Implementation of jnlp://

Wed May 3 16:57:51 UTC 2017

On 05/01/2017 02:26 PM, Nicolas Roduit wrote:
> Yes, the ITW message management is much better than JWS.
> 

Surprise \o/

> I've tried to build from sources and it works as expected. There is an issue in javaws.desktop.in, 

thanx a lot!
> it should be in:
> MimeType=x-scheme-handler/jnlp;x-scheme-handler/jnlps

Thanx, pushed today.
> 
> Is possible to cross-compile from Linux the Windows dist without the plugins: Here are my parameters:
> ./configure --prefix=/home/nicolas/opt/IcedTea-web --with-jdk-home=/usr/lib/jvm/java-8-oracle 
> --disable-native-plugin
Half correct:

The above crosscompilation will give you correct netx.jar (whole javaws) and (completely useless) 
plugi's jsobject.jar but not correct lunchers, and corrupted plugin.jar
Plugin.jar maybe useless for you -  but it is powering javaws -html someHtmlPageUrl. This alow sto 
parse applet form html page. But its powers are limited ( no js powered applets,no dom 
communication) so it is just pure adapter for old applet games/tools.

Plugin jar's build on windows have hack, to replace xembeddedframe by wembededframe. See 
http://icedtea.classpath.org/hg/icedtea-web/file/0c05ca429063/Makefile.am#l541 . You see to have 
this croscompilable, you need windows jdk's rt.jar on classpath,
If you patch this , plugin jar will work fine for you after crosscompile (also with jsobject.jar)

Recommended way to build ITW on windows is cygwin. But if you workaround autotools (cross-autogen or 
how to say it) it should build for you on windows - this is just story I heard, never verified, so I 
can not confirm generated files or docs (many seds and greps) were correct.

> 
> An the plugins with cross-compilers like mingw-w64?

No. Native plugin, even if you crosscompile will not work.  Native java plugin is dead thing 
anyway... I know there is fork of itw - https://github.com/GITNE/icedtea-web - where the guy 
attempted to do correct windows npapi plugin, but it is really huge job. And there is already no 
target audience.

hth (and really thanx for inputs)
  j.
> 
> Best,
> 
> Nicolas
> 
> On 28. 04. 17 14:44, Jiri Vanek wrote:
>> On 04/25/2017 06:52 PM, Nicolas Roduit wrote:
>>> In my opinion this feature is major against the slow death of the Java client side although its 
>>> implementation is very simple. It offers the possibility to launch an application from many 
>>> contexts (browsers, mail client...).
>>
>> Agree.
>>
>>>
>>> For some advanced applications, we still believe that running a Java client application is the 
>>> good choice (easy deployment, performance, maintainability).
>>>
>>> However, since  JRE 8_121  I'm disappointed about the security behaviors of JWS. I've filled out 
>>> a bug at https://bugs.openjdk.java.net/browse/JDK-8175981 but the guys from Oracle consider as a 
>>> "security" feature.  As I don't have the permission to post a comment, I can only think that 
>>> Oracle is killing the dynamic behavior of JWS. Hopefully, this issue doesn't appear with the new 
>>> jnlp protocol.
>>
>>
>> I'm afraid yours hopes are invalid.  Those checks are here,and really are annoying.  Itw had 
>> adapted them on its own way, and so it have its own ways how to get rid of it (more levels or 
>> verbosity/security/checks, different mechanism on remembering of already clicked yes/no (based on 
>> regex, so you can get rid of a lot)
>> In 1.6 this was little bit glued together, in 1.7 this was rewritten. You can even redirect 
>> "YES\nYES\nYES..." to stdin of javaws to walk around it completely :))
>>
>>>
>>> I've seen that the version 1.7 will have a windows experimental version. It would be nice if a 
>>> future version of the IcedTea-web could install the mechanism for the jnlp handler (on Windows it 
>>> is very simple).
>>
>>
>> Indeed it will have windows support. In all cases, it will be heavily experimental as my windows 
>> abilities are very limited.  In all cases it will be some beggining, andhopefuly the windows port 
>> will grow in right direction.
>> The binary release of ITW for windows will not be capable of registering the protocol (no msi, 
>> just zip), however any redistributor, will be bale to pack this blob and wrap to msi or whatever 
>> suitable. (note, this may change, it is still some way to walk before release)
>>
>> Also src release of ITW dont have any url registrations, but to RPM for fedora I'm maintaing, I 
>> will add it (thank you for hints)
>>
>>
>> HtH
>>   J.
>>>
>>> Best,
>>>
>>> Nicolas
>>>
>>> On 25. 04. 17 15:19, Jiri Vanek wrote:
>>>> On 04/24/2017 05:39 PM, Nicolas Roduit wrote:
>>>>> Hi Jiri,
>>>>>
>>>>> You can try with the link "Non square pixels"  at 
>>>>
>>>> TY!
>>>>
>>>>> https://dcm4che.atlassian.net/wiki/display/WEA/DICOM+Samples
>>>>> Note: In some browsers, the jnlp handler doesn't work if you copy directly the link in the URL 
>>>>> bar 
>>>>> (jnlp://launcher-weasis.rhcloud.com/weasis-pacs-connector/viewer?studyUID=2.16.756.5.5.100.397184556.14391.1373576413.1508). 
>>>>
>>>>
>>>>
>>>>
>>>> ugh. indeed. the protocol registration appeared pretty tricky. I tried 5 browsers, and if it 
>>>> worked, it had sidekicks. Anyway, your example worked for me.
>>>>
>>>> The change is so simple that it may bebackported to 1.6. Still, with 1.7 on way(it was supposed 
>>>> to release in February, but got stuck on translations), I would rather wait. How critical is 
>>>> this for you?
>>>>
>>>>>
>>>>>
>>>>> To configure the jnlp handler on Linux, see the end of this page: 
>>>>> https://docs.oracle.com/javase/9/deploy/overview.htm
>>>>>
>>>>> For making other examples, you only need to replace the scheme of a jnlp link (http => jnlp).
>>>>
>>>> Yup. I had several self testing implementations, but nothing beats real life.
>>>>>
>>>>> Best,
>>>>>
>>>>> Nicolas
>>>>>
>>>>> On 24. 04. 17 16:47, Jiri Vanek wrote:
>>>>>> On 04/24/2017 10:49 AM, Jiri Vanek wrote:
>>>>>>> On 03/08/2017 08:18 PM, Nicolas Roduit wrote:
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> Is there any plan in IcedTea-Web to follow this implementation: 
>>>>>>>> https://bugs.openjdk.java.net/browse/JDK-8055464 ?
>>>>>>>>
>>>>>>>> It should help a lot for having a better integration of JWS with most of browsers.
>>>>>>>>
>>>>>>>> Best regards,
>>>>>>>>
>>>>>>>> Nicolas
>>>>>>>>
>>>>>>> Up to now, There were no intentions.
>>>>>>>
>>>>>>>
>>>>>>> Now I will take a look, but can not promise.
>>>>>>>
>>>>>>
>>>>>>
>>>>>> Today I made a patch for this. Are you able to test it? As I myself dont have any examples of 
>>>>>> this, it would be extremely useful!!
>>>>>>
>>>>>> Tahnx!
>>>>>>
>>>>>>  J.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>
>>>>
>>>
>>
>>
> 

-- 
Jiri Vanek
Senior QE engineer, OpenJDK QE lead, Mgr.
Red Hat Czech
jvanek at redhat.com    M: +420775390109