From: William A. Rowe Jr <wrowe@apache.org>
Date: Thu, 13 Dec 2001 17:24:27 +0000 (+0000)
Subject:   Not really an XXX here
X-Git-Tag: 2.0.30~212
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=e844f598af4562a9441e5da5a4485f2b3f79e34b;p=thirdparty%2Fapache%2Fhttpd.git

  Not really an XXX here


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/trunk@92453 13f79535-47bb-0310-9956-ffa450edef68
---

diff --git a/modules/metadata/mod_cern_meta.c b/modules/metadata/mod_cern_meta.c
index 4bb69ec3bcb..e9ed8ab299e 100644
--- a/modules/metadata/mod_cern_meta.c
+++ b/modules/metadata/mod_cern_meta.c
@@ -361,10 +361,16 @@ static int add_cern_meta_data(request_rec *r)
 		 dconf->metasuffix ? dconf->metasuffix : DEFAULT_METASUFFIX,
 			   NULL);
 
-    /* XXX: it sucks to require this subrequest to complete, because this
+    /* It sucks to require this subrequest to complete, because this
      * means people must leave their meta files accessible to the world.
      * A better solution might be a "safe open" feature of pfopen to avoid
      * pipes, symlinks, and crap like that.
+     *
+     * In fact, this doesn't suck.  Because <Location > blocks are never run
+     * against sub_req_lookup_file, the meta can be somewhat protected by
+     * either masking it with a <Location > directive or alias, or stowing
+     * the file outside of the web document tree, while providing the
+     * appropriate directory blocks to allow access to it as a file.
      */
     rr = ap_sub_req_lookup_file(metafilename, r, NULL);
     if (rr->status != HTTP_OK) {