On 29.12.2016 02:29, Wookey wrote: [...] Hi Wookey, > So why didn;t my substition rules of > s/bcel/bcel/ s/bcel/bcel/ jar s/5.1/5.x/ > or > s/bcel/bcel/ s/bcel/bcel/ jar s/5.1/6.0/ > work? > > The upstream pom.xml has: > <dependency> > <groupId>bcel</groupId> > <artifactId>bcel</artifactId> > <version>5.1</version> > </dependency> > > The only thing wrong here seems tobe the version, so what should I do > here to make mh_make happy that libjava-bcel contains the right stuff? Don't try to hard to make mh_make happy because it can only give you a rudimentary skeleton debian directory. (For "easy" packages it works much better but more complicated projects won't work as expected). We usually use existing packages, which are quite similar, as templates. > I also got the complaint that it can't find: > backport-util-concurrent > > Do I care about that? Do I need to package that too? <sinking feeling emoji> Welcome to Java hell ^-^ > > Also how does this 'debian' version thing work? Is that like saying > 'generic/current' version? And we do that to deal with all these > packages asking for a specific version when they really don't actually > care (probably). So should I prefer to set the versions for all this > stuff to 'debian' as that's likely to stay working for longer? Java is version centric, that means you have to declare that your software project works exactly with a specified version and if you try an older or newer one you are basically on your own. :/ As you know in Debian we try to package only one version of software project X and thus we need to use a generic "debian" version (and patch packages that don't work with it). The debian tools (maven-debian-helper) always rewrite the version to "debian" by default. You don't have to change anything here. There are a few libraries which are known to horribly break things. Here we use a substitution rule like 1.x for versions ranging from 1.0 to 1.99 etc, instead of using the generic "debian" version because we know that version 2.x will break the package. > Cheers for pointers. I guess it would be easier if you gave me some link to your package and I try to figure out why it fails for you at the moment. Please also describe what you want to achieve, so that I get the bigger picture and if you're lucky I might be able to explain why it didn't work as expected. Cheers, Markus
Attachment:
signature.asc
Description: OpenPGP digital signature