On 2016-12-02 Fri 22:54 GMT-0800, Matthew Carter <m@ahungry.com> wrote:

> Michael Heerdegen <michael_heerdegen@web.de> writes:
>
>> Hong Xu <hong@topbug.net> writes:
>>
>>> In the current master branch, it seems that the variable shr-start has
>>> never been used in any sensible way: Every time its value is referred,
>>> it is bound to nil. Should it be considered to be removed?
>>
>> Are you sure it is always bound to nil when referenced in `shr-tag-a'?
>> I would expect it can be changed by side effect in the body, and the
>> let-binding is there to make that change temporary.
>>
>>
>> Michael.
>>
>
> It looks like it's bound at the top defvar level (and changed via setq
> in the img function), so those let's to nil in certain functions are
> likely to inhibit the current setting when those functions begin to
> run.

That's exactly what I meant: Every time it is referred, it is bound to
nil by a let.