On Fri, Jan 01, 2016 at 01:22:53PM +0100, Kornel Benko wrote: > Am Donnerstag, 31. Dezember 2015 um 19:33:59, schrieb Georg Baum > <georg.b...@post.rwth-aachen.de>
> > This is not a bug. The conversion 501->500 must be lossy, since the format Thanks for the explanation. Good to know there is no bug. > This is not causing problems for test. In fact, we do not try to export to > pdf/dvi > on such converted lyx files. Indeed. I ran these "tests" manually by doing the lyx2lyx. Although if we had the framework I would have been in favor of using it. In this case we would have just added the test to the inverted/ignored list. > > IMHO it does not make sense to blindly test more and more combinations of > > conversions of the existing documents. > > The recursive export test with lyx??x has proven to be helpful (uncovering > some lyx2lyx bugs). > > > I would rather put the energy in > > creating dedicated tests, e.g. for each lyx2lyx step. > > What we have now is _one_ step in this direction. Dedicated tests requires understanding of the underlying mechanism. I do not know much at all about lyx2lyx (I'm not sure Kornel does either?). But despite this lack of knowledge we were able to discover bugs through the use of the less-than-ideal tests. I would love to have dedicated tests but for them I think we need the people that implement features or fix bugs to make the tests. > > I must admit that I do > > not understand the current test strategy anymore, since there are several > > categories, and the test code became so complicated (e.g. running lyx2lyx > > in > > a loop or automagically changing document settings). > > 1.) loop: > What we do is (in checking lyx21x) > export xx.lyx to xx.21.lyx > export xx.21.lyx to xx.21.21.lyx > 1a) uses lyx2lyx to convert to newest lyx format > 1b) checks if xx.21.lyx is loadable > stop loop if xx.21.lyx is identical to xx.21.21.lyx > 2.) changing document settings: > for system fonts this was needed for older lyx versions. Now, if the > system fonts are specified in the > lyx-file, it is not done. I don't understand either *why* the above should discover bugs. But it did in the past and it is thus likely to do so in the future. Knowing only that makes me think it is a good idea so I'm glad Kornel implemented them. Scott
signature.asc
Description: PGP signature
