char(1) is padded with spaces with charset UTF8

[WarmBooter]

With a FB connection using charset UTF-8:

select cast('A' as char(1)) from rdb$database;

The result using ibase_fetch_object will be 'A ' ("A" padded with 3 spaces).

select cast('A' as varchar(1)) from rdb$database;

The result will be 'A', which is correct.

It seems that FB transfers the char(1) padded with spaces to fill the 4 bytes buffer, but for the end user, a char(1) should never return more than 1 character. IMHO, in such cases, the PHP driver should strip out any extra spaces before delivering the result to the final user in a way that the result would never has more characters than the declared field size.

[MartinKoeditz]

Yes, that's right. We should fix it.

[gedasys]

Is a solution already in sight?

[dab18]

pdo-firebird fixed with this: php/php-src#8576

[dab18]

--- ibase_query.c.orig  2025-03-21 14:18:17.753598142 +0400
+++ ibase_query.c       2025-03-21 14:20:48.012279537 +0400
@@ -947,6 +947,10 @@
        for (i = 0; i < sqlda->sqld; i++) {
                XSQLVAR *var = &sqlda->sqlvar[i];
 
+               if ((var->sqltype & ~1) == SQL_TEXT) {
+                       var->sqltype = SQL_VARYING | (var->sqltype & 1);
+               }
+
                switch (var->sqltype & ~1) {
                        case SQL_TEXT:
                                var->sqldata = safe_emalloc(sizeof(char), var->sqllen, 0);

tested on php: 5.5, 5.5, 7.1, 7.3
works like charm

[mlazdans]

Wouldn't that turn all CHAR types to VARCHAR?

[MartinKoeditz]

Well, basically yes. But it's only an internal conversion. I couldn't find any problems. I also created a test for that issue. Please check it out.