there is no SPI in the CDI spec which is not important for all users. Extensions are a core mechanism and needs to work.
We _must_ file a Weld issue in any case if we find an error. Maybe it might be our fault and they can help to clarify this... This is just like this funyn guy telling Matze that an 'average developer' will not code Extensions, thus -> low prio GF bug. This assumption is plain wrong and Glassfish folks already figured this. Because if an Extension developer cannot implement his Extension, then a 'casual developer' cannot use this extension... Who has stacktraces, info about this issue? Thomas or Gerhard one of you please file a Weld bug - big thanks! Of course if we find something in OWB which you think it should just work, then please file a jira issue against OWB too - txs! LieGrue, strub --- On Sat, 5/21/11, Gerhard Petracek <gerhard.petra...@gmail.com> wrote: > From: Gerhard Petracek <gerhard.petra...@gmail.com> > Subject: Re: AW: CODI @ViewAccessScoped not cleaned up? > To: "MyFaces Discussion" <users@myfaces.apache.org> > Date: Saturday, May 21, 2011, 10:26 AM > hi mark, > > it would be nice if we don't need workarounds for weld. > however, we already have several workarounds. without them > codi wouldn't > work in combination with weld. the same is true for > mojarra. and we can't > just ignore both reference implementations for > months/years. > > in this case it's a very special caching bug which came in > due to the > performance improvements in weld v1.1.1. usually you won't > see it - we just > have a very special constellation (since we (re-)use a spi > which was > designed for application developers). > > for sure we can file a weld issue. however, since it's a > quite special > constellation, they would give it a very low priority. > there are way more > weld bugs which are more important. > > regards, > gerhard > > http://www.irian.at > > Your JSF powerhouse - > JSF Consulting, Development and > Courses in English and German > > Professional Support for Apache MyFaces > > > > 2011/5/21 Mark Struberg <strub...@yahoo.de> > > > I'm not really conviced that we should implement > workarounds for such > > clearly broken weld issues. > > > > Thomas, did you already file a bug report in Weld > marking this as blocker? > > Really, we should at least link to them in our sources > (+jira) and asap get > > rid of this workaround in OUR source again. Otherwise > we will end up with > > unmaintainable CODI sources. > > > > LieGrue, > > strub > > > > --- On Sat, 5/21/11, Ing.Thomas Kernstock <t.kernst...@e-technologies.at> > > wrote: > > > > > From: Ing.Thomas Kernstock <t.kernst...@e-technologies.at> > > > Subject: AW: CODI @ViewAccessScoped not cleaned > up? > > > To: "'MyFaces Discussion'" <users@myfaces.apache.org> > > > Date: Saturday, May 21, 2011, 1:17 AM > > > that's what I expected :-) The > > > chances that weld causes the problem where > > > rather huge (as we discussed earlier). > > > have a good one > > > thomas > > > > > > > > > -----Ursprüngliche Nachricht----- > > > Von: Gerhard Petracek [mailto:gerhard.petra...@gmail.com] > > > > > > Gesendet: Freitag, 20. Mai 2011 21:08 > > > An: MyFaces Discussion > > > Betreff: Re: CODI @ViewAccessScoped not cleaned > up? > > > > > > hi thomas, > > > > > > this part of codi hasn't changed since the first > release > > > whereas weld > > > changed a lot. > > > > > > regards, > > > gerhard > > > > > > http://www.irian.at > > > > > > Your JSF powerhouse - > > > JSF Consulting, Development and > > > Courses in English and German > > > > > > Professional Support for Apache MyFaces > > > > > > > > > > > > 2011/5/20 Ing.Thomas Kernstock <t.kernst...@e-technologies.at> > > > > > > > Hi Gerhard, > > > > > > > > I Think you are right - this problem exists > either > > > since Codi 0.9.4 or > > > > Glassfish3.1(Weld 1.1.1). This explains why > I see > > > extra database > > > > updates (in my logfiles) of entities that > should have > > > been destroyed > > > > allready. On Glassfish 3.0.1 with CODI 0.9.3 > this > > > worked fine ! > > > > > > > > best regards > > > > Thomas > > > > > > > > ___________________________________ > > > > -----Ursprüngliche Nachricht----- > > > > Von: Gerhard Petracek [mailto:gerhard.petra...@gmail.com] > > > > Gesendet: Freitag, 20. Mai 2011 19:35 > > > > An: MyFaces Discussion > > > > Betreff: Re: CODI @ViewAccessScoped not > cleaned up? > > > > > > > > hi gernot, > > > > > > > > @PreDestroy will be called after rendering > the first > > > page which > > > > doesn't use the bean. > > > > it looks like a weld bug (i see the same > with weld > > > v1.1.1). > > > > i tested it with owb and it works as > expected. > > > > i'll have a look at it, if there is a > possible > > > workaround for weld. > > > > > > > > regards, > > > > gerhard > > > > > > > > http://www.irian.at > > > > > > > > Your JSF powerhouse - > > > > JSF Consulting, Development and > > > > Courses in English and German > > > > > > > > Professional Support for Apache MyFaces > > > > > > > > > > > > 2011/5/20 Gernot Pfingstl <gernot.pfing...@gmx.at> > > > > > > > > > Hi, > > > > > > > > > > I'm using Glassfish 3.1 and CODI 0.9.5 > > > > > > > > > > I've a bean A annotated with > @ViewAccessScoped. > > > > > > > > > > Bean A is used in page1.xhtml. If page1 > is called > > > from page1 the > > > > > state of the bean is preserved (as > expected). > > > > > > > > > > Then I navigate to page2.xhtml (which > does not > > > have any reference to > > > > > bean A), an then I navigate (from > page2) to > > > page3.xhtml (which also > > > > > does not have any reference to bean > A). > > > > > At this point I supposed bean A has > bean > > > invalidated/removed - but a > > > > > @PreDestroy is never calles > > > > > > > > > > Now I navigate (from page3) to > page1.xhtml - and > > > bean A has the > > > > > previous values and not an inital > state. > > > > > > > > > > Do I misunderstand the > @ViewAccessScoped or is > > > there something wrong? > > > > > > > > > > regards, > > > > > Gernot > > > > > > > > > > > > > > > > > > > > > >
