Web lists-archives.com

Re: [Fix v2] t5562: remove dependency on /dev/zero

Johannes Sixt <j6t@xxxxxxxx> writes:

> How many bytes are needed here? yes() in test-lib.sh generates only 99
> 'y', if I am not mistaken.

I think we will not use "yes" in the end for this topic, which makes
this comment totally irrelevant to the thread, but I wonder why we
have the limit of 99 there?  It cannot be "we do not want to worry
about sigpipe" affecting the end result of the test (after all the
reader may stop reading from after reading just one, and the status
of the upstream process that would die with sigpipe is lost anyway).

It turns out it is about sigpipe ;-) but in somewhat a different
way.  To prevent others from wasting their time wondering about
this, probably we want to have something like the attached?

 t/README      | 9 +++++++++
 t/test-lib.sh | 6 +++++-
 2 files changed, 14 insertions(+), 1 deletion(-)

diff --git a/t/README b/t/README
index 1326fd7505..f4e1a82657 100644
--- a/t/README
+++ b/t/README
@@ -927,6 +927,15 @@ library for your script to use.
    test_oid_init or test_oid_cache.  Providing an unknown key is an
+ - yes [<string>]
+   This is often seen in modern UNIX but some platforms lack it, so
+   the test harness overrides the platform implementation with a
+   more limited one.  Use this only when feeding a handful lines of
+   output to the downstream---unlike the real version, it generates
+   only up to 99 lines.
diff --git a/t/test-lib.sh b/t/test-lib.sh
index 42b1a0aa7f..541a37f4c0 100644
--- a/t/test-lib.sh
+++ b/t/test-lib.sh
@@ -1313,7 +1313,11 @@ then
-# Provide an implementation of the 'yes' utility
+# Provide an implementation of the 'yes' utility; the upper bound
+# limit is there to help Windows that cannot stop this loop from
+# wasting cycles when the downstream stops reading, so do not be
+# tempted to turn it into an infinite loop. cf. 6129c930 ("test-lib:
+# limit the output of the yes utility", 2016-02-02)
 yes () {
 	if test $# = 0