Subject: Re: Do I have to be an expert to get performance: CL versus Perl
From: rpw3@rpw3.org (Rob Warnock)
Date: Fri, 23 Dec 2005 23:00:54 -0600
Newsgroups: comp.lang.lisp
Message-ID: <i-6dndUYCJGbSzHeRVn-jg@speakeasy.net>
Paul Foley  <see@below.invalid> wrote:
+---------------
| [CMUCL used to return short, but that was fixed ages ago...I think]
+---------------

Actually, IIUIC, it never returned short for reads from plain files,
only from sockets or special files (such as TTYs), where even blocking
Unix "read()"s will return short if there's *some* data available
[returning zero only on end of file... usually]. But somewhere between
CMUCL-18e and CMUCL-19a the various low-level routines called by
READ-SEQUENCE changed from doing a single SYSTEM:READ-N-BYTES
[usually a wrapper around Unix "read()"] to having a loop around
SYSTEM:READ-N-BYTES that reads until the requested amount has been
seen or EOF, whichever comes first.


-Rob

-----
Rob Warnock			<rpw3@rpw3.org>
627 26th Avenue			<URL:http://rpw3.org/>
San Mateo, CA 94403		(650)572-2607