On Mon, Mar 19, 2007 at 09:05:11AM +0100, Werner LEMBERG wrote: > > > Maybe the real issue is that AT&T troff and its derivatives do not > > change to copy mode when reading names: > > > > .ds xx foo > > .ds \\f bar > > \*(\f2xx \n(.f > > > > prints "foo 2" with both 7th Edition troff and Heirloom troff. With > > groff, it prints "bar2xx 1". Consequently, Heirloom troff also does > > not change its mode when reading the contents of a "\[...]" bracket > > sequence. And thus, both occurences of "\E" in "\E*[\E*[foo]]" are > > interpreted as escape characters. > > This is now fixed in the CVS -- \E gets expanded in escape names. The > fix was much simpler than expected. Please test. >
a first feedback (not with regards to the \E stuff itself...): overall it seems to work OK. the small changes to grotty and X75 (inclusion of figure bounding box in grotty output, inclusion of postscript file name in bounding box (also in gxditview) are an improvement, I think, dito absense of the warnings concerning PSPIC when formatting for grotty. what seems buggy is the following: I now get "numeric overflow" errors/warnings for each `.PSPIC' call although the generated output looks alright. joerg
