Re: Setting max-cache-ttl to 0 does not disable caching

Hello,

Lucas Mulling wrote:
> A few users have questioned a change in behaviour from version 2.4.7 to
> 2.5.0 (commit 92de0387f04b1e87a4a49ed063323624f25ac3ef) where setting
> max-cache-ttl to 0 does not disable password caching anymore, is this
> change intended?

Sorry, it's not intended change.  I overlooked this use case of setting
max-cache-ttl to 0.

To disable caching, I think that you can do by:

	default-cache-ttl 0

Possible fix to recover the same semantics would be the following.

==========================
diff --git a/agent/cache.c b/agent/cache.c
index e8544205f..fbe1c1f14 100644
--- a/agent/cache.c
+++ b/agent/cache.c
@@ -318,6 +318,7 @@ compute_expiration (ITEM r)
   unsigned long maxttl;
   time_t current = gnupg_get_time ();
   time_t next;
+  int no_maxttl = 0;
 
   if (r->cache_mode == CACHE_MODE_PIN)
     return 0; /* Don't let it expire - scdaemon explicitly flushes them.  */
@@ -334,13 +335,16 @@ compute_expiration (ITEM r)
     {
     case CACHE_MODE_DATA:
     case CACHE_MODE_PIN:
-      maxttl = 0;  /* No MAX TTL here.  */
+      no_maxttl = 1;
+      /* No MAX TTL here.  */
       break;
     case CACHE_MODE_SSH: maxttl = opt.max_cache_ttl_ssh; break;
     default: maxttl = opt.max_cache_ttl; break;
     }
 
-  if (maxttl)
+  if (no_maxttl)
+    next = 0;
+  else
     {
       if (r->created + maxttl < current)
         {
@@ -351,8 +355,6 @@ compute_expiration (ITEM r)
 
       next = r->created + maxttl - current;
     }
-  else
-    next = 0;
 
   if (r->ttl >= 0 && (next == 0 || r->ttl < next))
     {
--