Re: FYI: libtool.m4 removal problem

[Charles Wilson]

Ralf Wildenhues wrote:

> Also, I'm entirely unconvinced as to his idea of parallel libtool
> installations.  I can see the value when developing libtool, but
> allowing/encouraging users to do so, I fear, would only increase
> libtool maintenance burden (having to support old releases for
> longer time).

Actually, no.  It's a non-intuitive result, but allowing a graceful 
transition actually SPEEDS acceptance of the new release.  For instance, 
consider autoconf-2.13 vs. 2.5x.  That transition took fer-stinkin'-EVER 
because "side-by-side" installation was left to the distributors to 
manage.  So individual developers resisted "taking the plunge" until ALL 
of the projects they worked on were ready to plunge with them.  But no 
project could "take the plunge" until all of its developers agreed. 
Hence, deadlock.

See
"Strategy Letter II: Chicken and Egg Problems"
http://www.joelonsoftware.com/printerFriendly/articles/fog0000000054.html

(Yes, there WERE ways to have both autoconfen installed, and there WERE 
ways to ensure that a given configure.in worked with both 2.13 and 2.5x. 
 But they were HARD.)

Side-by-side is GOOD.

--
Chuck