Because from the reliability point of view the native gwt solution is better than a third-party library and produces cleaner code while the styling drawback can easily be ameliorated, we have chosen it from the two and applied it for the gwt 1.4 upgrade branch.

We also didn't get any response from the gwt-tk mailing list regarding a future release. > Hi devs, > > We have planned the GWT 1.4 upgrade for the next XWatch milestone > at the > end of march. For this, we need to make a decision regarding the > web-gwt > dependencies (for the moment, the xwiki gwt dialogs rely on the gwt- > tk > library that does not yet have a release for gwt 1.4) so we must > choose > from: > - using another library to provide this functionality, particularly > the > MyGwt library > pros: nice looks and good API + the possibility of using the library > components in including projects. > cons: code changes required by the clean switch (updating all > involved > classes to use library API instead of GWT API), the lack of a maven > repository with all available MyGwt versions (only 0.5.0 rc). As > well, > while testing, I experienced a couple of rendering issues (caused, > seemingly by the use of the strict or xhtml DTD). > - using the native gwt modal dialogs introduced in 1.4: > pros: not styled (we totally control the styling process and can > build an > uniform look); widgets consistency, at least at this level. A big > advantage is that the web-gwt module will not depend on any other > library > anymore. > cons: not styled, the usual GWT API (which can be poor in some > situations); GWT also has some problems caused by the standard mode > interpretation in browsers (caused by the doctype declaration) but in > this situation the code seems to be stable. > > What do you think?