On Mon, Jun 19, 2017, 5:21 AM Nicolas Goaziou <mail@nicolasgoaziou.fr> wrote:
  There is absolutely no drawback in using lexical binding. Since Org
  9.0, it _is_ the default for Org core: almost every Org library
  activates it nowadays. 

Again, lexical binding has _no_ drawback and makes life of developers
easier (e.g., code is more readable, compiler reports more errors).
I moved almost every library in Org to lexical binding, some changes
being trivial, some painful, for a reason. I don't want to do a step
backward in that area without a very strong reason–to tell the truth,
even a strong reason wouldn't convince me.

Here are some of my observations on the topic of lexical-binding by following emacs-devel closely for the past few years. 

It has become a norm to write new elisp code that is lexical-binding friendly. 

I have seen this movement started on emacs master for a long time now (since 2011), and still there is a continuous effort to enable lexical-binding on more and more emacs core files: http://git.savannah.gnu.org/cgit/emacs.git/log/?qt=grep&q=lexical-binding

The same applies to Magit and many other external emacs packages ( https://github.com/search?q=lexical-binding+language%3A"Emacs+Lisp"&type=commits ). 

The idea is that once almost all the elisp code out there is lexical-binding compatible, the default of emacs can be changed to that and dynamic binding can be obsoleted.

I have also seen how a lexically bound package is more portable as there is no implicit reliance on global variables from multiple other packages.
--

Kaushal Modi