Re: How to add pseudo vector types

[Stefan Monnier]

>> If we copy the buffer's content to a freshly malloc area before passing
>> that to TS, then there should be no problem running TS in a separate
>> concurrent thread, indeed.
> Making a copy of the buffer is a non-starter from where I stand.  It
> doesn't scale, for starters.  I don't see any reason to go to such a
> complex design at this early stage.

I see absolutely no problem with scaling in making a copy: the extra
memory and CPU time taken by the copy will be a constant factor which
I don't expect to go much beyond 10%, which doesn't threaten scaling and
seems perfectly acceptable in return for being able to perform the
parse concurrently.

I'm not sure we'll want to do that, but I see no reason to consider it
a non-starter.

[ BTW, it's not clear to me if an update needs to be able to read the
  whole buffer or if it only needs access to the "update
  description".  ]


        Stefan