Jürgen Spitzmüller wrote: > Uwe Stöhr wrote: > > What is if we add a layout that did not exist before? Of course this breaks > > the backward compatibility. > > Yes, that's tolerated. Although it is an edge case (there are some good > arguments to limit the inclusion of new layouts to major releases).
Agreed, it is edge-case already. If this is questioned we can ask whether we should stop doing it -- not that it's an argument for including even more. > > I was even forced in a long discussion to install LyX in a folder named "LyX > > 2.0" instead of "LyX 2.0.5" with the argument that one doesn't need to have > > 2 different versions of LyX installed. Although I stated that we could have > > introduced a regression accidentally did not count. But now you tell me > > that distinguishing between bugfix releases is important. This is > > inconsistent. We consistently distinguish between development version (2.1, trunk) and branch (2.0.x), which should be bugfix releases and not route for new features. It would make some sense to allow 1.6.x and 2.0.x live together, it is nonsense to let 2.0.x versions installed together. Although we fail from time with some particular regression the basic attitude always was that switching between branch versions should be rather indistinguishible from users POV. Please make clear what is inconsistent here, I fail to follow your logic. > > Looking at other programs, please tell me one program that provides backward > > compatibility. Even programs like Firefox introduce new things with every > > release. If they add a new HTML5 feature you won't benefit from that if you > > downgrade. Of course your HTML5 file will look different with Firefor 16 if > > you are not using Firefox 17. And all the advisories are telling you that > > you should always use the latest program versions for security reasons, no > > matter what program. The same is with all Mozilla programs, No it is not and you clearly have no idea about Firefox release model, so lets get the facts first. Firefox introduce new features in 10,11,...,17/etc versions like we do with 2.0, 2.1, 2.2 AND THEN he bumps stable releases for chosen versions like 10.0.1, 10.0.2, 10.0.3,...; 17.0.1, 17.0.2, ... This for example means that the default stable firefox on my linux distrubution is 10.0.11 today and 17.0.1 is flagged as testing if you are adventurous enough. You can do this on windows as well and I'm sure any institution which looks for stablity/security goes through this 10.0.x,17.0.x,24.0.x cycles, not through 10,11,12... It is only that Firefox tries to beat other browsers with new features so that normal windows user have no idea about stable/security releases and about which Mozilla is pretty silent ;) That was just for example; if OpenOffice or MS Office does not provide stable releases it is not argument for us to stop release model we use and most of us (as far as I can read the thread) consider to be better or even reason why we stick around LaTeX world instead of Office world. Pavel
