with-eval-after-load with sh-mode

[lisa-asket]

>From: Arthur Miller <arthur.miller@live.com>
>To: lisa-asket@perso.be
>Subject: Re: with-eval-after-load with sh-mode
>Date: 02/07/2021 10:56:14 Europe/Paris
>Cc: help-gnu-emacs@gnu.org

>lisa-asket@perso.be writes:

>>From: Arthur Miller <arthur.miller@live.com>
>>To: lisa-asket@perso.be
>>Subject: Re: with-eval-after-load with sh-mode
>>Date: 02/07/2021 10:42:38 Europe/Paris
>>Cc: help-gnu-emacs@gnu.org
>
>>lisa-asket@perso.be writes:
>
>>> How can I use `with-eval-after-load` with sh-mode?
>>>
>>> Basically I want to defer call to sh-mode-map until sh-mode is loaded
>>>
>>> (with-eval-after-load 'sh
>
>>Try this:
>
>>(with-eval-after-load 'sh-script
>>(message "My cool one-time config follows here ..."))
>
>>sh-mode is defined in lisp/progmodes/sh-script.el.
>
> What is the difference between 
>
> (with-eval-after-load 'sh-script
> (myfun))
>
> and
>
> (add-hook 'sh-mode-hook #'myfun)

>Obviously sh-mode-hook gets executed every time you load sh-mode, i.e,
>when you open a shell file.

>eval after load will kick in when feature is loaded, i.e. when you read
>in (load or require) sh-script.el, so it least in theory it should get
>executed only once, unless you unload feature 'sh-script for some reason.



This leads to another consideraion.  There are certainly advantages about using

`with-eval-after-load`.  But, then, what are the circumstances for using 
`add-hook`?