJskes - QCG wrote: Thanks, had a look as well, and nothing looks like a forgotten dispose, in that component. I'm starting to think it is a broader problem. Maybe more a LAF thing, if i 'hide' the window (i dont know the proper term, the display manager shows a button with windowtitle in [] in the panel), the increment of the graphics contexts stops. Also it is probably dependend on the netbeans window content. After restarting netbeans it is down to 3 gc extra per second. Gr. Simon On 9/4/23 11:59, Peter Hull wrote: On Mon, 4 Sept 2023 at 10:01, Simon IJskes - QCG wrote: The memory graph in the performance toolbar claims a huge amount of X11 resources. To give an idea, it creates 30 X11 graphics contexts per second. For information - this is implemented in https://github.com/apache/netbeans/blob/0322fce7c3f5c9335f2bcb00e5751de81fdb7ffb/platform/openide.actions/src/org/openide/actions/HeapView.java#L62 I had a quick look and I can't see anything jumping out as being the problem, it's quite a simple pure Java class. The auto update is set to trigger every 1500 milliseconds so something else must be causing it to refresh at 30 times per second. - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists -- create firmware without programming https://www.bendyroad.com - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
It would be good if someone could try and reproduce this (I don't use netbeans on Linux) but it seems like it might not be Netbeans to blame? Also can you just clarify - are these X resources returned after a GC or is there some amount which is permanently leaked? On Mon, 4 Sept 2023 at 12:40, Simon IJskes - QCG wrote: > > The rate of GC increments depends on if the performance toolbar is the > last in the toolbars, when i add a custom toolbar at the end, it is down > to 3 per second. When not about 30+ per second. > > Using Metal Laf. > > Gr Simon > > On 9/4/23 13:19, Simon IJskes - QCG wrote: > > > > Thanks, > > > > had a look as well, and nothing looks like a forgotten dispose, in that > > component. > > > > I'm starting to think it is a broader problem. Maybe more a LAF thing, > > if i 'hide' the window (i dont know the proper term, the display manager > > shows a button with windowtitle in [] in the panel), the increment of > > the graphics contexts stops. Also it is probably dependend on the > > netbeans window content. After restarting netbeans it is down to 3 gc > > extra per second. > > > > Gr. Simon > > > > On 9/4/23 11:59, Peter Hull wrote: > >> On Mon, 4 Sept 2023 at 10:01, Simon IJskes - QCG wrote: > >>> The memory graph in the performance toolbar claims a huge amount of X11 > >>> resources. > >>> > >>> To give an idea, it creates 30 X11 graphics contexts per second. > >> > >> For information - this is implemented in > >> https://github.com/apache/netbeans/blob/0322fce7c3f5c9335f2bcb00e5751de81fdb7ffb/platform/openide.actions/src/org/openide/actions/HeapView.java#L62 > >> I had a quick look and I can't see anything jumping out as being the > >> problem, it's quite a simple pure Java class. > >> The auto update is set to trigger every 1500 milliseconds so something > >> else must be causing it to refresh at 30 times per second. > >> > >> - > >> To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org > >> For additional commands, e-mail: dev-h...@netbeans.apache.org > >> > >> For further information about the NetBeans mailing lists, visit: > >> https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists > >> > >> > >> > > > > -- > create firmware without programming https://www.bendyroad.com > > > - > To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org > For additional commands, e-mail: dev-h...@netbeans.apache.org > > For further information about the NetBeans mailing lists, visit: > https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists > > > - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
The rate of GC increments depends on if the performance toolbar is the last in the toolbars, when i add a custom toolbar at the end, it is down to 3 per second. When not about 30+ per second. Using Metal Laf. Gr Simon On 9/4/23 13:19, Simon IJskes - QCG wrote: Thanks, had a look as well, and nothing looks like a forgotten dispose, in that component. I'm starting to think it is a broader problem. Maybe more a LAF thing, if i 'hide' the window (i dont know the proper term, the display manager shows a button with windowtitle in [] in the panel), the increment of the graphics contexts stops. Also it is probably dependend on the netbeans window content. After restarting netbeans it is down to 3 gc extra per second. Gr. Simon On 9/4/23 11:59, Peter Hull wrote: On Mon, 4 Sept 2023 at 10:01, Simon IJskes - QCG wrote: The memory graph in the performance toolbar claims a huge amount of X11 resources. To give an idea, it creates 30 X11 graphics contexts per second. For information - this is implemented in https://github.com/apache/netbeans/blob/0322fce7c3f5c9335f2bcb00e5751de81fdb7ffb/platform/openide.actions/src/org/openide/actions/HeapView.java#L62 I had a quick look and I can't see anything jumping out as being the problem, it's quite a simple pure Java class. The auto update is set to trigger every 1500 milliseconds so something else must be causing it to refresh at 30 times per second. - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists -- create firmware without programming https://www.bendyroad.com - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
Thanks, had a look as well, and nothing looks like a forgotten dispose, in that component. I'm starting to think it is a broader problem. Maybe more a LAF thing, if i 'hide' the window (i dont know the proper term, the display manager shows a button with windowtitle in [] in the panel), the increment of the graphics contexts stops. Also it is probably dependend on the netbeans window content. After restarting netbeans it is down to 3 gc extra per second. Gr. Simon On 9/4/23 11:59, Peter Hull wrote: On Mon, 4 Sept 2023 at 10:01, Simon IJskes - QCG wrote: The memory graph in the performance toolbar claims a huge amount of X11 resources. To give an idea, it creates 30 X11 graphics contexts per second. For information - this is implemented in https://github.com/apache/netbeans/blob/0322fce7c3f5c9335f2bcb00e5751de81fdb7ffb/platform/openide.actions/src/org/openide/actions/HeapView.java#L62 I had a quick look and I can't see anything jumping out as being the problem, it's quite a simple pure Java class. The auto update is set to trigger every 1500 milliseconds so something else must be causing it to refresh at 30 times per second. - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists -- create firmware without programming https://www.bendyroad.com - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
On Mon, 4 Sept 2023 at 10:01, Simon IJskes - QCG wrote: > The memory graph in the performance toolbar claims a huge amount of X11 > resources. > > To give an idea, it creates 30 X11 graphics contexts per second. For information - this is implemented in https://github.com/apache/netbeans/blob/0322fce7c3f5c9335f2bcb00e5751de81fdb7ffb/platform/openide.actions/src/org/openide/actions/HeapView.java#L62 I had a quick look and I can't see anything jumping out as being the problem, it's quite a simple pure Java class. The auto update is set to trigger every 1500 milliseconds so something else must be causing it to refresh at 30 times per second. - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
Re: x11 resource claim of graph applet in performance toolbar
On 9/4/23 11:00, Simon IJskes - QCG wrote: The memory graph in the performance toolbar claims a huge amount of X11 resources. Some platform details: Netbeans 18, openjdk version "17.0.8.1" 2023-08-24 OpenJDK Runtime Environment (build 17.0.8.1+1-Ubuntu-0ubuntu122.04) OpenJDK 64-Bit Server VM (build 17.0.8.1+1-Ubuntu-0ubuntu122.04, mixed mode, sharing) X11: 2:21.1.4-2ubuntu1.7~22.04.1 Gr. Simon -- create firmware without programming https://www.bendyroad.com - To unsubscribe, e-mail: dev-unsubscr...@netbeans.apache.org For additional commands, e-mail: dev-h...@netbeans.apache.org For further information about the NetBeans mailing lists, visit: https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
