> > Torsten Curdt a écrit : >> >>> - the exceptions could extend IOException >>>> >>> Could - but why restrict it that way? (composition over inheritance) >>> >> >> I don't see this as a restriction. >> > > Well, it means all ArchiveException necessarily need to be IOExceptions. > That's restricts their type. > I don't see the benefit. >
I agree with Torsten. There may occur other exceptions than plain input/output errors. > > An issue during a compression or archive operation is an IO exception to >> me. The TrueZIP developers made the same assumption: >> >> >> https://truezip.dev.java.net/nonav/javadoc-6/de/schlichtherle/io/ArchiveException.html >> > > I don't realy see that being an argument :) ...they^^^^he did other very > weird design decisions, too. :) > We had a discussion at this list long time before. If you are interested, you may want to search the mailinglists. > > > - Ant has a JarMarker class that extends ZipExtraField, is there an >>>> equivalent ? >>>> >>> Could you elaborate? >>> >> >> I don't know what it is, but it seems the Ant source is more recent than >> the code in compress, it might be good to resync the code with Ant. >> > > +1 > Torsten: I will do that for the GIT repository. I also will clean up style issues then, i don't like hungarian notation too. Hm ...indeed. But IMO the changeset stuff is where it actually gets > interesting :) > >> It's also the most debatable topic, and it will delay the release. The other stuff is ready for a release. About releasing: i like that "release often" principle. I am not sure about why we didn't release compress in the past, but I guess that we hadn't enough community about that. A release 0.1 with just the cleaned-up streaming classes + testcases feels good to me. 0.2 could have that improved ChangeSet implementation then. Best Chris
