Re: [PATCH] to improve DBD::Sybase 'make test' when using freetds

2010/4/11 David Dick <[email protected]>:
> This patch redirects some logging statements from stderr to the tdsdump file
> and tries to improve the results of running 'make test' from DBD::Sybase
> when using the freetds libraries.  It allows every DBD::Sybase test to pass
> except for t/xblob.t.
>

Well... patch left after some commits




diff -Naur old/locales.conf new/locales.conf
--- old/locales.conf	2008-02-29 09:30:19.000000000 +1100
+++ new/locales.conf	2010-04-11 20:22:08.000000000 +1000
@@ -1,5 +1,5 @@
 [default]
-	date format = %b %e %Y %I:%M:%S:%z%p
+	date format = %b %e %Y %I:%M

 [en_US]
 	date format = %b %e %Y %I:%M:%S:%z%p



So default english...



diff -Naur old/src/ctlib/cs.c new/src/ctlib/cs.c
--- old/src/ctlib/cs.c	2007-12-24 08:12:02.000000000 +1100
+++ new/src/ctlib/cs.c	2010-04-11 20:22:08.000000000 +1000
@@ -471,7 +471,14 @@
 		    src_type, src_len, desttype, destlen);

 	if (!is_fixed_type(desttype) && (destlen <= 0)) {
-		return CS_FAIL;
+		destlen = srcfmt ? srcfmt->maxlength : 0;
+		if (destlen <= 0) {
+			tdsdump_log(TDS_DBG_FUNC, "cs_convert() could not determine the
maxlength for the destination field\n");
+			return CS_FAIL;
+		}
+	}
+	if ((is_numeric_type(src_type)) && (desttype == SYBCHAR)) {
+		destlen += 1; /* maxlength does not take into account the decimal
point in the resulting string */
 	}

 	dest = (unsigned char *) destdata;




not tested.... documentation does not state anything on the subject....




diff -Naur old/src/tds/convert.c new/src/tds/convert.c
--- old/src/tds/convert.c	2008-02-04 18:55:39.000000000 +1100
+++ new/src/tds/convert.c	2010-04-11 20:22:08.000000000 +1000
@@ -2070,6 +2070,7 @@
 			old_digits_left = digits_left;
 			digits_left = cr->n.scale;
 			places = 1;
+		} else if (*pstr == 0) {
 		} else if (*pstr == ' ') {
 			for (; pstr != pend && *pstr == ' '; ++pstr) ; /* skip contiguous blanks */
 			if (pstr == pend)



this is quite strange and should cause a syntax error (conversion
fail) if a NUL character is found in the string.... Am I right ?
I though this was handled and tested by some of our tests but
according to coverage
(http://freetds.sourceforge.net/up/out83/coverage/tds/convert.c.gcov.html)
it isn't.... I would check other libraries... or perhaps we could
"fix" only CTLib.

freddy77
_______________________________________________
FreeTDS mailing list
[email protected]
http://lists.ibiblio.org/mailman/listinfo/freetds