Raymond Toy pushed to branch issue-275-signal-underflow-in-float-reader at cmucl / cmucl Commits: 2034be96 by Raymond Toy at 2024-02-07T10:55:02-08:00 Update release notes. - - - - - 3b6a71be by Raymond Toy at 2024-02-07T11:03:49-08:00 Clean up impl Remove old code dealing with small numbers and refactor the code to a nicer form. - - - - - 2 changed files: - src/code/reader.lisp - src/general-info/release-21f.md Changes: ===================================== src/code/reader.lisp ===================================== @@ -1824,50 +1824,27 @@ the end of the stream." (t (error _"Internal error in floating point reader."))))) (defun make-float-aux (number divisor float-format stream) - (let ((ratio (/ number divisor)) - result) - (handler-case - (progn - (setf result (coerce ratio float-format)) - #+nil - (when (and (zerop result) (not (zerop number))) - ;; The number we've read is so small that it gets - ;; converted to 0.0, but is not actually zero. In this - ;; case, we want to round such small numbers to - ;; least-positive-foo-float. If it's still too small, we - ;; want to signal an error saying that we can't really - ;; convert it because the exponent is too small. - ;; See CLHS 2.3.1.1. - (let ((float-limit (ecase float-format - ((short-float single-float) - least-positive-single-float) - (double-float - least-positive-double-float) - #+double-double - (double-double-float - ext:least-positive-double-double-float)))) - (if (>= (* 2 ratio) float-limit) - (setf result float-limit) - (error _"Underflow")))) - result) - (floating-point-underflow (c) - (describe c) - ;; Got an underflow. Resignal it with the same - ;; operation/operands, but allowing a restart to set the value - ;; to 0. - (restart-case - (error 'floating-point-underflow - :operation (arithmetic-error-operation c) - :operands (arithmetic-error-operands c)) - (return-zero () - :report (lambda (stream) - (format stream _"Return ~A for ~A" - (coerce 0 float-format) - (read-buffer-to-string))) - (setf result (coerce 0 float-format))))) - (error () - (%reader-error stream _"Number not representable as a ~S: ~S" - float-format (read-buffer-to-string)))))) + (handler-case + (let ((ratio (/ number divisor))) + (coerce ratio float-format)) + (floating-point-underflow (c) + (describe c) + ;; Got an underflow. Resignal it with the same + ;; operation/operands, but allowing a restart to set the value + ;; to 0. + (restart-case + (error 'floating-point-underflow + :operation (arithmetic-error-operation c) + :operands (arithmetic-error-operands c)) + (return-zero () + :report (lambda (stream) + (format stream _"Return ~A for ~A" + (coerce 0 float-format) + (read-buffer-to-string))) + (setf result (coerce 0 float-format))))) + (error () + (%reader-error stream _"Number not representable as a ~S: ~S" + float-format (read-buffer-to-string)))))) (defun make-ratio (stream) ===================================== src/general-info/release-21f.md ===================================== @@ -40,6 +40,8 @@ public domain. * ~~#258~~ Remove `get-page-size` from linux-os.lisp * ~~#269~~ Add function to get user's home directory * ~~#266~~ Support "~user" in namestrings + * ~~#275~~ Handle floating point underflow in float reader + allowing user to flush the value to 0. * Other changes: * Improvements to the PCL implementation of CLOS: * Changes to building procedure: View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/compare/70586e47c6002ceef43fc05... -- View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/compare/70586e47c6002ceef43fc05... You're receiving this email because of your account on gitlab.common-lisp.net.