HH> first of all, we're not dealing with a bug, it's a feature; so, if some HH> other behavior is wanted, it should be an option of extension to existing HH> mechanisms, nog a replacement. HH> there is one thing we should keep in mind: naive users are not always aware HH> of tex's empty line == par, so this is why \start/\stop things are HH> consistent in absense or presense of them; HH> distinguishing between display and semi-display is therefore not that HH> trivial, and the best option is separate environments, like HH> \startquotation ... \stopquotation HH> (or \startdquotation ... \stopdquotation for display quotation) versus HH> \startpquotation ... \stoppquotation HH> for an in-paragraph one (like we have with formulas, see core-mat.tex) HH> that way, one knows what to expect, since it's clear from the coding, and HH> not from a (forgotten, or accidental) empty line HH> another option is to have a dedidated environment HH> \startinparagraph ... \stopinparagraph HH> or \startcontinue ... \stopcontinue HH> which nills the surrounding whitespace and inhibits the next indent. (this HH> should be not that hard to implement) This latter alternative sounds awfully 'redundant' (too much code for a simple thing); the first sounds somewhat better, but the best solution IMO is to have an option/conditional \obeypars or something like that. When true, empty lines after startstops denote paragraph and lack of empty lines denote "remaining in the same paragraph"; when false, current ConTeXt behaviour is kept. Maybe we may want to have a per-startstop option (obeypars=true,obeypars=false) together/in place of the global \obeypars/\noobeypars (or \obeyparstrue/\obeyparsfalse). Let's say that the best thing (again IMO) is to have a global option, overridable in each startstop: the obeypars key in startstops would then accept the values true (force true), false (force false), default (behave according to the global setting). Does this sound sensible? -- Giuseppe "Oblomov" Bilotta