MRC%PANDA@SUMEX-AIM.Stanford.EDU (04/02/87)
I fixed the summer time problem in TOPS-20 at about the same time all the messages were flying on Unix-Wizards about it. The vendor- supplied code had three cells; a tri-state flag (never summer time, always summer time, and use algorithm), a last possible day of year to start summer time, and a last possible day of year to end it. I added a vector of "years when rule effective", and made the two day cells be parallel vectors. Finally, I made the code run down the year vector to get the right index into the day vectors. This took me less than an hour of work. Most of it was spent writing documentation so some poor schmuck who has to add a new rule in the future would know what to do (nominally, add a single line such as DST 1993,LSTMAR,FSTNOV if Congress slides it a week each way from the current rule in 1993). This was in assembly language to boot. It is unbelievable that any competant vendor could possibly spend much more time fixing Unix (in C) or VMS (in BLISS). Good gawd, if I spent a day on it I could implement a full summer time engine that let you say which day of the week and add special year-based rules (such as the proposed rule to end PDT a week later on presidential election years). I guess that'd take half a day. I could spend the second half adding rules for all our non-North American friends. Any vendor who says that it will take time off from fixing bugs is making an excuse. I seriously question the competancy of ANY vendor who cannot at least provide the customer with a bare bones facility to set summer time to fit local rules. -------