Hey David, this is a troll... Concernant « Re: default inputenc/fontenc tight to language », David Carlisle écrit : « » But more interesting to me is to know » whether any differences are due to technical abilities of the two » systems of course not. VF supersedes mltex completely. Moreover, VF allows to kern differently accented letters from the base, which mltex can't. » or whether they are just different design choices by the authors » of the vf fonts. One single-user french group has been fighting for years against Jörg Knappen's DC/EC fonts. Among the reasons were, if i remember well: accents too flat, dieresis too low and bold (it's an umlaut, and not a dieresis, in fact), some inaccuracies in weight between misc symbols and text ones. Good fonts have different accents over upper case letters than over lowercase/small caps (the small caps accents are not reduced cap-accents, but same as lc). CM, either with mltex or VF, can't provide these refinements; EC could, but not as perfect as we would like. Discussing about that with Jörg during the years inbetween DC and EC, he talled me that he had got feedback from 2 or 3 french users about the glyphs concerning them. So it seems that nobody cares that much about the `quality': neither CM nor EC being complete top quality solutions for french typography. Automatically placed accents like with \accent are typically too high, and sometimes too much centered (the acute accents should be somewhat to the right of the width of an e, CM is essentially correct in this respect; the circumflex is too high and maybe too light). There is no reason why a VF should emulate \accent, in fact Thanh, when preparing fonts for czech with special accents, worked hard to obtain less slopy cap-accents, and better placed lc accents. » > i think to accent placement made dynamically by the output driver on » > the basis of \special commands » » Unless I am totally confused, that isn't what mltex does, is it? Nobody does it! It could be an opentype feature (reacting to a language tag). And how would you manage to make that device independant (you're talking of a _default_ system!)? That could be done with virtual fonts, and family (NOT encoding) switches depending on languages... Notice that I _never_ saw any argument between germans and frenchs about the dieresis in Times. Do they _really_ need distinct glyphs? » > Clearly i refuse any solution which would give me, defaultly, » > a less quality output, Bernard, I say it politely, you're plain wrong. We're talking about default encodings, and you're answering about existing font _shapes_. Choosing anything with less features than T1 as default encoding restricts the possible quality with any font set (once again, kerns + hyphenation \implies T1 or LY1). » Fair enough. Getting quality output is of course the aim of the game. But precisely, here the problem is to devise the good defaults that don't limitate quality. We're not expecting from article.cls that it be an example of `good layout quality', it's only here to define the standard markup for preprints. We're not expecting that the default font used be nice to look at, we're just expecting that the right glyph be in the right slot, and that everything that should end on the page be where it should, so that we can safely read the preprint, or print it as _copy_. An actual typesetter will use his own system to generate the films, using his own defaults, that may or may not be latex's ones. Aesthetics are totally irrelevant in the current discussion, imho. Thierry Bouche __ « Ils vivent pour vivre, et nous, hélas ! nous vivons pour savoir. » Charles Baudelaire, Paris.