![]() |
I haven't had or taken the time yet to review what the script is going to do, nothing good I suspect. I know hylafax.org has a comment from someone who has run the clock forward and seen faxes continue to go, so I guess the most likely problem is reporting breaking. My first impression is that at worst it might be necessary to flush the xferfaxlog on Jan 1 so that all dates within are "00" dates, thinking that at least all dates will then be monotonically increasing. I know what you mean wrt the send and recv log runner scripts... :-O Obviously the scripts have to be fixed in tandem with the faxq/sendfax/sendpage programs, otherwise reporting will fall apart. -Alan Robert Colquhoun wrote: > > Hi Alan, > > At 04:48 PM 12/13/99 -0800, Alan Sparks wrote: > >I can't find a mention of this in the archives... Is there a patch for > >HylaFAX 4.0pl2 to fix the Y2K bug in xferfaxlog and the faxcron script? I > >note that the xferfaxlog BUGS section states, "The date format will > >``break'' in the year 2000." Hmm.. > > > >Patches? Suggestions? Thanks in advance. > > The xferfaxlog file records the day with 2 digits for the year ie MM/DD/YY > HH:MM > > The xferfaxstats script parses it to produce a summary report....does > xferfaxstats break if fed output where some of dates have a '00' year? > > If needed i can quite easily fix it so that the faxq/faxsend output the > date with four digits, but i am not really very keen on modifying the > xferfaxstats awk script it is way out of control ;-) > > Cutting from line 184: > > # > # Convert MM/DD/YY hh:mm to seconds. > # NB: this does not deal with leap years. > # > function cvtDateTime(s) > { > yday = substr(s,7,2)*365 + substr(s,4,2) - 1; > mon = substr(s,0,2) + 0; > for (i = 0; i < mon; i++) > yday += daysInMonth[i]; > return yday*FULLDAY + cvtTime(substr(s,10) ":00"); > } > > ...which offhand looks like it will break! > > PS At worst this should just affect the fax reporting not the fax servers > operation, i think. > > - Robert