IIRC it's due to dlopen holding the reference to the deleted file in memory until it's closed - it does not check with the filesystem again in the mean time. Windows will prevent the lib from being deleted. For *nix it's generally prevented by restricted access to the lib (and assuming people with higher privilege won't do such things). <div> <br> </div><div>Cheers,</div><div>yc</div><div><div><br><div class="gmail_quote">On Fri, Aug 6, 2010 at 8:22 AM, Noel Welsh <span dir="ltr"><<a href="mailto:noelwelsh@gmail.com" target="_blank">noelwelsh@gmail.com</a>></span> wrote:<br> <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> AFAIK it is unavoidable.<br> <br> N.<br> <div><div></div><div><br> On Fri, Aug 6, 2010 at 4:08 PM, John Clements <<a href="mailto:clements@brinckerhoff.org" target="_blank">clements@brinckerhoff.org</a>> wrote:<br> > It appears that the results of ffi-lib are persistent across evaluations. Specifically, if I use ffi-lib to load a library, then go to the filesystem and replace this library with garbage, subsequent evaluations still cheerfully use the version that I just destroyed.<br> ><br> > I'm guessing that this behavior is unavoidably inherited from the OS (this is mac OS X). If so, may I add a warning to the ffi-lib docs that states this? If this is not something unavoidable, of course, I'd much rather just change it.<br> ><br> > John<br> </div></div><div>_________________________________________________<br> For list-related administrative tasks:<br> </div> <a href="http://lists.racket-lang.org/listinfo/dev" target="_blank">http://lists.racket-lang.org/listinfo/dev</a><br> </blockquote></div><br><br clear="all"><br>-- <br>Cheers,<br>yc<br><br>Taming the software dragon - <a href="http://dragonmaestro.com" target="_blank">http://dragonmaestro.com</a><br><br> </div></div>