On Monday, September 3, 2012 12:29:22 AM UTC+2, Ray Cromwell wrote:
>
> The version pings are actually pretty important in terms of showing 
> Google management continued interest in GWT as well. Being able to 
> have a rough estimate on how many active monthly users we have is a 
> good way to argue for continued investment.
>

Would you mind sharing the numbers? (at least with the Steering Committee)
Also, while I understand the usefulness of the pings (I'll rollback my 
changes in the gwt-maven-plugin for the DevMode mojo), that doesn't answer 
the question about the need for the JNI access to WinINet. It looks to me 
like recent versions of Java (at least starting with 1.5, for the 
Sun/Oracle JDK) use the WinINet settings by default (that's what I 
understand from Oracle's documentation). Glancing at the OpenJDK code, it 
looks like they use the settings from the registry, but do not use WinINet 
directly. Does that make a difference in practice? (maybe for corporate 
proxies that require authentication?)

Oh, and you'd probably want to add the ping to the SuperDevMode then?



 

>
> On Sat, Sep 1, 2012 at 10:41 PM, John A. Tamplin <j...@jaet.org<javascript:>> 
> wrote: 
> > On Sat, Sep 1, 2012 at 9:57 AM, Thomas Broyer 
> > <t.br...@gmail.com<javascript:>> 
> wrote: 
> >> 
> >> As you all know, the we decided to move to Maven as the build system 
> for 
> >> GWT. The first step is cleaning up dependencies. 
> >> 
> >> Digging into the code, I stumbled upon a bit of JNI: GWT checks for 
> >> updates every day, and on Windows it tries to use a bit of JNI to glue 
> it to 
> >> the IE COM object, and therefore make sure to use the WinINet proxy 
> >> settings. 
> >> For that to work, it requires the gwt-ll.dll inside the "install dir" 
> of 
> >> the GWT SDK. 
> >> 
> >> So, first, it means that it won't work if you use GWT from Maven (or 
> Ivy, 
> >> or Gradle, or anything that uses the artifacts from The Central 
> Repository 
> >> rather than an SDK bundle downloaded from code.google.com or as an 
> Eclipse 
> >> plugin), and will fallback to using "plain Java" in that case. 
> >> But I wonder if this bit of JNI is at all needed to being with, as Java 
> >> seems to be using (or able to use, at least) the WinINet proxy 
> settings: 
> >> http://www.java.com/en/download/help/proxy_setup.xml 
> >> 
> >> So, first question: is that bit of JNI really needed after all? 
> > 
> > 
> > IIRC, yes to get proper handling of proxy settings on Windows. 
> > 
> >> 
> >> Next question: is the version check needed? useful? 
> >> I recently changed the gwt-maven-plugin to unconditionally pass the 
> >> -XdisableUpdateCheck, as you generally don't want that check to be done 
> on 
> >> your CI server, and there are other means to check for updates (e.g. 
> "mvn 
> >> versions:display-dependency-updates" to stay in the Maven world, but 
> there's 
> >> also an RSS feed for the downloads, and Sonatype's Insight –e.g. 
> >> http://www.sonatype.com/Products/Insight-App-Health-Check–, and 
> probably 
> >> other means as well). 
> > 
> > Doing away with the update check would also mean losing the ability to 
> tell 
> > what versions are active from the pings. 
> > 
> > -- 
> > John A. Tamplin 
> > 
> > -- 
> > http://groups.google.com/group/Google-Web-Toolkit-Contributors 
>

-- 
http://groups.google.com/group/Google-Web-Toolkit-Contributors

Reply via email to