On Wed, Jul 15, 2009 at 03:49:13PM -0700, Brock Pytlik wrote: > The bug report contains the details, but essentially, I looked at all > the packages on my system and took statistics over the length of each > line. I also plotted a histogram of the length of lines and it looks > like a roughly fairly sharp distribution (if I remember my statistic > terms right) with the biggest peak around 200 characters, and smaller > peaks around 80 characters and 275 characters. > > I agree that picking constants is always an area that's fraught with > danger, but my question is whether you think the average length of our > actions is going to consistently grow over time? If so, then maybe this > is a bad idea. My guess is that by and large, 20 years from now, our > actions are probably going to be on the same order of magnitude. If > that's not true, we can always come back and tweak this.
Sorry, I forgot to look at the bug report. If you've tied this to something that has a physical representation -- the length of an action in a manifest -- then that's probably sufficient for now. It might be wise to leave an XXX in the code here to revisit the buffer sizing if the length of actions changes substantially. I'm afraid I don't have any concept of how we might expect action length to grow. I can't comment with any authority on that subject. I'm fine with the change at this point, provided you've fixed up the mode stuff too. -j _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
