From jonathan.gibbons at oracle.com Fri Nov 6 16:47:49 2020 From: jonathan.gibbons at oracle.com (jonathan.gibbons at oracle.com) Date: Fri, 06 Nov 2020 16:47:49 +0000 Subject: hg: code-tools/jtreg: add final newline Message-ID: <202011061647.0A6GlnCv020764@aojmv0008.oracle.com> Changeset: 77b292ffdb73 Author: jjg Date: 2020-11-06 08:47 -0800 URL: https://hg.openjdk.java.net/code-tools/jtreg/rev/77b292ffdb73 add final newline ! src/share/classes/com/sun/javatest/regtest/agent/GetSystemProperty.java From jonathan.gibbons at oracle.com Fri Nov 6 19:23:18 2020 From: jonathan.gibbons at oracle.com (jonathan.gibbons at oracle.com) Date: Fri, 06 Nov 2020 19:23:18 +0000 Subject: hg: code-tools/jtreg: 2 new changesets Message-ID: <202011061923.0A6JNIPZ005047@aojmv0008.oracle.com> Changeset: 1ce981167447 Author: jjg Date: 2020-11-06 11:01 -0800 URL: https://hg.openjdk.java.net/code-tools/jtreg/rev/1ce981167447 Add support for Automatic-Module-Name in jar files ! src/share/classes/com/sun/javatest/regtest/exec/Action.java ! src/share/classes/com/sun/javatest/regtest/exec/RegressionScript.java Changeset: ce4ab2ed1cc5 Author: jjg Date: 2020-11-06 11:20 -0800 URL: https://hg.openjdk.java.net/code-tools/jtreg/rev/ce4ab2ed1cc5 JT Harness to 6.0-b14 ! make/build-support/jtharness/version-numbers ! make/build-support/version-numbers From jonathan.gibbons at oracle.com Mon Nov 9 17:05:28 2020 From: jonathan.gibbons at oracle.com (jonathan.gibbons at oracle.com) Date: Mon, 09 Nov 2020 17:05:28 +0000 Subject: hg: code-tools/jtreg: Remove Version.java from installed copty of testng.jar Message-ID: <202011091705.0A9H5SgF016373@aojmv0008.oracle.com> Changeset: 4297b4403673 Author: jjg Date: 2020-11-09 09:04 -0800 URL: https://hg.openjdk.java.net/code-tools/jtreg/rev/4297b4403673 Remove Version.java from installed copty of testng.jar ! make/jtreg.gmk From jonathan.gibbons at oracle.com Tue Nov 10 19:34:22 2020 From: jonathan.gibbons at oracle.com (jonathan.gibbons at oracle.com) Date: Tue, 10 Nov 2020 19:34:22 +0000 Subject: hg: code-tools/jtreg: ignore failure if no Version.java to delete Message-ID: <202011101934.0AAJYMFc000867@aojmv0008.oracle.com> Changeset: 4fa2589db463 Author: jjg Date: 2020-11-10 11:33 -0800 URL: https://hg.openjdk.java.net/code-tools/jtreg/rev/4fa2589db463 ignore failure if no Version.java to delete ! make/jtreg.gmk From jonathan.gibbons at oracle.com Tue Nov 10 23:55:45 2020 From: jonathan.gibbons at oracle.com (jonathan.gibbons at oracle.com) Date: Tue, 10 Nov 2020 23:55:45 +0000 Subject: hg: code-tools/jtreg: fix for tests that need to add non-default system modules Message-ID: <202011102355.0AANtj1X025927@aojmv0008.oracle.com> Changeset: 78e6684c6c33 Author: jjg Date: 2020-11-10 15:54 -0800 URL: https://hg.openjdk.java.net/code-tools/jtreg/rev/78e6684c6c33 fix for tests that need to add non-default system modules ! src/share/classes/com/sun/javatest/regtest/exec/MainAction.java From tiago.daitx at canonical.com Mon Nov 16 15:10:24 2020 From: tiago.daitx at canonical.com (Tiago Daitx) Date: Mon, 16 Nov 2020 12:10:24 -0300 Subject: Suggestion: make jtdiff more flexible on summary file naming Message-ID: Hi, Currently jtdift is very strict about the naming of a summary.txt file. In src/share/classes/com/sun/javatest/diff/ReportReader.java it uses file.getName().equals(SUMMARY_TXT) twice. Is there a particular reason for it being that strict about the file name? Assuming there is no need for it to be so strict, would it be acceptable to make it a tad more flexible? My first suggestion would be replacing the equals with endsWith(). And the second one would be adding some extra logic (system/java property, command line argument) that will either ignore the 'equals()' check or use something else instead. As for a use case, I don't have anything particularly strong on this. In some situations I find it easier to manage files with a particular naming convention than having each of them in their own separate directories. Best regards, Tiago -- Tiago St?rmer Daitx Software Engineer tiago.daitx at canonical.com PGP Key: 4096R/F5B213BE (hkp://keyserver.ubuntu.com) Fingerprint = 45D0 FE5A 8109 1E91 866E 8CA4 1931 8D5E F5B2 13BE From bakaliosdim at gmail.com Mon Nov 16 14:26:05 2020 From: bakaliosdim at gmail.com (=?UTF-8?B?zpTOt868zq7PhM+BzrfPgiDOnM+AzrHOus6xzrvOuc6/z4I=?=) Date: Mon, 16 Nov 2020 16:26:05 +0200 Subject: run specific jtreg tests under gdb Message-ID: Hello all, I am extending GraalVM to support ShenandoahGC and i am testing it using jtreg but some tests fail with segmentation fault.So i would like to use gdb with those specific tests but i can't find a way to do it.The command that jtreg has on its results forks a new process so that doesnt help.I run the tests using this command jtreg -v -jdk:/graal/graal-shenandoah-jdk -vmoptions:"-XX:+UnlockDiagnosticVMOptions -XX:+UseShenandoahGC -XX:-InlineNatives" /jdk/test/hotspot/jtreg/gc/shenandoah Thanks in advance, i hope i have given enough infos for you to help me From jonathan.gibbons at oracle.com Mon Nov 16 16:32:34 2020 From: jonathan.gibbons at oracle.com (Jonathan Gibbons) Date: Mon, 16 Nov 2020 08:32:34 -0800 Subject: run specific jtreg tests under gdb In-Reply-To: References: Message-ID: <1995010a-ebe6-73d2-3715-211f98377e96@oracle.com> Hello, In the .jtr file that is generated when you run a test, jtreg will generate a `rerun` section, containing a shell script to run the test standalone. This section is intended for use in cases like yours, where you want to modify how the test is run, such as to run it under a debugger. Note that in the section in the .jtr file, characters are escaped, and so characters like `\` appear as `\\`. You can either copy and paste the text into a text editor to unescape the text, or you can use the jtreg `-show` option, which can be used to extract content from a .jtr file and write it to the console, or piped into a file of your choosing. I hope that helps. -- Jon On 11/16/20 6:26 AM, ???????? ????????? wrote: > Hello all, > I am extending GraalVM to support ShenandoahGC and i am testing it using > jtreg but some tests fail with segmentation fault.So i would like to use > gdb with those specific tests but i can't find a way to do it.The command > that jtreg has on its results forks a new process so that doesnt help.I > run the tests using this command > jtreg -v -jdk:/graal/graal-shenandoah-jdk > -vmoptions:"-XX:+UnlockDiagnosticVMOptions -XX:+UseShenandoahGC > -XX:-InlineNatives" /jdk/test/hotspot/jtreg/gc/shenandoah > Thanks in advance, i hope i have given enough infos for you to help me From jonathan.gibbons at oracle.com Mon Nov 16 16:36:29 2020 From: jonathan.gibbons at oracle.com (Jonathan Gibbons) Date: Mon, 16 Nov 2020 08:36:29 -0800 Subject: Suggestion: make jtdiff more flexible on summary file naming In-Reply-To: References: Message-ID: <0bd38659-1337-3281-ac12-18692a919229@oracle.com> Tiago, It is not so much that jtreg is being strict as much as it is about looking at the argument and inferring the kind of content, based on the standard output generated by the main jtreg tool. It would probably be reasonable to provide a way to force the command-line arguments to be treated as summary files. -- Jon On 11/16/20 7:10 AM, Tiago Daitx wrote: > Hi, > > Currently jtdift is very strict about the naming of a summary.txt > file. In src/share/classes/com/sun/javatest/diff/ReportReader.java it > uses file.getName().equals(SUMMARY_TXT) twice. Is there a particular > reason for it being that strict about the file name? > > Assuming there is no need for it to be so strict, would it be > acceptable to make it a tad more flexible? My first suggestion would > be replacing the equals with endsWith(). And the second one would be > adding some extra logic (system/java property, command line argument) > that will either ignore the 'equals()' check or use something else > instead. > > As for a use case, I don't have anything particularly strong on this. > In some situations I find it easier to manage files with a particular > naming convention than having each of them in their own separate > directories. > > Best regards, > Tiago > From tiago.daitx at canonical.com Mon Nov 16 18:21:52 2020 From: tiago.daitx at canonical.com (Tiago Daitx) Date: Mon, 16 Nov 2020 15:21:52 -0300 Subject: Suggestion: make jtdiff more flexible on summary file naming In-Reply-To: <0bd38659-1337-3281-ac12-18692a919229@oracle.com> References: <0bd38659-1337-3281-ac12-18692a919229@oracle.com> Message-ID: Hi Jonathan, On Mon, Nov 16, 2020 at 1:38 PM Jonathan Gibbons wrote: > > Tiago, > > It is not so much that jtreg is being strict as much as it is about > looking at the argument and inferring the kind of content, based > on the standard output generated by the main jtreg tool. > > It would probably be reasonable to provide a way to force the > command-line arguments to be treated as summary files. Is there any preference on doing that as another command line argument (so that it does show in --help) or could this be done as simply as looking for a system or java property? My only concern about doing it as a property is how to make people aware that it exists. Do you have any suggestions where that should/could be mentioned? Somewhere in the documentation? In or close to the "File not recognized" error message? I am willing to submit patches for this proposal, just trying to figure out what is considered an acceptable solution. Regards, Tiago -- Tiago St?rmer Daitx Software Engineer tiago.daitx at canonical.com PGP Key: 4096R/F5B213BE (hkp://keyserver.ubuntu.com) Fingerprint = 45D0 FE5A 8109 1E91 866E 8CA4 1931 8D5E F5B2 13BE From jonathan.gibbons at oracle.com Mon Nov 16 18:47:54 2020 From: jonathan.gibbons at oracle.com (Jonathan Gibbons) Date: Mon, 16 Nov 2020 10:47:54 -0800 Subject: Suggestion: make jtdiff more flexible on summary file naming In-Reply-To: References: <0bd38659-1337-3281-ac12-18692a919229@oracle.com> Message-ID: <4ffdac30-d681-ebc5-b11f-e86db5925766@oracle.com> I think a command line option would be better, so that it shows up in command-line help.? I haven't looked at the code to see the best form of such as option, but I envisage two possibilities: 1. A boolean flag that declares that the file args are summary files. This allows flexibility for summary files only. 2. An option that declares the kind of the file args. This allows flexibility for any of the supported file types. -- Jon On 11/16/20 10:21 AM, Tiago Daitx wrote: > Hi Jonathan, > > On Mon, Nov 16, 2020 at 1:38 PM Jonathan Gibbons > wrote: >> Tiago, >> >> It is not so much that jtreg is being strict as much as it is about >> looking at the argument and inferring the kind of content, based >> on the standard output generated by the main jtreg tool. >> >> It would probably be reasonable to provide a way to force the >> command-line arguments to be treated as summary files. > Is there any preference on doing that as another command line argument > (so that it does show in --help) or could this be done as simply as > looking for a system or java property? > > My only concern about doing it as a property is how to make people > aware that it exists. Do you have any suggestions where that > should/could be mentioned? Somewhere in the documentation? In or close > to the "File not recognized" error message? > > I am willing to submit patches for this proposal, just trying to > figure out what is considered an acceptable solution. > > Regards, > Tiago > > -- > Tiago St?rmer Daitx > Software Engineer > tiago.daitx at canonical.com > > PGP Key: 4096R/F5B213BE (hkp://keyserver.ubuntu.com) > Fingerprint = 45D0 FE5A 8109 1E91 866E 8CA4 1931 8D5E F5B2 13BE