Greetings! For now, I'd just go with the value that is too large, providing of course that you can control the size of your heap in lisp. We need to come up with a memory division schema that allows for an unbounded heap, but reserves known space for stack, (immediate fixnums?), etc. I.e., configure needs to be able to determine the stack start and direction, shared memory loading addresses, .text start, and whether any of these are changeable, and to then determine the optimal stack limits, immediate fixnum table limits and location, and heap limits if absolutely unavoidable. For example, on the 32bit standard Linux, we're stuck with a heap below 1G, so why not move text down to 0 and place the fixnum table above the shared memory and beneath the stack? Etc. As always, creative suggestions most welcome.
Take care, Robert Boyer <[EMAIL PROTECTED]> writes: > > I would multiply the MAXPAGES abouve by 7/8 or somewhat less. I hope to > > make this determination more robust soon. > > I just tried > ./configure --disable-statsysbfd --enable-locbfd --enable-maxpage=229376 > --disable-readline > followed by > make > > on a copy your GCLtwc files and got > > GCL (GNU Common Lisp) April 1994 229376 pages > MAXPAGE is not a power of 2. Recompile: Success > make[1]: *** [saved_pre_gcl] Error 1 > rm init_pre_gcl.lsp raw_pre_gcl > make[1]: Leaving directory > `/v/filer2/boyer/gcl/gcl-2.6.6twc/gcl-2.6.6twc/unixport' > make: *** [unixport/saved_pre_gcl] Error 2 > > Want to suggest any other number for MAXPAGE? I'm afraid that the power of > two lower than 262144 is the normal default 131072. > > Bob > > P. S. (* 7/8 262144) = 229376 > > > -- Camm Maguire [EMAIL PROTECTED] ========================================================================== "The earth is but one country, and mankind its citizens." -- Baha'u'llah _______________________________________________ Gcl-devel mailing list [email protected] http://lists.gnu.org/mailman/listinfo/gcl-devel
