use dprintf(1,..) instead of buffered stderr

[libc-test] / common / t.c

diff --git a/common/t.c b/common/t.c
index ecaaf7f..2373b09 100644 (file)
--- a/common/t.c
+++ b/common/t.c
@@ -29,10 +29,12 @@ void reset_timer() { errtimer(); }
 void error__(const char *n, int l, const char *s, ...) {
        va_list ap;
 
+       if (failed == 0 && nfailed == 0)
+               dprintf(1, "FAIL\n", n);
        failed = 1;
-       fprintf(stderr, " ERROR %s %s:%d: ", name, n, l);
+       dprintf(1, " ERROR %s %s:%d: ", name, n, l);
        va_start(ap, s);
-       vfprintf(stderr, s, ap);
+       vdprintf(1, s, ap);
        va_end(ap);
 }
 
@@ -44,7 +46,7 @@ static void run(const char *n, void (*f)()) {
        failed = 0;
        name = n;
        if (verbose)
-               fprintf(stderr, "running %s:\n", name);
+               dprintf(1, "running %s:\n", name);
 
        pid = fork();
        if (pid == 0) {
@@ -66,21 +68,21 @@ static void run(const char *n, void (*f)()) {
 
        if (failed) {
                nfailed++;
-               fprintf(stderr, "FAILED %s\n", name);
+               dprintf(1, "FAILED %s\n", name);
        } else if (verbose)
-               fprintf(stderr, "PASSED %s\n", name);
+               dprintf(1, "PASSED %s\n", name);
 }
 
 static int summary() {
        if (nfailed)
-               fprintf(stderr, "FAIL:%d (all: %d)\n", nfailed, count);
+               dprintf(1, "FAIL (%d out of %d tests)\n", nfailed, count);
        else
-               fprintf(stderr, "PASS (all: %d)\n", count);
+               dprintf(1, "ok (%d tests)\n", count);
        return !!nfailed;
 }
 
 static void usage() {
-       fprintf(stderr, "usage: ./t [-vs]\n");
+       dprintf(1, "usage: ./t [-vs]\n");
        exit(1);
 }