read-process-output-max (was: Emacs HTTP libraries)

[Stefan Monnier]

>> While all other things get discussed, could we perhaps have url.el be
>> updated to locally bind read-process-output-max to a larger value, say
>> 1mb?
>
> I don't think this would be TRT, because that variable affects all the
> communications with all the subprocesses running at that time.

BTW, what are the factors at play here?
I understand there is a per-chunk overhead which pushes towards a higher
value, but I'm not sure what factors are pushing towards smaller values.
I can imagine that a 1MB buffer could be considered more than useful
(hence wasteful) but I don't immediately see what would be the downside
of setting it to 64kB instead of 4096B.

The docstring seems to strongly recommend not to increase it, so
I suspect there's a much stronger reason that "not be wasteful" in play.


        Stefan