Uwe Stöhr wrote: > Nevertheless one must discuss how useful the tests are if they make so > many troubled what anybody just changed a layout file without touching > tex2lyx. I still think that the main purpose of the tex2lyx tests should > be to test tex2lyx capabilities.
Exactly, and this is what they are doing. In the past, tex2lyx was more or less independent from LyX, and one could hack around in LyX without the need to touch tex2lyx at all. The big disadvantage of this approach was that tex2lyx was always lagging behind, and a lot of code duplication was needed. Therefore, more and more code was shared, but this created a dependency of tex2lyx on many parts of LyX, including the layout files. I should probably add a section about the history to Development.lyx, maybe this is then easier to understand. Today we have the situation that a simple change to a layout file can make tex2lyx produce an invalid file format. Therefore the tests need to be run and updated even for such a change. > The point is that every new rule increases the barrier to attract new > developer and to hold developers active. True, and I think we all agree that we want to have barriers as low as possible. However, the tex2lyx tests have already been payed off a lot. For them, the barrier is only in the initial setup, once you are used to running them it is very little additional effort. I guess you remember the time some years ago when both you and I invested a huge amount of time to bring tex2lyx up to date regarding the file format. This was a very un-fun type of work, and when it was finished I was sure that I never want to do this again. This is the reason for having the tests and the rule to keep them up to date. In terms of total cost it is the most efficient way to keep tex2lyx up to date. > In my case I knew I would have > maximal 2 days not knowing when I will be able to LyX again. And I of > course knew that the first LyX 2.2 cannot be far away and then a > fileformat change for simple things like layout changes are then no > longer possible. So sometimes just putting things in is sensible. (I > know Georg hates me for statements like that.) No. This is not a personal issue. I do however seriously ask you to think about the statement "sometimes just putting things in is sensible" again in the context of the questions Scott asked: What if others would think alike? What would that mean for the project, and the working atmosphere? How would you feel personally if others did act like that? > However, now I have the > situation I feared but my feature is in and there is time to fix > possible issues in the alpha/beta cycle. I don't see the big problems > despite that I stole your time with the tex2lyx test issue. IMNSHO, stealing this amount of time _is_ a big problem. I hope that this is now clear. > So finally apologies to Georg, Guillaume, Kornel and Günter that you > need to spend time for tex2lyx. > I hope we will find a solution soon that I can do things right also on > Windows too. It depends on you. You need to be willing to spend some initial setup effort, and if you are, then it is easy, and we can walk you step by step through the process. The result will be a batch file that you can run whenever you need to execute the tests, and you can have another one for updating the references. Georg
