summaryrefslogtreecommitdiff
path: root/network_io
diff options
context:
space:
mode:
authortrawick <trawick@13f79535-47bb-0310-9956-ffa450edef68>2003-08-01 11:06:52 +0000
committertrawick <trawick@13f79535-47bb-0310-9956-ffa450edef68>2003-08-01 11:06:52 +0000
commita3d005c0ea4cd14bf97670bf5e9ce0fe10272814 (patch)
tree5378664923dec1341864f5b679e631f779e58fd2 /network_io
parentdc29c36d94aa2980cef1be5696664dac12e95870 (diff)
downloadlibapr-a3d005c0ea4cd14bf97670bf5e9ce0fe10272814.tar.gz
leave a comment about an interesting and possibly unexpected
behavior git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64576 13f79535-47bb-0310-9956-ffa450edef68
Diffstat (limited to 'network_io')
-rw-r--r--network_io/unix/sendrecv.c8
1 files changed, 8 insertions, 0 deletions
diff --git a/network_io/unix/sendrecv.c b/network_io/unix/sendrecv.c
index 46bbe7271..587774187 100644
--- a/network_io/unix/sendrecv.c
+++ b/network_io/unix/sendrecv.c
@@ -741,6 +741,14 @@ apr_status_t apr_socket_sendfile(apr_socket_t * sock, apr_file_t * file,
/* Ignore flags for now. */
flags = 0;
+ /* word to the wise: by default, AIX stores files sent by send_file()
+ * in the network buffer cache... there are supposedly scenarios
+ * where the most recent copy of the file won't be sent, but I can't
+ * recreate the potential problem, perhaps because of the way we
+ * use send_file()... if you suspect such a problem, try turning
+ * on the SF_SYNC_CACHE flag
+ */
+
/* AIX can also send the headers/footers as part of the system call */
parms.header_length = 0;
if (hdtr && hdtr->numheaders) {