On 10/14/2015 05:25 PM, Denis Silakov wrote: > 14.10.2015 21:18, Russ Allbery пишет: >> Debian also relies heavily on upstream software maintainers to handle >> testing and debugging individual software components rather than doing a >> lot of distribution-level testing. > > Meanwhile, a good solution here is to integrate LSB tests in upstream > development cycles. This topic was discussed a lot, but unfortunately > there was only a little progress in this area. One of the major problems > is that test suite is often a separate complex product that requires > additional integration and maintenance work. Another problem is that > interpreting test results often requires significant time and experience. > > A good example is testing glibc with LSB tests at upstream - > http://sourceware.org/glibc/wiki/Testing/LSB_Core. That work is > performed by a volunteer who participated in LSB test development long > time ago. The tests are useful and do detect bugs in glibc, but this > activity completely depends on a single person - if he doesn't have > enough time to run the tests and to analyze the results, then LSB tests > are not launched at all.
LSB has taken some criticism for the tests being too hard to run and too hard to interpret. It's partly true as Denis notes above, but good testing is not simple, and LSB had a fair bit of contributed code which came with its own testing methods, you do what you can. Still, I feel like productive collaboration with upstreams could make tests that work for abi/api stability both in the upstream's environment and still be usable in the LSB test environment. The "market" for such things is highlighted in this notice: https://lists.ubuntu.com/archives/ubuntu-devel-announce/2015-September/001152.html Look in particular at reference [3] and look for the changed line 57. _______________________________________________ lsb-discuss mailing list [email protected] https://lists.linuxfoundation.org/mailman/listinfo/lsb-discuss
