On 2019-05-20, Kornel Benko wrote: > Am Montag, 20. Mai 2019, 12:20:11 CEST schrieb Jean-Marc Lasgouttes: >> Le 19/05/2019 à 08:45, Jürgen Spitzmüller a écrit : >> > Am Freitag, den 17.05.2019, 20:47 +0200 schrieb Günter Milde: >> >> commit 05877e0e6019144e5f484ca52e89113ae37843a1 >> >> Author: Günter Milde <mi...@lyx.org> >> >> Date: Fri May 17 14:03:28 2019 +0200 >> >> >> >> ctests: sort dedicated test samples >> > >> > Makefile change is missing.
>> I fixed that now, Thank you. >> but I wonder: shouldn't we just drop the makefile and >> stop this very partial distribution that makes no sense? This is my suggestion. >> Or should we distribute everything? > For me it makes sense. We distribute data to also be able to build with > cmake. Why should the tests be disabled in that case? IMO, the dedicated test samples are developer tools -- they are there to ensure we don't have recessions during code review and cleanup or when introducing new features. As the test setup/configuration only creates test instances for actually found documents, leaving out the dedicated test cases will not prevent people working from the distribution (as opposed to a repository checkout) to run the test suite (it will just be a bit smaller). >> So, s the autotest corpus something we want to distribute with LyX? It >> is small compared to the rest anyway. > The question is, which tests have to be possible. > All, > no one, > a subset (but who wants to decide?) My suggestion is "no one" but I can also live with "all". Deciding on a subset adds a maintenance burden that is not justified by its added value. Günter
