Jari Aalto wrote: > The problem is the accessibility: to know that option, it was buried.
It's documented on the man page along with some other variables that are not in the rules file. > | The only exception to this rule, as documented in the debhelper man page > | is this it might possibly be worth putting in the rules file if the > | upstream tarball contains CVS or .svn diretories. The cases where that's > | atually useful is a very small (possibly nonexistant) subset of the > | small set of upstream sources that are so broken. Putting it in the > | rules file will encourage inappropriate use. > > How's that promote inapropriate use? It makes people think "oh, I build from svn, let's uncomment this". > If the tarball includes CVS, > repackaging is considered no-no. It's there for reason. Best practice is to get upstream to fix it. Practically, the kind of problems that DH_ALWAYS_EXCLUDE works around are much less likely to be present if upstream builds the package with CVS dirs. > This is exactly the case. The upstream tarball includes CVS > directories and it took quite some time to figure out how to educate > debhelper to ignore those. The problem was that the familiad debuild > '-i' option was not available in dh_builddeb and following the SEE > ALSO references did not really reveal anything (only that it was top > level program to X, which used Y, which used Z ..., no way to pass > that 'ignore' along way). debbuild -i operates on source packages, not binary packages. -- see shy jo
signature.asc
Description: Digital signature
