Niels, On Tuesday, August 13, 2024 9:02:43 AM MST Niels Thykier wrote: > Soren Stoutner: > > Andrea, > > > > On Tuesday, August 13, 2024 12:33:57 AM MST Andrea Pappacoda wrote: > >> [...] > >> > >> That's a false positive. Rules-Requires-Root: no is already implied by the > >> build dependency on dpkg-build-api (= 1). Please see dpkg-build-api(7) and > >> bug #1057176. I also think that adding a lintian override is wrong since > > > > this > > > >> should really be fixed lintian-side. > > > > Even so, it is considered good practice in Debian to explicitly state > > "Rules- > > Requires-Root: no” in the control file, and there are no downsides to doing > > so. It shows that the developer has explicitly considered the need for root > > during building time and is confident it isn’t needed. Setting > > "dpkg-build-api (= 1)" might or might not indicate the maintainer has > > specifically considered whether root is needed during build time. It ought > > to indicate that, but it could also indicate the developer specified that > > for other reasons without being fully aware of all the implications. > > Drive by remark: I agree with Andrea and disagree with Soren on this. > > The "dpkg-build-api (= 1)" makes `Rules-Requires-Root: no` the default. > Lintian should fix its false positives rather than people teaching each > other to "work around" lintian. I also do not want people to add > `Rules-Requires-Root` for packages that does not need `debian/rules` > (see #1077936). So I feel very strongly about not just sprinkling magic > fairy dust into the `debian/control` to work around `lintian` not having > a proper definition of whether a package needs `fakeroot`.
I do not consider this to be simply a “lintian” issue. Rather, I consider it best practice as documented in the Debian Policy Manual. https://www.debian.org/doc/debian-policy/ch-controlfields.html#rules-requires-root For me, personally, this is a significant enough of an issue that I would not sponsor a package that did not specify “Rules-Requires-Root”. Other Debian Developers may feel differently on this issue, and I would have no hard feelings if they decided to sponsor the package without it. But, for me, I would not want to have my name attached. -- Soren Stoutner so...@debian.org
signature.asc
Description: This is a digitally signed message part.
