On Sat, 8 Dec 2001, Ken Hornstein wrote:
> >I'd say it still only warrants a 1.1. There are insufficient new
> >features added or changed functionality. Leave 2.0 for a major
> >rewrite.
I will side with Doug on this (Sorry if I'm being difficult ;-( ). My
reasons are explained below.
> Are you sure? Have you looked at the changes? There was a whole lot
> of cleaning up that was done, and I don't think the security stuff was
> insignificant either (but I'm obviously biased).
Certainly not insignificant, but we had APOP and KPOP before. While SASL
extends this to another level, it is an evolutionary change from the
previous part, not as revolutionary as may merit a doubling of release
versions. Then again, nmh-0.27 -> nmh-1.0 doesn't look like there was a
complete rewrite either.
As for the changes:
34627 lines of unified diffs
---------------------------------
2096 are from new config.{sub,guess} from FSF
2811 are from a regenerated configure
22802 are from man pages. There were not significant content
changes (except for a few cases). Mostly a lot of formatting
fixes and standardization
982 mts changes. Standardization of transport and SASL
2800 uip changes. SASL POP, header path changes
667 sbr changes. nothing interesting.
These statistics are wildly innacurate, because they count some CVS files,
and may not reflect that many source files were moved around. What I
really wanted to show is that you cannot infer the amount of functional
changes by diff. The man pages have some changes, true, but diff | wc
-l contributed by them should not be justification for 2.0.
Better would be to comb through the changelog, distill the differences by
category, eliminate duplicates ("fixed A." "oops, fixed A for real"), and
rank in importance. Doublessly, SASL will be one of the biggies, but I
think there was a fair amount of SMTP transport unification (Done by
Ruud?).
> >I think bumping a version number simply to draw attention to a
> >realease is a bit of a lie. The nmh code is still quite a mess. I'd
> >rather see 2.0 be a nice clean release. Why not kick this out as 1.1.0
> >(with a 1.1.1 coming soon after with the additional patches from the
> >archives) and instead write up a roadmap of what 2.0 should do?
>
> I'll think about it; I don't have really strong feelings that this should
> be 2.0, but enough people disagree with this one that maybe it's the wrong
> call. I guess I'll save my energy for my first controversial CVS commit
> that I know is going to get people screaming :-)
I am stronly in favor of 1.1.0, followed by 1.0.5. Really, this could have
been released 1.5 years ago, and at that point, it would have been 1.0.5.
Shantonu
