On Fri, Oct 11, 2024 at 02:51:57AM -0700, karthik nayak wrote:

Patrick Steinhardt [off-list ref] writes:

quoted

@@ -1077,8 +1078,10 @@ static void t_reftable_stack_auto_compaction_with_locked_tables(void)
 	 * size, we expect that auto-compaction will want to compact all of the
 	 * tables. Locking any of the tables will keep it from doing so.
 	 */
-	strbuf_reset(&buf);

However here it is different, since we still use the strbuf. I guess it
should be okay, since 'buf' is initialized using 'STRBUF_INIT' and that
still keeps the buf.len to 0.

quoted

-	strbuf_addf(&buf, "%s/%s.lock", dir, st->readers[2]->name);
+	strbuf_addstr(&buf, dir);
+	strbuf_addstr(&buf, "/");
+	strbuf_addstr(&buf, st->readers[2]->name);
+	strbuf_addstr(&buf, ".lock");
 	write_file_buf(buf.buf, "", 0);

So when we do 'strbuf_addstr(&buf, ...)' it should allocate the required
memory. But the reset removal did catch my eye.

Yeah, I removed it while at it as it is completely unnecessary. I've
updated the commit message to point this out.

Patrick