>>>>> "Herbert" == Herbert Voss <[EMAIL PROTECTED]> writes:
Herbert> Lars Gullik Bj�nnes wrote: >> "J�rgen Vigna" <[EMAIL PROTECTED]> writes: | Well I didn�t see the patch, >> but I can imagine what you did. While I agree | that the pwidth can >> be omited if the string is empty I don�t think we should | omit >> "booleans". I think we had this discussion already and decided to | >> leave them in as if we would change defaults in a future version we >> would | have to maintain more code as we would have now. Yes, that >> is what we decided. And I also think that we need the table code to >> settle a bit before doing this. I also suspect some interanal >> tabular work that will perhaps be quite large. Herbert> Remember that I never changed the read-code of the tabular. Herbert> So it's a good question, why to write a lot of code when it's Herbert> not important for reading??? Because if we decide in lyx 1.3.0 that some bool should be true by default instead of true, then we have big problems when reading old files. What could be done maybe is to output at the beginning things like <defaults> <column alignment="center" valignment="top" leftline="true" rightline="false" width="0pt" special=""> <trow topline="true" bottomline="true" endhead="false" endfirsthead="false" endfoot="false" endlastfoot="false" newpage="false"> <cell multicolumn="0" alignment="center" valignment="top" topline="true" bottomline="false" leftline="true" rightline="false" rotate="false" usebox="none" width="0pt" special=""> </defaults> So that when reading back, we know what assumptions have been made. Does that make sense? JMarc
