Unless you've also contributed a custom data type analzyer, I'm surprised it works at all on any of your pages. As for it working once, then an exception, then working, I'm not sure why it works the first time and then gives you an exception, but the reason it's working after the refresh is because the exception causes the existing page to be chucked from the pool, so when you refresh, you get a fresh instance of the page, and so whatever weird state you got in before is now fixed. Come to think of it, I saw a very similar issue when I upgraded an app of mine from 5.1.0.5 to 5.2.0-SNAPSHOT. It didn't have anything to do with BeanEditForm. But I did see behavior of every other render being fine, with the intervening renders throwing exceptions. In my case, it was caused by a page field that I wasn't properly cleaning up. There appears to be a small change in behavior with regard to non-persisted page fields in 5.2.0. Unfortunately, this was awhile ago, and I don't recall exactly what the issue was, but I can say that I remember after looking into it, I was amazed that the code was working at all in 5.1.0.5. That is, it appears that the new behavior in T5 is more "correct" (and in line with documented behavior) than in 5.1.0.5. In any event, as a starting point, it's worth checking on how you're storing your model parameter.
HTH, Robert On May 22, 2010, at 5/224:55 AM , Vangel V. Ajanovski wrote: > On 05/22/2010 10:16 AM, Jim O'Callaghan wrote: >> Is anyone using a version newer than 5.1.0.5 in production for their >> application? There are a few nice to have more recent additions that would >> be useful in our application but it would be interesting to hear first-hand >> experience of any stability issues. >> > I intend to run the latest version of our app build arround the latest > 5.2.0-SNAPSHOT in production from next week with a selected group of a > 200-300 users, but at this moment there is just one small bug that stops > me that I should try to fix, for which I am not sure if it is a problem > with our application or with Tapestry, and I don't yet have the time to > check. > > In BeanEditForm one should not "add" properties that already exist. But > when such properties are not atomic like string or int, for example a > property I have is of the class FacilityType, they are not visible. Of > course they are not because Tapestry does not know how to show them and > you have to create a "parameter" block to specifiy how to display them. > But this is not enough for some pages, and is enough for others in our > version with 5.2.0-snapshot. Under 5.1.0.5 it was all ok. In 5.2.0 i > have to "add" these properties alhough they already exist. But when I > "add" them, then Tapestry only runs OK for the first time I submit the > form. The next time I open the same page Tapestry complains that the > property already exists in the model, but after a refresh in the browser > everything seems normal and you can continue working. On the other hand > there are other pages where I am not forced to "add" the already > existing property and the "parameter" block is enough. This is just a > first report and I am not yet sure if this is a problem with Tapestry or > with how our pages are built. But since it's not a stopper bug, and you > can continue working after a refresh, it is down the priority for us at > this moment and I will not try to find it and state the nature til I am > sure and that will happen only after several days. > --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tapestry.apache.org For additional commands, e-mail: users-h...@tapestry.apache.org
