On Mon, May 23, 2005 at 06:23:03PM +0200, Kurt Roeckx wrote: > On Mon, May 23, 2005 at 08:39:12AM +0000, Gerrit Pape wrote: > > On Sat, May 21, 2005 at 02:40:13PM +0200, Kurt Roeckx wrote: > > > Package: libdjbdns > > > Version: 1.05-3 > > > Severity: important > > > > Hi, you think a non-perfect library selftest is an important bug? What > > severity do you use when reporting bugs against all those library > > packages that don't include any selftest at all? > > I could set this to serious too because it caused a build failure > for me. It's causing a segfault.
Hi Kurt, then please say so and do so. And please include the build log and information about the build environment. The package builds fine for me, and built fine on the autobuilders. Maybe you're running a kernel with some security features. > And I hardly call this a selftest, it doesn't even check the > return values of the functions, I wonder why you even bother to > run it. If you want to improve the selftest, I would be happy to review and consider a patch. The test ensures that programs using the public library functions do actually compile and successfully link, and don't fail with undefined references or so. See debian/copyright, I added the test while identifying ``all necessary lower-level .[ch] files''. Funny that you want me to justify a selftest; any selftest, regardless of the quality, is a good thing to check for regressions, if it succeeds reproducibly. Gerrit. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
