[PHP-CVS-DAILY] cvs: php-src / ChangeLog
changelog Sat May 29 20:32:49 2004 EDT Modified files: /php-srcChangeLog Log: ChangeLog update http://cvs.php.net/diff.php/php-src/ChangeLog?r1=1.1642r2=1.1643ty=u Index: php-src/ChangeLog diff -u php-src/ChangeLog:1.1642 php-src/ChangeLog:1.1643 --- php-src/ChangeLog:1.1642Fri May 28 20:32:18 2004 +++ php-src/ChangeLog Sat May 29 20:32:48 2004 @@ -1,3 +1,77 @@ +2004-05-29 Wez Furlong [EMAIL PROTECTED] + +* ext/standard/proc_open.c: + Turn this off then + +2004-05-29 Greg Beaver [EMAIL PROTECTED] + +* pear/tests/PEAR_test_mock_pearweb.php.inc + pear/tests/pear_downloader_invalid.phpt + pear/tests/php_dump.php.inc: + modify php_dump.php, .cvsignore ignores *.php + move include to the place where it is used + +* pear/tests/pear_installer2.phpt: + update _installFile() test for channels + +* pear/tests/pear_installer1.phpt: + update checkDeps/extractDownloadFilename tests for channels + +* pear/tests/pear_registry.phpt: + update registry test for channels + +* pear/tests/pear_downloader_invalid.phpt + pear/tests/pear_downloader_new.phpt + pear/tests/pear_downloader_old.phpt: + use fake pearweb - now anyone can run the download unit tests. + + These tests require channels to pass + +* pear/tests/depnoreleases-1.0.tgz + pear/tests/depunstable-1.0.tgz + pear/tests/pkg1-1.1.tgz + pear/tests/pkg1-2.0b1.tgz + pear/tests/pkg2-1.1.tgz + pear/tests/pkg3-1.1.tgz + pear/tests/pkg3-1.4.tgz + pear/tests/pkg4-1.1.tgz + pear/tests/pkg4AndAHalf-1.3.tgz + pear/tests/pkg5-1.1.tgz + pear/tests/stabilitytoolow-0.3.tgz + pear/tests/stabilitytoolow-0.5.tgz + pear/tests/stabilitytoolow-0.6beta.tgz + pear/tests/stabilitytoolow-1.0b1.tgz + pear/tests/stabilitytoolow-2.0a1.tgz + pear/tests/stabilitytoolow-2.0b1.tgz + pear/tests/stabilitytoolow-2.0dev.tgz + pear/tests/stabilitytoolow-3.0dev.tgz: + fake pearweb package files, for unit testing downloads without an installed + pearweb + +* pear/tests/pear_config.phpt + pear/tests/pear_config_1.1.phpt: + config unit tests, ready for channels + +* pear/tests/PEAR_test_mock_pearweb.php.inc + pear/tests/download_test.config.inc + pear/tests/download_test_classes.php.inc + pear/tests/merge2.input + pear/tests/pear_channelfile.phpt + pear/tests/pear_installer_installFile_channels.phpt + pear/tests/pear_installer_install_channels.phpt + pear/tests/pear_registry_1.1.phpt + pear/tests/user3.input: + new unit tests + pearweb mock object is complete + +* pear/tests/pear_installer4.phpt: + update for conflicting file test + +2004-05-29 Wez Furlong [EMAIL PROTECTED] + +* ext/standard/proc_open.c: + Don't rely on TIOCNOTTY. + 2004-05-28 Ard Biesheuvel [EMAIL PROTECTED] * ext/interbase/interbase.c @@ -4189,7 +4263,7 @@ 2004-03-18 Pierre-Alain Joye [EMAIL PROTECTED] * ext/gd/tests/bug27582_2.phpt: - - Fix the test description and $Id: ChangeLog,v 1.1642 2004/05/29 00:32:18 changelog Exp $ + - Fix the test description and $Id: ChangeLog,v 1.1643 2004/05/30 00:32:48 changelog Exp $ 2004-03-18 Derick Rethans [EMAIL PROTECTED]
Re: [PHP-CVS] cvs: php-src /ext/standard proc_open.c
On Fri, 28 May 2004, Wez Furlong wrote: wez Fri May 28 09:25:51 2004 EDT Modified files: /php-src/ext/standard proc_open.c Log: Hopefully resolve proc_open build issues. If proc_open.c is compiled by default (it currently is), you are taking a huge risk here. To put it differently, Hopefully does not break the PHP 5 release (fingers crossed). #ifdef-based portability solutions are infinitely inferior to autoconf-checks, because they only mature by breaking the build for someone to initiate a feedback cycle. Doing this for an initial, widely anticipated release of a product is basically a guarantee for a lot of such feedback. The current platform check for linux, sun, and irix will catch a huge number of systems which either don't support pts or have not them enabled. Are there any compelling reasons why you would not want to do this properly? I also just had a quick look at the code. It assumes that if a C library contains code for grantpt etc, that the actual system also has a working /dev/ptmx. This assumption is, of course, completely unreliable. For example, the default Linux 2.4 kernel setting is pts disabled -- although glibc contains the support code you check for. Thus, the current code will fail on a default Linux kernel. With other words: The pts support of proc_open.c is anything but mature and should not be enabled for the initial PHP 5 release. - Sascha -- PHP CVS Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
[PHP-CVS] cvs: php-src /ext/standard proc_open.c
wez Sat May 29 05:27:18 2004 EDT Modified files: /php-src/ext/standard proc_open.c Log: Turn this off then http://cvs.php.net/diff.php/php-src/ext/standard/proc_open.c?r1=1.26r2=1.27ty=u Index: php-src/ext/standard/proc_open.c diff -u php-src/ext/standard/proc_open.c:1.26 php-src/ext/standard/proc_open.c:1.27 --- php-src/ext/standard/proc_open.c:1.26 Fri May 28 20:30:11 2004 +++ php-src/ext/standard/proc_open.cSat May 29 05:27:18 2004 @@ -15,9 +15,9 @@ | Author: Wez Furlong [EMAIL PROTECTED] | +--+ */ -/* $Id: proc_open.c,v 1.26 2004/05/29 00:30:11 wez Exp $ */ +/* $Id: proc_open.c,v 1.27 2004/05/29 09:27:18 wez Exp $ */ -#if defined(__linux__) || defined(sun) || defined(__IRIX__) +#if 0 (defined(__linux__) || defined(sun) || defined(__IRIX__)) # define _BSD_SOURCE /* linux wants this when XOPEN mode is on */ # define _BSD_COMPAT /* irix: uint */ # define _XOPEN_SOURCE 500 /* turn on Unix98 */ @@ -56,7 +56,7 @@ * */ #ifdef PHP_CAN_SUPPORT_PROC_OPEN -#if HAVE_PTSNAME HAVE_GRANTPT HAVE_UNLOCKPT HAVE_SYS_IOCTL_H HAVE_TERMIOS_H +#if 0 HAVE_PTSNAME HAVE_GRANTPT HAVE_UNLOCKPT HAVE_SYS_IOCTL_H HAVE_TERMIOS_H # include sys/ioctl.h # include termios.h # define PHP_CAN_DO_PTS1 -- PHP CVS Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php