Uwe Stöhr <uwesto...@web.de> writes: > It is not acceptable that this tex2lyx issue > http://www.lyx.org/trac/ticket/6059 > eventually delays LyX 1.6.4. I therefore reverted now my controversial > commit in branch and opt for the following solution that was basically > proposed by Vincent:
Thanks. I am not sure why I did not see your message before replying to the previous one. Sorry. > - Use JMarc's solution as default. > - When this fails because the TeX file was externally modified, look > in the lookup table. I am not sure how we can determine that it 'fails'... When there are no comments? With the patch I posted to 6127, we should be in a situation where your method is used in addition to mine. > Additionally: > - fix the tex2lyx bug that this TeX file cannot be compiled after reimporting: > > http://www.lyx.org/trac/browser/lyx-devel/trunk/src/tex2lyx/test/test-insets.tex > - add a comment for each lyx-generated preamble command, no structure just > a comment saying what will follow Why is this better? I also would like personally to have comments that are (also) machine readable, to avoid hardcoding them in tex2lyx (that would also allow to translate them through gettext). > - switch from \newcommand definitions in LaTeXFeatures.cpp to \providecommand > to avoid potential problems with roundtrips Yes, why not, although we cannot always demand that from textclass authors. My own shopping list would be: * make sure that layout files preamble lines are treated as correctly as internal ones. In particular for private layout files. * make as much effort as possible to avoid preamble inflation on import. This might not be a problem in many case, but accumulation of preamble lines is a bug waiting to happen. For example, currently \newlength use will produce a typeset error when repeated. JMarc
