Community
Participate
Working Groups
I have tried using both RAC and IAC and the manual test client does not come up. It prints an exception like the following: org.eclipse.swt.SWTError: No more handles [gtk_init_check() failed] at org.eclipse.swt.SWT.error(SWT.java:2968) at org.eclipse.swt.widgets.Display.createDisplay(Display.java:757) at org.eclipse.swt.widgets.Display.create(Display.java:746) at org.eclipse.swt.graphics.Device.<init>(Device.java:141) at org.eclipse.swt.widgets.Display.<init>(Display.java:429) at org.eclipse.swt.widgets.Display.<init>(Display.java:420) at org.eclipse.tptp.test.manual.runner.ui.internal.ManualTestView.<init>(ManualTestView.java:125) at org.eclipse.tptp.test.manual.runner.core.internal.ManualTestRunner.main(ManualTestRunner.java:194)
Configuration: RHEL v4 with the IBM JRE v1.4.2.
SWT requires the following libraries with the specified versions or later: -GTK 2.2.1 -ATK 1.2.0 -glib 2.2.1 -Pango 1.2.1 -Freetype 2.1.3 The test machine is at gtk2-2.4.13-12 and an uplevel of all other packages.
This line of code in the Manual Test View simply creates a new display: Display display = new Display(); Confirming no other SWT dependencies required for Linux.
Created attachment 29620 [details] Test class. SWT libraries shipped in the Agent Controller or the workbench are required to run this test class.
(In reply to comment #4) Test class does not produce the symptoms on the test machine.
See https://bugs.eclipse.org/bugs/show_bug.cgi?id=106396#c7
[Eclipse Project Release Notes, Release 3.1.1] gtk_init_check and X11 socket failure when using the IBM 1.4.2 JRE (GTK only) Under RHEL 3.1 with the IBM 1.4.2 JRE and a large number of plugins, Eclipse may fail to launch with an exception from gtk_init_check along with this error: _X11TransSocketOpen: socket() failed for local _X11TransSocketOpenCOTSClient: Unable to open socket for local A workaround is to set the environment variable JAVA_HIGH_ZIPFDS to a value of 500 before starting Eclipse. (bug 106396)
Potential for a readme entry. Originator is testing this potential workaround. Potential readme entry: For Linux, you must create an environment variable called JAVA_HIGH_ZIPFDS and set its value to a high number, such as 500, to tell the IBM JRE to create that many file descriptors in a separate area of memory for later use.
Highly desirable and planned for this release, but not stop ship Prevents function from being used, but a work around is possible
This is only happening on the originator test machine. As such, it appears to be a configuration issue or SWT defect. Still investigating....
Two different test machines, each with the required SWT dependencies, produce differing test results: Test machine 1 (Originator's machine - purple): OS: Linux purple 2.6.9-5.ELsmp #1 SMP Wed Jan 5 19:30:39 EST 2005 i686 i686 i386 GNU/Linux GTK: gtk2-2.4.13-12 ATK: atk-1.8.0-2 glib: glib-1.2.10-15/glib2-2.4.7-1 Pango: pango-1.6.0-7 Freetype: freetype-2.1.9-1 Test machine 2 (Tester's machine - dungeon): OS: Linux dungeon 2.6.9-11.ELsmp #1 Fri May 20 18:17:57 EST 2005 i686 i686 i386 GNU/Linux GTK: gtk2-2.4.13-14 ATK: atk-1.8.0-2 glib: glib-1.2.10-15/glib2-2.4.7-1 Pango: pango-1.6.0-9 Freetype: freetype-2.1.9-1
(In reply to comment #11) NOTE: The machine producing the symptoms is a multiprocessor machine with a different Linux kernel.
(In reply to comment #8) Potential workaround did not work.
Opened SWT bug report (https://bugs.eclipse.org/bugs/show_bug.cgi?id=115822). Since there is no solution to this problem and the problem is with a TPTP dependency, deferring to the TPTP v4.2 release.
Can you run Eclipse/TPTP or Manual Test Client directly (not through IAC or AC) the dual CPU machine?
(In reply to comment #15) Eclipse/TPTP works independent of IAC or Agent Controller. However, the Manual Test View does not work independent of IAC or Agent Controller (org.eclipse.swt.SWTError: No more handles [gtk_init_check() failed]).
My question was if you were able to run Eclipse or TPTP on that machine or you get the same error? You should be able to run Manual Test client independently, please let me know if you need help to do that.
(In reply to comment #17) See Comment #16.
Is it yes or not? sorry but I cannot infer that anwser from Comment #16 :-)
(In reply to comment #19) The Manual Test View does not work independent of IAC or Agent Controller (org.eclipse.swt.SWTError: No more handles [gtk_init_check() failed]).
(In reply to comment #20) This defect occurs when running the Manual Test View on SMP Linux with the following scenarios: 1) Manual Test View executed by the IAC. 2) Manual Test View executed by the Agent Controller. 3) Manual Test View executed stand-alone (see defect #115822).
Targeting to the next iteration since this defect is blocked by SWT defect 115822.
No change in state for dependent SWT defect. As such, cannot contain in TPTP V4.2.0 (i2).
Update the manual test runner to return an error message to the launching client workbench instructing the user of the issue and any possible solution.
Created attachment 38323 [details] Test class with a SWT Browser.
(In reply to comment #25) > Created an attachment (id=38323) [edit] > Test class with a SWT Browser. > Hi ! I noticed the same problem with non-SMP kernel in FreeBSD. I am FreeBSD user, and I have SWT library (3.1) installed on two computers. One is working ok, but other is not. Only difference betwen computers is that one is connected to internal network (that one is working), and the other one is connected to internet directly (through other internet line). I have tried to "fix" all 4 points and I have somewhat succeeded. 1. The DISPLAY environment variable is unset I have set it to localhost:0 2. The host/port information is wrong I have added my hostname to /etc/hosts and X doesn't complain anymore 3. The X server is down IS NOT. 4. The socket to the X server is on too high of a file descriptor Mhm. i can't do anything with this. I tried to set that JAVA_HIGH... variable, but it doesn't work on FreeBSD I am using FreeBSD 6.1 (RC1 on first computer and RELEASE on second), Java version 1.5.0_06 (I tried with both versions, with home build and that that is available as download now). Application producing this error is Azureus. Computer is AMD 2.2 Gb (Single core 32-bit), 512 Mb of memory. I have also tried with this sample in this forum, and I get the same problem. Andy
*** Bug 138596 has been marked as a duplicate of this bug. ***
This bug is blocking bidi testing activities, we are expecting a fix for this bug before eclipse 3.2 is shipped in order to verify and close these bugs.
(In reply to comment #26) Andy, thanks for your comments. This defect (SMP Linux IA32) has been verified as working as designed using TPTP V4.2.0. Please open a SWT defect for your issue with the sample code from this defect (https://bugs.eclipse.org/bugs/attachment.cgi?id=38323).
Using the TPTP-4.2.0-200605290100 workbench and Agent Controller (SWT V3.2 RC6) on the SMP test machine (Linux purple 2.6.9-5.ELsmp #1 SMP Wed Jan 5 19:30:39 EST 2005 i686 i686 i386 GNU/Linux) and IBM JRE V1.5.0 (SR 2), the Manual Test View is working as designed after verifying the following: 1. The DISPLAY environment variable is set. 2. The host/port information is correct (e.g. added to /etc/hosts). 3. The X server is up. Opened defect https://bugs.eclipse.org/bugs/show_bug.cgi?id=144777 to address the information dialog (https://bugs.eclipse.org/bugs/show_bug.cgi?id=115648#c24).
ACTION: Please verify/close this defect.
Closing.
I ran into this problem as well. I could not solve until I noticed a peculiarity in the environment tab of the run configuration dialog: normally, the radio buttons "Append environment to native environment" and "Replace native environment with the specified environment" are available, but in this case they are both disabled and grayed out. I suppose this means the environment is replaced rather than appended, which is most certainly not what I want. In any case, I worked around this by manually configuring the DISPLAY and HOME environment variables. I looked at /proc/$pid/environ to confirm these variables did not exist in the environment. I suggest that these radio buttons should not be disabled, and append should be the default, as it is when not running under the profiler.
(In reply to comment #33) Thanks Shane for this information. When the user launches a TPTP Manual Test using the Test Launch Configuration (opposed to a SWT Applciat5ion Launch Configuration), the Environment tab is not exposed since the Manual Test View is launched by the Integrated/Agent Controller using its configuration. Ultimately, the DISPLAY and HOME environment variables must be configured on the machine (as well as Mozilla) for the Manual Test View to successfully launch.
*** Bug 201850 has been marked as a duplicate of this bug. ***
