Bug #68797 [Com]: Number 2.2250738585072012e-308 converted incorrectly

Edit report at https://bugs.php.net/bug.php?id=68797&edit=1 ID: 68797 Comment by: exploringbinary at gmail dot com Reported by: exploringbinary at gmail dot com Summary: Number 2.2250738585072012e-308 converted incorrectly Status: Closed Type: Bug Package: Math related Operating System: Windows PHP Version: 5.5.20 Assigned To: ab Block user comment: N Private report: N New Comment: I still couldn't find the description of the fix but I can tell from the new zend_strtod.c that the fix was to refresh it with the latest copy of David Gay's dtoa.c (with PHP-specific enhancements merged in). I have not tested it but plan to at some point. Thanks. Previous Comments: ------------------------------------------------------------------------ [2015-03-25 15:06:31] ab@php.net @exploringbinary, yes, the fix was put into master, you can easly check and clone it from github. It is also noted in the NEWS. A windows binary can be fetched from the link in my previous post or directly http://windows.php.net/downloads/snaps/master/ . Thanks. ------------------------------------------------------------------------ [2015-03-25 14:32:59] exploringbinary at gmail dot com Hi, Can you point me to the changed source code (zend_strtod.c presumably)? http://snaps.php.net/ is a dead link. I don't see this bug in any of the change logs. Can you provide a description of the fix as well? Thanks. ------------------------------------------------------------------------ [2015-03-17 23:58:41] ab@php.net The fix for this bug has been committed. Snapshots of the sources are packaged every three hours; this change will be in the next snapshot. You can grab the snapshot at http://snaps.php.net/. For Windows: http://windows.php.net/snapshots/ Thank you for the report, and for helping us make PHP better. Should be fixed in master now, please check. Thanks. ------------------------------------------------------------------------ [2015-01-12 16:39:28] ab@php.net I can reproduce this. But from what I see on the internets like here http://patrakov.blogspot.de/2009/03/dont-use-old-dtoac.html the version you point to can possibly have issues with gcc. Lets see if there's a commit history somewhere for that file. Thanks. ------------------------------------------------------------------------ [2015-01-11 00:53:10] exploringbinary at gmail dot com Description: ------------ The conversion of the number 2.2250738585072012e-308 (equivalent to DBL_MIN) to double-precision floating-point is done incorrectly; it is one ULP too low. It prints back as 2.2250738585072009E-308; it should be 2.2250738585072014e-308 (DBL_MIN). (The nearby decimal numbers 2.2250738585072013e-308, 2.2250738585072014e-308, 2.2250738585072015e-308, and 2.2250738585072016e-308 all convert correctly to DBL_MIN.) This is likely due to PHP's old copy of David Gay's dtoa.c; see these change entries at http://www.netlib.org/fp/changes -- they are probably what is needed: Wed Feb 12 00:40:01 EST 1997 dtoa.c: strtod: on IEEE systems, scale to avoid intermediate underflows when the result does not underflow; compiling with -DNO_IEEE_Scale restores the old logic. Fix a bug, revealed by input string 2.2250738585072012e-308, in treating input just less than the smallest normalized number. (The bug introduced an extra ULP of error in this special case.) Fri May 15 07:49:07 EDT 1998 dtoa.c: strtod: fix another glitch with scaling to avoid underflow with IEEE arithmetic, again revealed by the input string 2.2250738585072012e-308, which was rounded to the largest denormal rather than the smallest normal double precision number. Test script: --------------- <?php ini_set("precision", "17"); echo 2.2250738585072012e-308; ?> Expected result: ---------------- 2.2250738585072014e-308 Actual result: -------------- 2.2250738585072009E-308 ------------------------------------------------------------------------ -- Edit this bug report at https://bugs.php.net/bug.php?id=68797&edit=1