Uwe Stöhr wrote: > > I'm very surprised about this. The longtable doc does not mention that > > \endhead needs a prior \endfirsthead. > > It is the opposite way: \endfirsthead has to be before \endhead
That's exactly what I wrote. > > I see why the problem occurs, but I think the fix is too general. I > think the > fix, whatever it is, should only apply to cases where captions > are actually > used. > > When you have a header but no first header, the the first header is empty. > In this case the first header is now also set as empty in the dialog. I > think that this is uncontroversial. > > It was already the case that you get an \endhead when the first header was > set to empty. So I haven't introduced anything new. Hm? I don't understand. You changed the behaviour, this is something new. I'm not sure having and automatic \endfirsthead has unwanted side effects. Remember, we had bugs in the past due to the longtable dialog trying to be smart. > Having \endhead before \endfirsthead is always correct, no matter what you > are using. Yo can test this by setting the first header as empty in LyX > 1.6.3. You will see that you still can compile all documents - this has to > be the case, otherwise the empty option would have to be removed. > > So as we already provide the empty option it is safe to use it. But you use it automatically, and I'm not sure this is always desired. If this really has to be used in order to fix caption behaviour, you should at least restrict this to the case when captions are used. BTW, the testcase you provide on trac still shows the problem for me even after your fix in trunk. In any case, I'd like to understand the actual problem first. Jürgen
