Derek Price wrote:


So, if the history file path were configurable and run through
strftime(), I can either (1) cache the history file name the first time
it is determined and gaurantee that all the operations from the same
server execution are logged to the same history file, or (2),
re-strftime() each time history_write() is called, and guarantee that
given two consecutive lines of any single history file, there was no CVS
operation that occurred in the span of time between the lines.

I'm leaning towards (1) since it is a little easier to code, only
requires one set of system calls per CVS server run, and CVS always
sorts history lines before displaying them anyhow.  Anyone have any
different opinions?



Bug-cvs mailing list

Reply via email to