There was a discussion around mixins which happened years ago. I actually implemented a POC 18 months ago: https://github.com/apache/maven/pull/1209 Coupled with XInclude, which works quite well imho, despite what some seem to indicate, I think it can already support a bunch of use cases.
I'd be also very happy to provide the needed API hooks to support tiles use cases. The goal with the immutable model was not to forbid such things, but rather to open the API at specific points so that Maven can support them better at the end. But supporting that in Maven can be done in 4.1 very easily. And it won't break the whole ecosystem. The build/consumer POM feature allows hiding changes in the build POM. So we can very easily add mixins support because they would be inlined and translated into the consumer POM when reaching the repository. Le dim. 16 mars 2025 à 09:42, Fred Cooke <fred.co...@gmail.com> a écrit : > I'd like to revive this discussion on tiles and possible alternatives. > > I'll say it bluntly: Maven without tiles or an equivalent is not something > I want to use. And yet there's nothing else that comes close, either. > > I've been using Maven since 2006 (2, then 3) and tiles since 2017, > intensely since 2019. > > The parent hierarchy style is a maintenance nightmare that makes flat out > duplicated configurations look good. > > Once you go tiles you never go back. > > I feel like it's time to accept that parent hierarchies are not good enough > and offer some kind of tiles like composition behaviour in core maven as a > first class feature. A core part of the pom. > > Is it too late? Could I dust off my 8 year unused dev skills and help? > > This is a huge deal. Tiles takes maven from a good tool to a genuine > masterpiece. M4 without tiles would be a huge step backwards even if it > improves numerous other things. > > I have read through the following page, and most of it sounds good: > https://maven.apache.org/whatsnewinmaven4.html > > What's the path forward? > > Cheers, > > Fred. > > On Fri, 24 Jan 2025 at 01:39, Elliotte Rusty Harold <elh...@ibiblio.org> > wrote: > > > If we're going to use something like > > mvn:org.apache.maven:maven-snippet:1.0 let's make sure we define the > > scheme with the IANA. > > > > Yes, I suppose you can use id or element pointers without going full > XPath. > > > > On Thu, Jan 23, 2025 at 12:24 PM Guillaume Nodet <gno...@apache.org> > > wrote: > > > > > > The XInclude spec does not forbid non local references. > > > See https://www.w3.org/TR/xinclude/#include-location > > > So it could definitely be something like: > > > mvn:org.apache.maven:maven-snippet:1.0 > > > Also, pointers are actually supported as shown by the unit tests [1]. > > > > > > That said, we did choose to only support relative URis (in [2]) > > > and I'm sure it would be a good idea to open it to absolute URIs... > > > > > > [1] > > > > > > https://github.com/apache/maven-xinclude-extension/blob/main/src/test/java/org/apache/maven/xinclude/stax/XIncludeTest.java#L111-L117 > > > [2] > > > > > > https://github.com/apache/maven-xinclude-extension/blob/main/src/main/java/org/apache/maven/xinclude/LocalXmlResolver.java > > > > > > Le jeu. 23 janv. 2025 à 12:55, Elliotte Rusty Harold < > elh...@ibiblio.org> > > a > > > écrit : > > > > > > > On Thu, Jan 23, 2025 at 12:16 AM Mark Derricutt <m...@talios.com> > > wrote: > > > > > > > > > > > > > > If the xinclude support could use some form of GAV HREF, rather > than > > > > just a > > > > > local file, that might cover the basic requirements/use cases (tho > > some > > > > > users seem to make heavy use of plugin configuration merging over a > > > > series > > > > > of parents). > > > > > > > > XInclude is defined by the XInclude spec, not by Maven. You'd need to > > > > define a completely different inclusion model or perhaps define a new > > > > URI type for Maven GAV coordinates, then add support for this to the > > > > XInclude engines. I'm also not sure what that might mean. Do you > > > > really want to include an entire pom.xml? Probably not, and XPath > > > > fragment identifiers aren't well understood or supported. > > > > > > > > > > > > > > > > -- > > > > Elliotte Rusty Harold > > > > elh...@ibiblio.org > > > > > > > > --------------------------------------------------------------------- > > > > To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org > > > > For additional commands, e-mail: dev-h...@maven.apache.org > > > > > > > > > > > > > > -- > > > ------------------------ > > > Guillaume Nodet > > > > > > > > -- > > Elliotte Rusty Harold > > elh...@ibiblio.org > > > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org > > For additional commands, e-mail: dev-h...@maven.apache.org > > > > > -- ------------------------ Guillaume Nodet
