From bourges.laurent at gmail.com Thu Jun 6 18:41:52 2019 From: bourges.laurent at gmail.com (=?UTF-8?Q?Laurent_Bourg=C3=A8s?=) Date: Thu, 6 Jun 2019 20:41:52 +0200 Subject: IcedTeaWeb Roadmap In-Reply-To: References: <25ac8df3-d0f4-8782-85fb-fc7ec171e416@redhat.com> Message-ID: Hi, Sorry for being off, I need to focus on my work + family in june. Hopefully I will have more free time in july. Laurent Le mar. 21 mai 2019 ? 15:19, Jiri Vanek a ?crit : > hi! > > Now is sorry on my side. I fell sick, and am just slowly recovering. > The bats are not updated. They are waiting for your hand. > > Dont hesitate to criticise the variables + mvn + configure.sh + build.sh > approach/style. You are > currently primary customer of mine :) > > If you are going to work on the bats, then wait a while for a while, as > I'm going to publish one > more important PR today or tomorrow - the embedded mode foor 1.8 and 2.0 > (and a bit more for 2.0) > > In meantime, do not hesitate to experiment with : > JRE=/what/ever mvn clean install -Plaunchers > > > O:) > > J. > > > On 5/10/19 5:13 PM, Laurent Bourg?s wrote: > > Sorry for the noise, > > It seems you did the job already ! > > > > Did you manage updating windows launchers (bat) too ? > > > > Laurent > > > > Le ven. 10 mai 2019 ? 09:24, Laurent Bourg?s > > a ?crit : > > > > Dear Jiri, > > Sorry for my long silence, I was tired and stayed off. I will get > the latest code and have a look. > > > > I may have some time soon, but I have no clue how to proceed with > shell launcher 'build' or > > 'install'. > > > > I can second you or review such change, if it helps. > > > > Cheers, > > Laurent > > > > Le ven. 26 avr. 2019 ? 15:29, Jiri Vanek jvanek at redhat.com>> a ?crit : > > > > Hi Laurent, here is PR whcih enables rust and natice builds > again. > > > > https://github.com/AdoptOpenJDK/icedtea-web/pull/208 > > > > Will you be able to follow it with bats? > > > > Thank you in advance. If not, jsut let me know, and Iwil do my > best. > > > > J. > > On 4/18/19 1:22 PM, Jiri Vanek wrote: > > > Hello! > > > > > > On 4/18/19 10:24 AM, Laurent Bourg?s wrote: > > >> Hi, > > >> > > >> Since ITW 1.7 & 1.8 have been publicly released and the > icedtea-web repository migrated to > > >> AdoptOpenJDK github, I wonder what is the current roadmap for > IcedTeaWeb maintenance (1.8 > > branch) > > > > > > 1.8 should remain same. Backports will go in, and its separate > release will happen from > > time to time > > > (I would say currently is the timeto do so, but no one seems > to be interested) > > > > > > I should be responsible for the releases, and will publish > them on github, or still on > > > classpath.org . Also I wills send > announcement. > > > > > >> and its future 1.9... > > > > > > 2.0 :) > > >> > > >> Could you publicly explain what are the current plans on: > > >> - AdoptOpenJDK integration of ITW package into Public OpenJDK > binary builds ? > > > > > > I keep saying, that integration of ITW into any JDK is bad > idea, however it is easy to > > make ITW > > > embeddable in such way. Currently such a patch is not in > place. (but isactlually oneliner > > searching > > > first for its own jdk, or better third mode next to > DISTRIBUTION and BUNDLED - EMBEDDED) > > > > > > If anybody will do that, I'm not against. Also if any JDK > vendor will bundle ITW, its > > theirs choice. > > > > > >> - provide an itw installer on top of any OpenJDK install ? > > > > > > I wish to keep maintaining (with help of windows itw > community) ITW standalone installer > > and zips. > > > Similarly I wish to keep ITW as linux-distribution friendly. > > > > > >> - ITW code evolution in the karakun's OpenWebStart project ? > > > > > > I believe karakun have good intentions with ITW. They have > several full-timers on it, > > thus they can > > > afford to do breaking changes, which, if done in previous > model, could remain unfinished > > for long > > > time. Also they can help to fix long standing hard bugs like > iLo console or so. > > > > > > All changes are going through PR. Anybody can stop any PR at > any time. If you disagree > > with some > > > evolution, dont hesitate to rebuke. > > >> > > >> I noticed the current github head branch contains only java > code now. Where is the code > > providing > > >> launchers (shell & rust) ? > > > > > > Current github have both java and shell and rust codebase. > Java is built by maven. The > > shells and > > > rusts are currently "just there" but I'm already working on > integrate them with current > > maven build. > > > > > > Not much willchane at the end. Maven is generating jars. The > shell and rust laucnhers must > > put them > > > to classpath, in fboth standalone and distribution-like way. > I'm going to keep that > > alive. All > > > dsitros and community standalon builds are depnding on those > launchers. They are not going > > to be > > > left behind (unless somebody provides better ones) > > > > > >> How is made the packaging now ? > > > > > > > > > Currently there is nothing. On my machine PR is slowly rising > which will have script, > > which will > > > prepare distribution. I thougth it will be one afternoon work > but I overcalcualted quite a > > lot. I > > > hope to have it finished for rust in end of net week, shell > launchers to follow. > > > > > > > > > I'm going to fix the outstanding issues you did not like so > much - mainly only one libdir > > with all > > > jars for portable build, so quite a simplification for rust > and shell portable tarballs/msi. > > > > > > > > > > > > Hope that explains. Dont hesitate to shout if you disagree or > if you wish to do something > > differently. > > > > > > > > > J. > > > > > >> > > >> Cheers, > > >> Laurent Bourges > > > > > > > > > > > > -- > > Jiri Vanek > > Senior QE engineer, OpenJDK QE lead, Mgr. > > Red Hat Czech > > jvanek at redhat.com M: +420775390109 > > > > > -- > Jiri Vanek > Senior QE engineer, OpenJDK QE lead, Mgr. > Red Hat Czech > jvanek at redhat.com M: +420775390109 > -------------- next part -------------- An HTML attachment was scrubbed... URL: From vasvir2 at gmail.com Mon Jun 10 10:11:08 2019 From: vasvir2 at gmail.com (Vassilis Virvilis) Date: Mon, 10 Jun 2019 13:11:08 +0300 Subject: No subject Message-ID: Hi, I am new to the list. I researched all the questions below to the best of my abilities. If this is not the right place for such queries I apologize upfront and I would appreciate it if you could point me to the right direction. So the webstart is dead for us after java 11 upgrade and I have been tasked to research alternatives. The story so far. In linux we were already using IcedTea-Web. Nothing to be done here webstart wise. Task completed. Great. In Windows I used the AdoptOpenJdk msi installer and the IcedTea-Web msi installer. Worked great. Kudos to everybody and big thanks. In MacOSX now I was able to make it work with minimal patching on the portable build. Various questions were accumulated and I wasn't sure if I should spam github. I decided to ask here first. Let me know if you want me to create a github issue for any of this. 1) In MacOSX someone can use JRE=`/usr/libexec/java_home` to detect java home. Do you want me to try create a pull request along these lines like: [ -x /usr/libexec/java_home ] && JRE=`/usr/libexec/java_home`; 2) Is there any way to have a console with my application's messages and exceptions but without all the debug messages of the world? I wasn't able to find such a setting. 3) Having two consoles is annoying (great for debugging) but people are not used to it. I found how to turn off the one by specifying -Xnofork. Are there are any problems with that approach that I should be aware of? 4) I tried to give my webstart application another title but in the Application Menu just says Boot. That bit was working before. We may have apple deprecating features in accordance with Java 9+ changes at play here though. Looking around I found: a) My previous approach. It was working in the past but not currently. I will retry to make sure though. System.setProperty("apple.laf.useScreenMenuBar", "true"); System.setProperty("com.apple.mrj.application.apple.menu.about.name", "MyAppName"); b) -Xdock:name=MyAppName That one works but requires different javaws for different jnlp. Furthermore it doesn't play with -Xnofork obviously. c) java-vm-args attribute in the resource/java jnlp element looks like that is not honored, is it? Are there any plans to support java-vm-args in the feature? So what I would like to have: I) Config option to close/hide the first console without resorting to -Xnofork II) Config option to turn off debugging in the second console while keeping the application's messages such as the original netx implementation III) Automatically assign application name via -Xdock:name based on the jnlp title in MacOSX or via any other working way possible that I currently do not know. Will it help to break them in github issues or is it too much? Thanks again for reading thus far and sorry for the long mail. Regards Vassilis Virvilis From vasvir2 at gmail.com Mon Jun 10 10:19:21 2019 From: vasvir2 at gmail.com (Vassilis Virvilis) Date: Mon, 10 Jun 2019 13:19:21 +0300 Subject: IcedTeaWeb questions Message-ID: Sorry this time with subject Hi, I am new to the list. I researched all the questions below to the best of my abilities. If this is not the right place for such queries I apologize upfront and I would appreciate it if you could point me to the right direction. So the webstart is dead for us after java 11 upgrade and I have been tasked to research alternatives. The story so far. In linux we were already using IcedTea-Web. Nothing to be done here webstart wise. Task completed. Great. In Windows I used the AdoptOpenJdk msi installer and the IcedTea-Web msi installer. Worked great. Kudos to everybody and big thanks. In MacOSX now I was able to make it work with minimal patching on the portable build. Various questions were accumulated and I wasn't sure if I should spam github. I decided to ask here first. Let me know if you want me to create a github issue for any of this. 1) In MacOSX someone can use JRE=`/usr/libexec/java_home` to detect java home. Do you want me to try create a pull request along these lines like: [ -x /usr/libexec/java_home ] && JRE=`/usr/libexec/java_home`; 2) Is there any way to have a console with my application's messages and exceptions but without all the debug messages of the world? I wasn't able to find such a setting. 3) Having two consoles is annoying (great for debugging) but people are not used to it. I found how to turn off the one by specifying -Xnofork. Are there are any problems with that approach that I should be aware of? 4) I tried to give my webstart application another title but in the Application Menu just says Boot. That bit was working before. We may have apple deprecating features in accordance with Java 9+ changes at play here though. Looking around I found: a) My previous approach. It was working in the past but not currently. I will retry to make sure though. System.setProperty("apple.laf.useScreenMenuBar", "true"); System.setProperty("com.apple.mrj.application.apple.menu.about.name", "MyAppName"); b) -Xdock:name=MyAppName That one works but requires different javaws for different jnlp. Furthermore it doesn't play with -Xnofork obviously. c) java-vm-args attribute in the resource/java jnlp element looks like that is not honored, is it? Are there any plans to support java-vm-args in the feature? So what I would like to have: I) Config option to close/hide the first console without resorting to -Xnofork II) Config option to turn off debugging in the second console while keeping the application's messages such as the original netx implementation III) Automatically assign application name via -Xdock:name based on the jnlp title in MacOSX or via any other working way possible that I currently do not know. Will it help to break them in github issues or is it too much? Thanks again for reading thus far and sorry for the long mail. Regards Vassilis Virvilis -- Vassilis Virvilis From jvanek at redhat.com Wed Jun 12 15:31:47 2019 From: jvanek at redhat.com (Jiri Vanek) Date: Wed, 12 Jun 2019 17:31:47 +0200 Subject: IcedTeaWeb questions In-Reply-To: References: Message-ID: <719ea849-b587-6518-a926-02831428272a@redhat.com> On 6/10/19 12:19 PM, Vassilis Virvilis wrote: > Sorry this time with subject > > Hi, > > I am new to the list. I researched all the questions below to the > best of my abilities. If this is not the right place for such queries > I apologize upfront and I would appreciate it if you could point me > to the right direction. Thank you for reching ITW. I hope I will explain all your questions > > So the webstart is dead for us after java 11 upgrade and I have been > tasked to research alternatives. > > The story so far. > > In linux we were already using IcedTea-Web. Nothing to be done here > webstart wise. Task completed. Great. TY:) > > In Windows I used the AdoptOpenJdk msi installer and the IcedTea-Web > msi installer. Worked great. Kudos to everybody and big thanks. Hi! They come from here https://github.com/AdoptOpenJDK/IcedTea-Web/releases/tag/icedtea ? Interesting, I was not part of the release, which may not meter, but the release missed my Testing. LAt release Iwas testing was: http://mail.openjdk.java.net/pipermail/distro-pkg-dev/2019-March/041321.html This shoould not have any efect, I'm just ensuring what release you are on. > > In MacOSX now I was able to make it work with minimal patching on the > portable build. Various questions were accumulated and I wasn't sure > if I should spam github. I decided to ask here first. So shell launchers. Yah:( they are not future complete. > > Let me know if you want me to create a github issue for any of this. > > 1) In MacOSX someone can use > JRE=`/usr/libexec/java_home` > to detect java home. Do you want me to try create a pull request along > these lines like: > [ -x /usr/libexec/java_home ] && JRE=`/usr/libexec/java_home`; In native launchers, PATH is also scanned for possible JRE. Would this solve your issue? If yes, then I would go with PATH approach. also JAVA_HOME should bealready working for you, and also you can set JRE in $CONFIG_HOME/icedtea-web/deployment.properties. Would that work for you? If not, then your solution may go to play. > > 2) Is there any way to have a console with my application's messages > and exceptions but without all the debug messages of the world? I > wasn't able to find such a setting. Console have details button leading to details panel, where you can disable output of itw, and kepponly output of application. > > 3) Having two consoles is annoying (great for debugging) but people > are not used to it. I found how to turn off the one by specifying > -Xnofork. Are there are any problems with that approach that > I should be aware of? Each ITW have its console. And it have its purpose. The first console can actually have interesting info. Firs one belongs to first JVM process, which downloads JNLP, and founds eg XMX in it. Then it knows, it must spawn new vm, to apply thi setting. Second vm again downloads jnlp and the jars and run app. Soyou can see, xnofork is disabling some functionality. Nothing is preventing you from closing the first console, does it? If you wish to fix it, then likely new deployment property can be introduced, to close the console of forking VM . Looking forward for PR :) > > 4) I tried to give my webstart application another title but in the > Application Menu just says Boot. That bit was working before. We may > have apple deprecating features in accordance with Java 9+ changes > at play here though. Looking around I found: > > a) My previous approach. It was working in the past but not > currently. I will retry to make sure though. > > System.setProperty("apple.laf.useScreenMenuBar", "true"); > System.setProperty("com.apple.mrj.application.apple.menu.about.name", > "MyAppName"); > > b) -Xdock:name=MyAppName That one works but requires different javaws > for different jnlp. Furthermore it doesn't play with -Xnofork obviously. I had never before hear about those two. sorry. > > c) java-vm-args attribute in the resource/java jnlp element looks > like that is not honored, is it? Are there any plans to support > java-vm-args in the feature? The title belongs to jnlp's element title. Is it not honored too? speaking abot java-vm-args, not sure what you are reffering to. -J params are supported. > > So what I would like to have: > 0) find JRE on mac? ..should be answered > I) Config option to close/hide the first console without resorting > to -Xnofork I guess that was answered too. > > II) Config option to turn off debugging in the second console while > keeping the application's messages such as the original netx > implementation Also thsi was hopefully answered. ....Oh! now I realised - By console - you mean ITW console or terminal of your OS? For terminal, ITW should be really silent. Only in --verbose mode it shoudl spawn you with messages. What bothers you here? Maybe you have debugging proeprty enabled? As ITW supports headless support, so it supports pipes, if yuo see anything here in non debug mode, it is bug. Similarly, I had also wrongly answered your first question. This is your OS behaviour and os setup. I doubt there is anything we can do around. The second VM is again spawned via javaws.sh. And your os is opening terminal for each os. Using the javaws binary would solve this, but it means that somebody have to build native rust launchers on mac. Are yu voulenteer? it would be awesome :) > > III) Automatically assign application name via -Xdock:name based > on the jnlp title in MacOSX or via any other working way possible > that I currently do not know. This is first time I hear about -Xdoc. Jnlp's is not what you desire? You can pass jdk arguments to jvm via javaws -J-Xdoc, but I really do not see -Xdoc in openjdk's swithces. > > Will it help to break them in github issues or is it too much? Github is nice place to cooperate. We can continue the dsicussion a for a whil ehere, and once we agree what is bug, what feature, what is PR and what canbe workarounded then move to github. > > Thanks again for reading thus far and sorry for the long mail. > TYVM for the usage report! It is highly appreciated. J. From vasvir2 at gmail.com Thu Jun 13 11:34:59 2019 From: vasvir2 at gmail.com (Vassilis Virvilis) Date: Thu, 13 Jun 2019 14:34:59 +0300 Subject: IcedTeaWeb questions In-Reply-To: <719ea849-b587-6518-a926-02831428272a@redhat.com> References: <CAKbOjEy54j=Ww-0ircZ48uCH9CBHwZq1U3Ekn5Qf3wqUFDfLyQ@mail.gmail.com> <719ea849-b587-6518-a926-02831428272a@redhat.com> Message-ID: <CAKbOjExF07fS3c3Ap_HvuHGyY81sfYDp2sxbodXq7quq4gMUEg@mail.gmail.com> Thanks for the reply. It helps a lot. I will try to clarify inline. On Wed, Jun 12, 2019 at 6:31 PM Jiri Vanek <jvanek at redhat.com> wrote: > This shoould not have any efect, I'm just ensuring what release you are on. I went here: https://icedtea.classpath.org/wiki/IcedTea-Web#1.8 then here: https://mail.openjdk.java.net/pipermail/distro-pkg-dev/2019-March/041321.html (which is an e-mail from you I believe) In that e-mail provides links to the msi installer: http://icedtea.wildebeest.org/download/icedtea-web-binaries/1.8/windows/itw-installer.msi > In native launchers, PATH is also scanned for possible JRE. Would this solve your issue? > If yes, then I would go with PATH approach. > also JAVA_HOME should bealready working for you, and also you can set JRE in > $CONFIG_HOME/icedtea-web/deployment.properties. Would that work for you? > If not, then your solution may go to play. I don't know if you are familiar with MacOSX. Apple provides some wrapper for java and javaws that basically say: *Please install a JRE if you don't have one*. If you have multiple JRE installed then you can select between them, set the default etc via /usr/libexec/java_hom. In this scheme java is always in /usr/bin/java (the wrapper) and javaws is in /usr/bin/javaws. /usr/bin/javaws looks for the actual javaws in: /Library/Internet\ Plug-Ins/JavaAppletPlugin.plugin/Contents/Home/bin/javaws. JRE itself is located on /Library/Java/JavaVirtualMachines/adoptopenjdk-11.jre/Contents/Home So I looked a bit at launchers/rust-launcher/src/utils.rs : get_jdk_from_given_path_testable but I am not sure what it would in the MacOSX case. I can't tell without testing if such an approach would work. Thanks for $CONFIG_HOME/icedtea-web/deployment.properties. I hadn't realized that you set the JRE there. This implies that javaws runs with a JRE and you can configure another to run the application. Unfortunately it doesn't help you to pickup the right JRE to run javaws. > If you wish to fix it, then likely new deployment property can be introduced, to close the console > of forking VM . Looking forward for PR :) Thanks for confirming that there is no such setting. I will consider a PR when the itch grows to pain levels. > > I had never before hear about those two. sorry. Oh here there are: https://alvinalexander.com/java/java-application-name-mac-menu-bar-menubar-class-name > > > > c) java-vm-args attribute in the resource/java jnlp element looks > > like that is not honored, is it? Are there any plans to support > > java-vm-args in the feature? > > The title belongs to jnlp's element title. Is it not honored too? > speaking abot java-vm-args, not sure what you are reffering to. -J params are supported. -J needs to be specified in the command line when javaws is executed. java-vm-args is set on the JNLP, so javaws forks the java with these arguments. Somewhat risky I may add but it is there: https://docs.oracle.com/javase/8/docs/technotes/guides/javaws/developersguide/syntax.html > > ....Oh! now I realised - By console - you mean ITW console or terminal of your OS? > I mean ITW console with the advance panel and the details button. So that would require two PRs 1) Close ITW console on startup 2) Start with details off in the second console > Similarly, I had also wrongly answered your first question. This is your OS behaviour and os setup. > I doubt there is anything we can do around. The second VM is again spawned via javaws.sh. And your > os is opening terminal for each os. Using the javaws binary would solve this, but it means that > somebody have to build native rust launchers on mac. Are yu voulenteer? it would be awesome :) I think you answered correctly. I don't have two terminals or two shells. I just have two java consoles. > > This is first time I hear about -Xdoc. Jnlp's <title> is not what you desire? > You can pass jdk arguments to jvm via javaws -J-Xdoc, but I really do not see -Xdoc in openjdk's > swithces. Here it is: https://www.oracle.com/technetwork/articles/java/javatomac-140486.html The problem is that title does not set the application name in MaxOsx. Yes I can do it in the command line but I want to do it from the JNLP. That means to seth -Xdock:name when jnlp title is set. That would require a third PR. I really appreciate your input. Now I am understanding things much better. Thanks again. You have covered my questions 100% If you need any other clarification let me know. -- Vassilis Virvilis From jvanek at redhat.com Thu Jun 13 13:47:43 2019 From: jvanek at redhat.com (Jiri Vanek) Date: Thu, 13 Jun 2019 15:47:43 +0200 Subject: IcedTeaWeb questions In-Reply-To: <CAKbOjExF07fS3c3Ap_HvuHGyY81sfYDp2sxbodXq7quq4gMUEg@mail.gmail.com> References: <CAKbOjEy54j=Ww-0ircZ48uCH9CBHwZq1U3Ekn5Qf3wqUFDfLyQ@mail.gmail.com> <719ea849-b587-6518-a926-02831428272a@redhat.com> <CAKbOjExF07fS3c3Ap_HvuHGyY81sfYDp2sxbodXq7quq4gMUEg@mail.gmail.com> Message-ID: <e5deec51-ce9e-ff13-51a5-e2e2fd67dded@redhat.com> > I went here: https://icedtea.classpath.org/wiki/IcedTea-Web#1.8 > then here: https://mail.openjdk.java.net/pipermail/distro-pkg-dev/2019-March/041321.html > (which is an e-mail from you I believe) > In that e-mail provides links to the msi installer: > http://icedtea.wildebeest.org/download/icedtea-web-binaries/1.8/windows/itw-installer.msi ok. TY! > >> In native launchers, PATH is also scanned for possible JRE. Would this solve your issue? >> If yes, then I would go with PATH approach. >> also JAVA_HOME should bealready working for you, and also you can set JRE in >> $CONFIG_HOME/icedtea-web/deployment.properties. Would that work for you? >> If not, then your solution may go to play. > > I don't know if you are familiar with MacOSX. Apple provides some Not at all! I think it must be obvius:( Sorry and thanx for explanations. > wrapper for java and javaws that basically say: *Please install a JRE > if you don't have one*. If you have multiple JRE installed then you > can select between them, set the default etc via > /usr/libexec/java_hom. In this scheme java is always in /usr/bin/java > (the wrapper) and javaws is in /usr/bin/javaws. /usr/bin/javaws looks That mean it is on path I belive. So enhance shell launchers like to traverse path maybe good idea. > for the actual javaws in: /Library/Internet\ > Plug-Ins/JavaAppletPlugin.plugin/Contents/Home/bin/javaws. > JRE itself is located on > /Library/Java/JavaVirtualMachines/adoptopenjdk-11.jre/Contents/Home > > So I looked a bit at launchers/rust-launcher/src/utils.rs : > get_jdk_from_given_path_testable but I am not sure what it would in > the MacOSX case. I can't tell without testing if such an approach > would work. I owuld bet it will. Still worthy to add into shell lauchers. In shell launchers, the issue is that the waterfall rustlaunchers do when seeking for JRE, is producing unreadable code. > > Thanks for $CONFIG_HOME/icedtea-web/deployment.properties. I hadn't > realized that you set the JRE there. This implies that javaws runs > with a JRE and you can configure another to run the application. > Unfortunately it doesn't help you to pickup the right JRE to run > javaws. Ugh... That is exactly what it should do. If you have JRE specified here, it will be second most promoted after java_home. javaws and application are alwyas run in same jdk. Really. > >> If you wish to fix it, then likely new deployment property can be introduced, to close the console >> of forking VM . Looking forward for PR :) > > Thanks for confirming that there is no such setting. I will consider a > PR when the itch grows to pain levels. > >> >> I had never before hear about those two. sorry. > > Oh here there are: > https://alvinalexander.com/java/java-application-name-mac-menu-bar-menubar-class-name Oh I see. That can be fixed. If -Xdoc will be in jnlpfile, then it will - on mac - enforce forking of new jvm, which will have correct title. The disadvantage is, that it is mac only, and proper check for mac (java -Xdoc must not exit, as it does onlinux and win) wil be necessary. > >>> >>> c) java-vm-args attribute in the resource/java jnlp element looks >>> like that is not honored, is it? Are there any plans to support >>> java-vm-args in the feature? >> >> The title belongs to jnlp's element title. Is it not honored too? >> speaking abot java-vm-args, not sure what you are reffering to. -J params are supported. > > -J needs to be specified in the command line when javaws is executed. > java-vm-args is set on the JNLP, so javaws forks the java with these And java-vm-args are supported. But are based on enumeration. Xdoc is now misisng (especially because being mac specific) > arguments. Somewhat risky I may add but it is there: > https://docs.oracle.com/javase/8/docs/technotes/guides/javaws/developersguide/syntax.html > >> >> ....Oh! now I realised - By console - you mean ITW console or terminal of your OS? >> > > I mean ITW console with the advance panel and the details button. > > So that would require two PRs > > 1) Close ITW console on startup Yes, to enable hiding of first console after second vm is launched is good patch. > 2) Start with details off in the second console Not sure with this. It canbe easily added as new deployment property to specify what to see, but as defualt, I would alwyas vote for all. WDYT? > >> Similarly, I had also wrongly answered your first question. This is your OS behaviour and os setup. >> I doubt there is anything we can do around. The second VM is again spawned via javaws.sh. And your >> os is opening terminal for each os. Using the javaws binary would solve this, but it means that >> somebody have to build native rust launchers on mac. Are yu voulenteer? it would be awesome :) > > I think you answered correctly. I don't have two terminals or two > shells. I just have two java consoles. Thanx for clarification! > >> >> This is first time I hear about -Xdoc. Jnlp's <title> is not what you desire? >> You can pass jdk arguments to jvm via javaws -J-Xdoc, but I really do not see -Xdoc in openjdk's >> swithces. > > Here it is: https://www.oracle.com/technetwork/articles/java/javatomac-140486.html > The problem is that title does not set the application name in > MaxOsx. Yes I can do it in the command line but I want to do it from > the JNLP. That means to seth -Xdock:name when jnlp title is set. > That would require a third PR. Indded. As I wrote above, should be easily doable. Again thanx for confirmation. > > I really appreciate your input. Now I am understanding things much better. > > Thanks again. You have covered my questions 100% If you need any other > clarification let me know. > happy to help! Loking forward to review your PRs:) It is starting to be obvious that we need rust launchers on mac too.... J. -- Jiri Vanek Senior QE engineer, OpenJDK QE lead, Mgr. Red Hat Czech jvanek at redhat.com M: +420775390109 From vasvir2 at gmail.com Fri Jun 14 14:32:17 2019 From: vasvir2 at gmail.com (Vassilis Virvilis) Date: Fri, 14 Jun 2019 17:32:17 +0300 Subject: IcedTeaWeb questions In-Reply-To: <e5deec51-ce9e-ff13-51a5-e2e2fd67dded@redhat.com> References: <CAKbOjEy54j=Ww-0ircZ48uCH9CBHwZq1U3Ekn5Qf3wqUFDfLyQ@mail.gmail.com> <719ea849-b587-6518-a926-02831428272a@redhat.com> <CAKbOjExF07fS3c3Ap_HvuHGyY81sfYDp2sxbodXq7quq4gMUEg@mail.gmail.com> <e5deec51-ce9e-ff13-51a5-e2e2fd67dded@redhat.com> Message-ID: <CAKbOjEwS8tYaN3whzcA3fpsbATXicVpzd3G6oH=4XRGmMHkM9g@mail.gmail.com> On Thu, Jun 13, 2019 at 4:48 PM Jiri Vanek <jvanek at redhat.com> wrote: > > I don't know if you are familiar with MacOSX. Apple provides some > > Not at all! I think it must be obvius:( Sorry and thanx for explanations. Me neither. This is what I figured the last week trying to support some users. > Ugh... That is exactly what it should do. If you have JRE specified here, it will be second most > promoted after java_home. javaws and application are alwyas run in same jdk. Really. Yes I can see it now. deployment.properties is read by the shell. Nice. > And java-vm-args are supported. But are based on enumeration. Xdoc is now misisng (especially > because being mac specific) In my tests I wasn't able to pass any args to the application jvm via jnlp. I don't know what 'based on enumeration' means though. > Not sure with this. It canbe easily added as new deployment property to specify what to see, but as > defualt, I would alwyas vote for all. WDYT? I would think to have it open only for developers or users that are debugging jnlp loading issues. For normal users with working setups I would prefer it to be closed. The original netx implementation had only one java console not two and this surprises and annoys some users. Anyway if there was a config option to auto close after jvm startup that would be great and more than enough. Thank you very much for all the insights and the welcoming atmosphere. I will try to play with application name setups and the rust launcher and maybe I can provide feedback in PR form. Vassilis Vassilis From jvanek at redhat.com Fri Jun 14 14:48:19 2019 From: jvanek at redhat.com (Jiri Vanek) Date: Fri, 14 Jun 2019 16:48:19 +0200 Subject: IcedTeaWeb questions In-Reply-To: <CAKbOjEwS8tYaN3whzcA3fpsbATXicVpzd3G6oH=4XRGmMHkM9g@mail.gmail.com> References: <CAKbOjEy54j=Ww-0ircZ48uCH9CBHwZq1U3Ekn5Qf3wqUFDfLyQ@mail.gmail.com> <719ea849-b587-6518-a926-02831428272a@redhat.com> <CAKbOjExF07fS3c3Ap_HvuHGyY81sfYDp2sxbodXq7quq4gMUEg@mail.gmail.com> <e5deec51-ce9e-ff13-51a5-e2e2fd67dded@redhat.com> <CAKbOjEwS8tYaN3whzcA3fpsbATXicVpzd3G6oH=4XRGmMHkM9g@mail.gmail.com> Message-ID: <47ffb387-9310-2dea-2f60-d6881f9af588@redhat.com> On 6/14/19 4:32 PM, Vassilis Virvilis wrote: > On Thu, Jun 13, 2019 at 4:48 PM Jiri Vanek <jvanek at redhat.com> wrote: > >>> I don't know if you are familiar with MacOSX. Apple provides some >> >> Not at all! I think it must be obvius:( Sorry and thanx for explanations. > > Me neither. This is what I figured the last week trying to support some users. > >> Ugh... That is exactly what it should do. If you have JRE specified here, it will be second most >> promoted after java_home. javaws and application are alwyas run in same jdk. Really. > > Yes I can see it now. deployment.properties is read by the shell. Nice. And by rust ones. > >> And java-vm-args are supported. But are based on enumeration. Xdoc is now misisng (especially >> because being mac specific) > > In my tests I wasn't able to pass any args to the application jvm via > jnlp. I don't know what 'based on enumeration' means though. The netx is searching for xmx, xms and few others. If it founds them, then it spawn new JVM. Xdoc is another one who will need to be added here. With exception of being mac only. > >> Not sure with this. It canbe easily added as new deployment property to specify what to see, but as >> defualt, I would alwyas vote for all. WDYT? > > I would think to have it open only for developers or users that are > debugging jnlp loading issues. For normal users with working setups I > would prefer it to be closed. The original netx implementation had > only one java console not two and this surprises and annoys some > users. Anyway if there was a config option to auto close after jvm > startup that would be great and more than enough. I moreover agree. Still the default is enabled+hidden. So once you show it by purpose, it assumes you are already developer, and thus you wish to see all the issues. and that means issues from both runs. It could be read via some socket or pipe (as it was from plugin) but that can leak platform independence with unwanted overhead. > > Thank you very much for all the insights and the welcoming atmosphere. > I will try to play with application name setups and the rust launcher > and maybe I can provide feedback in PR form. TYVM for this effort! Corssing fingers! J.