Re: [PATCH] t9300-fast-import: don't hang if background fast-import exits too early

SZEDER Gábor [off-list ref] writes:

quoted

quoted

-	test_when_finished "
+	test_when_finished '
 		exec 8>&-; exec 9>&-;
-		kill $(cat V.pid) && wait $(cat V.pid)
-		true"
+		kill $(cat V.sh.pid) && wait $(cat V.sh.pid)
+		kill $(cat V.fi.pid) && wait $(cat V.sh.pid)
+		true'

The original interpolates the PID of the fast-import when
"when-finished" program is registered, so it is OK if somebody else
removed V.pid file; the new one interpolates when "when-finished"
program is run, reading from V.??.pid, so somebody needs to make
sure these pid files will stay around.  I do not think it is an
issue as I suspect we've left it to the global clean-up procedure
that removes the trash directory to remove the pid file.

In the original the same shell process starts 'git fast-import',
writes its pidfile, and registers the test_when_finished commands, so
we can be sure that the pid file is already present when the shell
runs the $(cat V.pid) command substitutions.

Yes.  It also means that V.pid file was not very useful in the
when-finished handler in the original.  We could have just used a
shell variable.

With this patch that's not the case anymore, because the background
subshell starts 'git fast-import' and writes the pidfile, but the main
shell process registers the test_when_finished commands.  IOW these
two shell processes are racing, and it's possible that the
test_when_finished command is executed before the background subshell
can write the pidfile.  So double quotes around the block of
test_when_finished commands are not good.

Oh, I was not questioning that.  I wanted to make sure, and I was
doing so aloud, that these files are

 (1) created way before, and 
 (2) left on the filesystem

when-finished handler actually runs, because the original did not
need any guarantee for (2), but now the sq version does.