On Nov 30, 2007 10:51 AM, Vin Shelton &lt;<a href="mailto:acs@alumni.princeton.edu">acs@alumni.princeton.edu</a>&gt; wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Mike, thank you for upgrading XEmacs&#39;s JDE package. &nbsp;With a suitable<br>21.5 executable, I can again build all the XEmacs packages. &nbsp;I have<br>observed a problem, however. &nbsp;If I try to build all the XEmacs<br>packages using an XEmacs which has builtin gmp support (
i.e. one<br>configured --with-bignum=gmp), the &quot;make bytecompile&quot; step for the<br>semantic package will chew up memory until either I or the linux<br>kernel kills the XEmacs process. &nbsp;Here&#39;s the output from today&#39;s build
<br>at work (which I killed by hand):<br></blockquote></div><br clear="all">I echo the thanks to Mike.&nbsp; That&#39;s a big pile of work that we&#39;ve needed for some time.&nbsp; I&#39;m glad you&#39;ve stepped up to the plate to do it.
<br><br>I don&#39;t know what is going on off-hand.&nbsp; If there is code somewhere that relies on integer overflow to work correctly, we need to find and fix it.&nbsp; Is there any chance that you could get a C backtrace from an XEmacs that is busy chewing up memory?&nbsp; I probably won&#39;t have time to try reproducing this myself until early next week.
<br><br>Thanks,<br>-- <br>Jerry James<br><a href="http://loganjerry.googlepages.com/">http://loganjerry.googlepages.com/</a>