gwyn@smoke.BRL.MIL (Doug Gwyn) (02/04/90)
In article <22266@mimsy.umd.edu> chris@mimsy.umd.edu (Chris Torek) writes: >... I think it was a mistake to add line-buffering to stdio at all: ... >... The right solution is to put fflush calls into the offending program. Well, well, Chris and I found something else we agree upon. To my mind this automatic-line-buffering-on-terminal-streams feature is yet another example of some software designer (maybe even DMR in the case of stdio) trying to make support functions more convenient to use WITH RESPECT TO A "TYPICAL USAGE" MODEL. Unfortunately, as usually happens in such cases, the "convenience" feature gets in the way of unusual but interesting applications. Just because an open file descriptor is associated with a device that understands terminal control commands does not mean that there is an interacting human being on the other end. Just because an open file descriptor is associated with a device that does NOT understand terminal control commands does not mean that there is NOT an interacting human being on the other end. The model that associates these aspects is too limited.