bug#30675: Ask the user what to do when shr-make-table: Variable binding depth exceeds max-specpdl-size

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

> I'd feel much better with a lower factor, e.g. some value that is just
> enough to cover this case plus some slack.  Bonus points for providing
> a defcustom with the factor, so that users could change that.

The problem is that HTML may nest arbitrarily deep, so there isn't
really any way to cover this completely.  But the larger the stack size
is, the lower the possibility for hitting the limit is.

FWIW, I haven't hit the limit in about a year, so it's rare for people
to encounter HTML that's that deeply nested.

> The idea of asking the user whether to increase by N levels sounds OK
> to me, provided that its implementation is not out-worldly hard.

I kinda feel that stack size is something that a user has little idea of
what it would mean.  Emacs: "We hit a magical limit!  Raise the limit?"
User: "Sure!"  Emacs: "Now it worked!"  I mean...  why consult the user
at all?  :-)

> Well, if I need to give my opinion on this, then blindly
> increasing the limit ten-fold is something that'd make me worry about
> a potential C stack overflow. 

Hm...  On what architectures is there a potential for a C stack
overflow?  I thought that these days, C stack overflows basically didn't
happened, and we only had the max-specpdl-size as a sanity check to
avoid deep lisp backtraces and stuff, but I may well have misunderstood
things.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no