Thanks for that. Just one remark about transitive dependencies automatically being embedded as well, there is the excludeTransitive flag which can be set per embed or subpackage (https://jackrabbit.apache.org/filevault-package-maven-plugin/generate-metadata-mojo.html#embeddeds). Usually it is a good idea that the all package only embeds direct dependencies to prevent transitive packages from being accidentally embedded. Probably that helps already. Konrad
> On 17. Jul 2023, at 16:16, Roy Teeuwen <r...@teeuwen.be> wrote: > > Hey Konrad, > > OK, I created a ticket to follow this up: > https://issues.apache.org/jira/browse/JCRVLT-711 > > I had an initial implementation ready, but this only worked when the > container archive was exploded / extracted. Trying to see how I can get it to > work when the container archive is also a zip > > You can see the initial effort here: > https://github.com/royteeuwen/jackrabbit-filevault/pull/new/feature/JCRVLT-711 > > Greets, > Roy > > >> On 14 Jul 2023, at 12:16, Konrad Windszus <k...@apache.org> wrote: >> >> Hi Roy, >> >> Sure you can easily implement such a FileVault validator rule, but that IMHO >> doesn’t require any changes on the Validation API nor any skip logic. >> I would just make the rule configurable to check for banned content package >> ids. >> That way it is really quick and doesn’t do harm if it is executed for every >> sub package. >> I would appreciate if such a rule could be donated to ASF FileVault so other >> could use it too easily. >> >> Konrad >> >>> On 7. Jul 2023, at 07:31, Roy Teeuwen <r...@teeuwen.be> wrote: >>> >>> Hey Konrad, >>> >>> Sure, the case is the following: >>> >>> I have the following reactor module: >>> >>> - all >>> - core >>> - ui.apps >>> - ui.content >>> - it.content >>> >>> I only want the it.content to be installed to specific environments, namely >>> local and an automated builds environment. To do this, the it.content is >>> defined in the embedded section of the filevault plugin, but I check if an >>> environment variable is available and profile-wise add the it.content >>> dependency. The setting failOnMissingEmbed is set to 'false'. >>> >>> I now had the case that another submodule created a dependency on >>> it.content, making it.content available as dependency to the 'all' package >>> and installed on the wrong environment. >>> To fix this, I'd like to see if I can add a FileVault validation rule to >>> the 'all' package to state one of the following (whichever one is doable): >>> >>> - it should not contain an embedded package named it.content >>> - it should not contain any subpackage that has as a filter >>> /content/${mysite} (preferable, because that way you actually really check >>> what you don't want to happen, even if it would be added on accident to any >>> other package) >>> >>> Thanks! >>> Roy >>> >>> >>>> On 6 Jul 2023, at 07:51, Konrad Windszus <k...@apache.org> wrote: >>>> >>>> Hi Roy, >>>> Usually one needs to distinguish between sub packages provided from >>>> outside Maven reactor (which should be skipped) and sub packages coming >>>> from reactor modules. As the latter ones are already checked individually >>>> in that case using skipSubPackageValidation is usually sufficient as >>>> checking the sub packages again won’t emit different validation issues. >>>> Maybe you can elaborate a bit on your use case... >>>> Feel free, though, to open a JIRA issue and (in the best case) provide a >>>> PR for skipping sub packages with specific Maven coordinates. >>>> Thanks, >>>> Konrad >>>> >>>>> On 5. Jul 2023, at 21:13, Roy Teeuwen <r...@teeuwen.be> wrote: >>>>> >>>>> Hey all, >>>>> >>>>> I'd like to skip the validation for only a certain subpackage (with other >>>>> subpackages contained in that subpackage). Is this possible? I see that >>>>> there is a skipSubPackageValidation, but this skips all sub packages >>>>> while I only want to do one. I also see you can make custom >>>>> validatorSettings, but if I understand this correctly, I'd have to >>>>> specify every validator to isDisabled true, which would become a long >>>>> list. So i'm wondering if there is a shorter way? >>>>> >>>>> Greets, >>>>> Roy >>>> >>> >> >
