[Git][cmucl/cmucl][master] Update according to logs.
![](https://secure.gravatar.com/avatar/cc13150cabd87c26f35cb4b0ea78d66d.jpg?s=120&d=mm&r=g)
Raymond Toy pushed to branch master at cmucl / cmucl Commits: 95f2932b by Raymond Toy at 2016-01-07T17:37:42Z Update according to logs. - - - - - 1 changed file: - src/general-info/release-21b.txt Changes: ===================================== src/general-info/release-21b.txt ===================================== --- a/src/general-info/release-21b.txt +++ b/src/general-info/release-21b.txt @@ -44,18 +44,32 @@ New in this release: Thus, src/code/unix-glibc2.lisp is no longer used. * Micro-optimize modular shifts on x86. * Update lisp-unit to commit e6c259f. + * Added EXT:WITH-FLOAT-TRAPS-ENABLED to complement + WITH-FLOAT-TRAPS-MASKED. + * (EXPT 0 power) doesn't throw INTEXP-LIMIT-ERROR anymore for any + integer value of power. + * Starting cmucl with "-dyanmic-space-size 0" means using the + maximum possible heap size for the platform. * ANSI compliance fixes: + * PATHNAME-MATCH-P did not accept search-lists. * Bugfixes: * Linux was missing unix-setitimer which prevented saving cores. + * Generate inxact exceptions more carefully. + * Fix FP issue when building with Xcode 7.2 (and newer versions of + clang). (See ticket #12.) * Trac Tickets: * Gitlab tickets: * Ticket #10 fixed: setting an element of a 1, 2, or 4-bit array with a constant index did not always set the element - appropriately. + appropriately. + * Ticket #12 fixed. It looks like a possible compiler bug, but + worked around by explicitly setting inexact instead of using FP + instructions to generate inexact. + * Ticket #16 fixed: search-lists are handled correctly. * Other changes: View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/commit/95f2932bc350b3a89930f45acb...
participants (1)
-
Raymond Toy