Uwe Stöhr wrote: > > I do not understand this, and I think this is wrong. What if "iiii" is > indeed > my longest label? > > "iiii" is always the shortest 4 character string, while "WWWW" is always > the widest one.
Well, if I only have one 4-char nomencl inset, and this has "iiii", I do not want the length to be "WWWW", do I? > > + // assure that the width is never below the predefined > > > Why do you forbid to go under the predefined? > > To have a consistent output. In most cases the 1 cm width is enough, so we > have to assure that this width is not undergone, for example when you have > several nomenclatures. I'll send you this evening a LyX example file, to > show what I mean. I still don't understand. Why shouldn't I be allowed to set a narrower width? This looks plain arbitrary to me. > > Also, I do not like the complicated ifthen statement at all. > > Why is ifthen complicated? It adds unnecessary LaTeX code and an unnecessary package dependency. > > I've changed this to use theFontMetrics instead. I think this is a much > better > approximation (and the best we can get). > > I don't think so, because you don't know the font metrics of the output > font, and this is what counts. My solution assured, the there is always > enough space, no matter what font is used. The only case where this might > produce a too large width is when your widest symbol is "iiiiiii", but this > is an extreme and absurd example. No. A too wide result will be returned for every string who has many rather small letters, and given the frequency of "i" and "l", I think this is no exceptional case at all. No. I think assuming always "W" produces much more bad results. I assume that the width relation of two given chars in different fonts does not differ too much. IOW I guess that (apart from fixed width fonts) the string "iiii" is always shorter than "WWWW" in any font. > >> > + // assure that the width is never below the > predefined >> > >> Why do you forbid to go under the predefined? Also, I do not like the > >> complicated ifthen statement at all. > > > > Now that we have a param "auto" and "none", I think this should go. If > the > user selects "auto", we should always use that and not care about the > > predefined length. > > No we have to assure that the width is never below the predefined value. > But to be correct, I'll change the hardcoded "1cm" to the one that is > actually defined in nomencl.cfg, that's only one more line. So you parse the nomencl.cfg file? Jürgen
