2007/7/27, Junio C Hamano <gitster@xxxxxxxxx>:
> Carlos Rica <jasampler@xxxxxxxxx> writes:
> > Signal SIGPIPE is ignored because the program sometimes was
> > terminated because that signal when writing the input for gpg.
>
> This "sometimes" does not give confidence to readers, I am
> afraid.
>
> What is happening is perfectly normal, not "sometimes it gets
> the signal mysteriously, so we need to paper it over by ignoring
> it".
>
> You invoke gpg, giving it a file that is supposed to contain a
> detached signature, and start feeding the payload that ought to
> verify Ok with the signature. If the tag is not signed, after
> gpg has read the detached signature, it already knows that the
> signature will not verify and it can exit without reading the
> payload from its standard input. When you try to write the
> payload to the pipe, you would get SIGPIPE.
By using "sometimes" I was requesting an answer to this strange
behaviour (for me) to you and the mailing list. Now I could reproduce it
removing the call to signal and running this:
#!/bin/sh
echo "creating vtag1..."
git tag -s -m "a signed tag" vtag1
echo "creating vtag2..."
git tag -m "an annotated non-signed tag" vtag2
echo "creating vtag3..."
git tag -m "another annotated non-signed tag is " vtag3
# 141 returned sometimes when gpg got
# a non-signed file as detached signature:
# (it refuses to read from the input when
# no signature is given, sometimes)
for i in 1 2 3 4 5 6 7 8 9 10
do
./git verify-tag vtag1 vtag2 vtag3
echo "git verify-tag vtag1 vtag2 vtag3 exit code: $?"
echo
done
git tag -d vtag1
git tag -d vtag2
git tag -d vtag3
In my system, some of the tests give 141 and others give 1 as exit code.
Dscho said that it could depend on the CPU current load of the computer,
since he got always 141 as you said, so perhaps it's me.
-
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
