Hi guys, Yeah, I was thinking along the same lines. If we had a FilterLink attribute (continuing the analogy of a link in the chain :P). In response to your earlier post Gert, I think it would be better if we hold off on filterset's and strive for filterchains, even if it doesn't make the 0.8.4 release.
A filterlink discover mechanism like the task discovery mechanism would be awesome I think - its one of the nice things about NAnt I think. Will the replace task still be useful for 0.8.4? ---------------------------------------- - Mitch Denny - [EMAIL PROTECTED] - http://www.monash.net - +61 (414) 610141 - > -----Original Message----- > From: Ian MacLean [mailto:[EMAIL PROTECTED] > Sent: Wednesday, October 08, 2003 12:52 AM > To: Gert Driesen > Cc: Mitch Denny; [EMAIL PROTECTED] > Subject: filterchains/sets other was: [Fwd: Ready to tackle > next release] > > I think I need to read up on filterchains a bit more. Is the > functionality of filterchains a superset of filtersets ? > > It should be possible to use the same type discovery > mechanism we now use for Tasks and referencable elements to > discover filterreader classes in a given assembly. Or > somthing similar using attributes. > > Ian > > Gert Driesen wrote: > > >Mitch, > > > >The filterset implementation currently does not allow for > other types > >of filters to be set, that's what the filterchain support > was intended for. > > > >I'm not sure on how to proceed with this, we could rework > the filterset > >implementation (which, right now, is more or less a port of the Ant > >filterset) to allow custom filtersets but this would > actually mean that > >the filterset implementation is moving toward a filterchain > >implementation (which is not bad ofcourse) .... > > > >It would change the build layout of the filtersets considerately > >ofcourse ... It would also mean that we would not be able to use the > >Element initialization code that we use for all other types, as this > >would mean that we'd need have a separate property for each possible > >type of filterreader in the filterchain, and this would > prevent further > >extensibility by third parties ... > > > >Ant support two notations for built-in filterreaders : > > > > <filterchain> > > <filterreader > classname="org.apache.tools.ant.filters.ReplaceTokens"> > > <param type="token" name="DATE" value="${TODAY}"/> > > </filterreader> > > </filterchain> > > > >and > > > > <filterchain> > > <replacetokens> > > <token key="DATE" value="${TODAY}"/> > > </replacetokens> > > </filterchain> > > > >For third party filterreaders only the first notation is supported, > >ofcourse. > > > >If we decide to go for the filterchain implementation instead of the > >fitlerset, I might some need help as I don't think I'll be > having much > >time (at least not enough, I think) in the next week or weeks ... > > > >I certainly would like to avoid having to support both > filtersets and > >filterchains, like Ant is forced to do because of backward > >compatibility, so we should decide on this as soon as possible ... > > > >I'd definitely prefer going for the filterchain support, > even if that > >means that I've wasted my effort on the filterset > implementation, but > >as I said, I could use a hand ... > > > >Gert > > > > > >----- Original Message ----- > >From: "Mitch Denny" <[EMAIL PROTECTED]> > >To: "Gert Driesen" <[EMAIL PROTECTED]>; "Ian MacLean" > ><[EMAIL PROTECTED]> > >Cc: <[EMAIL PROTECTED]> > >Sent: Tuesday, October 07, 2003 11:58 AM > >Subject: RE: [nant-dev] [Fwd: Ready to tackle next release] > > > > > >Cool, so should I bother with the replace task? One of the > tasks modes > >is NAnt and Regex (as well as simple text substitution). The > begin and > >end token attributes wouldn't make sense in that case and it would > >(possibly) be better to do something like: > > > ><filtersets> > ><filterset> > ><propertyfilter> > ><propertyset> > ><exclude > >property="nant.project.basedir" /> > ></propertyset> > ></propertyfilter> > ></filterset> > ></filtersets> > > > > > > > > > > ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ nant-developers mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/nant-developers
