Re: Dynamic modules: emacs-module.c and signaling errors

[Paul Eggert]

On 11/25/2015 12:50 AM, Daniel Colascione wrote:
> Modules in C already have to handle checking for failures of their
> internal allocations. What makes checking for failures of Emacs-side
> allocations so much worse?
Allocation is more-common in Lisp-like programs than in C-like programs, 
so it needs to be catered to more. Traditionally, C dynamic allocation 
was often done on the stack, where the programmer doesn't need to 
manually insert code to check whether it succeeds. Lisp and module code 
should be as convenient as that, for 'cons'.

This raises another issue, though. Suppose Emacs detects C stack 
overflow and signals an exception when in the middle of a module. How 
does that work? We need to solve the stack-exhaustion problem 
independently of the details of the module API for memory allocation. 
And whatever method we come up with, should make it easier for us to 
provide convenient heap allocation too, because stack and heap 
exhaustion can be handled by similar mechanisms.