ID: 29292 User updated by: jmelville at selectaustralasia dot com dot au Reported By: jmelville at selectaustralasia dot com dot au Status: Bogus Bug Type: MSSQL related Operating System: Windows 2000 Server SP4 PHP Version: 4.3.8 New Comment:
> in the bottom line make sure php > use updated ntwdblib.dll file. Thanks momo. I've read #25777 and I understand that there are problems with PHP using older libraries to talk to SQL Server. I have ntwdblib.dll versions 7.00.839 (this is the one that ships with PHP including 4.3.9RC3) and 8.00.194 (which comes with SQL 2000) and the problem described in this bug shows up with both those libs. Is there an even newer version that I should be using? Previous Comments: ------------------------------------------------------------------------ [2004-09-21 11:21:48] [EMAIL PROTECTED] ilia introduced this bug on 1.86.2.25 fixing #25777. all the meaning was discussed there. in the bottom line make sure php use updated ntwdblib.dll file. ------------------------------------------------------------------------ [2004-09-17 17:30:20] dcrignon at adequasys dot com I confirm this problem, under Windows 2003 Server, with Apache 2 or 1.3 and PHP 4.3.8 or 5.0. The only way is to go back to PHP 4.3.3 ... Not the good solution for me. Do you know if someone is working on this bug to fix it? We can adapt our product and trim() any variable to recover good values, but is it the good solution?.. Thank you for your help! David ------------------------------------------------------------------------ [2004-08-30 17:41:45] vikinoha at yahoo dot com PHP 4.3.8 I am experiencing the same problem with non-empty string returned by a query that should return an empty string. I have tried to replace the php_mssql.dll file in version 4.3.8 with the one from 4.3.3, but the problem remains with no change whatsoever:( If there is someone familiar with PHP sources, does the php_mssql.c 1.138 revision changes deal with this bug, or it is something completely unrelated? And also, if the 'illia_0' change (pointed out by richard dot quadling at bandvulc dot co dot uk) effectively removes this bug (does it?), why is it then if'd out? Thanks, Viktor ------------------------------------------------------------------------ [2004-08-13 12:22:27] richard dot quadling at bandvulc dot co dot uk Hi. If the column contains NULL (using Enterprise Manager to enter CTRL+0), then the output is correct. But having to convert all '' to NULLs is not right. Richard. ------------------------------------------------------------------------ [2004-08-13 09:48:32] richard dot quadling at bandvulc dot co dot uk Hi. Following script ... <?php $rConn = mssql_connect('localhost','PHPBB_User','PHPBB_User'); $rResults = mssql_query('SELECT user_icq,LEN(user_icq) AS user_icq_len FROM PHPBB_User.phpbb_users'); while ($row = mssql_fetch_assoc($rResults)) { var_export($row); echo '<br />Length of user_icq = ' . strlen($row['user_icq']) . '<br /><br /><br />'; } mssql_free_result($rResults); mssql_close($rConn); ?> produces output of ... array ( 'user_icq' => ' ', 'user_icq_len' => 0, ) Length of user_icq = 1 array ( 'user_icq' => '1711757', 'user_icq_len' => 7, ) Length of user_icq = 7 array ( 'user_icq' => ' ', 'user_icq_len' => 0, ) Length of user_icq = 1 Which is clearly wrong! The length being returned by SQL is 0, the data being returned by PHP is ' '. The data via enterprise manager is ''. I'm using Sambar Server V6.1 Beta 3, PHP V5.0.0 (about to upgrade to V5.0.1). MS SQL 2000 SP3 Developer Edition all on a Windows XP Pro. But none of this seems to be making any difference. According to all the comments made, this was broken a LONG time ago. It seems that a small patch HAS been applied to the source (php_mssql.c line 793-797), but is under a compiler directive of ilia_0. (Is this Ilia Alshanetsky? If so, great article in PHP|Architect about contexts!!!). The code is in the php_mssql_get_column_content_with_type() function. ... case SQLTEXT: { int length; char *data = charcol(offset); length=dbdatlen(mssql_ptr->link,offset); #if ilia_0 while (length>0 && data[length-1] == ' ') { /* nuke trailing whitespace */ length--; } #endif ZVAL_STRINGL(result, data, length, 1); break; } ... The macro ZVAL_STRINGL is ... #define ZVAL_STRINGL(z, s, l, duplicate) { \ char *__s=(s); int __l=l; \ (z)->value.str.len = __l; \ (z)->value.str.val = (duplicate?estrndup(__s, __l):__s); \ (z)->type = IS_STRING; \ } If someone can recompile the code, can they do so WITHOUT the directive first but put in some debugs (forget thread safeness, only testing) and show what the value of length in the assignment from dbdatlen(mssql_ptr->link,offset) and what happens if the macro is called with a 0 rather than a 1. ------------------------------------------------------------------------ The remainder of the comments for this report are too long. To view the rest of the comments, please view the bug report online at http://bugs.php.net/29292 -- Edit this bug report at http://bugs.php.net/?id=29292&edit=1