`)read' and large inputs

Hi,

`)read' behaves unexpectedly when passed large input files such as the
one available at:

  http://www.laas.fr/~lcourtes/tmp/axiom-large.input.gz

Basically, when passed this file, `)read' aborts with a "parse error"
message.  However, when the very same file is dumped through the
standard input of `axiom', no error occurs:

  $ axiom < axiom-large.input

Could it be a bug or limitation of `)read'?

Thanks,
Ludovic.