diff options
| author | James Cook <james@lindenlab.com> | 2010-06-01 11:06:16 -0700 |
|---|---|---|
| committer | James Cook <james@lindenlab.com> | 2010-06-01 11:06:16 -0700 |
| commit | 88b9808de6bf576807435c4c74126bcda523d0b8 (patch) | |
| tree | ee8612128c0c39a48ce3b06026346e31c96bd30f /indra/llmessage/llavatarnamecache.cpp | |
| parent | f9110055b97801249dfdad3e742539ded281b82f (diff) | |
DEV-50013 Don't add delays to batch up name requests
In practice, bulk name requests are all added in a single frame,
triggered for example by opening a group membership dialog. There's
no point to waiting to batch up more requests, and this slows down
the usual single-name lookup. Reviewed with Simon.
Diffstat (limited to 'indra/llmessage/llavatarnamecache.cpp')
| -rw-r--r-- | indra/llmessage/llavatarnamecache.cpp | 11 |
1 files changed, 6 insertions, 5 deletions
diff --git a/indra/llmessage/llavatarnamecache.cpp b/indra/llmessage/llavatarnamecache.cpp index 0c577053a3..912f34fcae 100644 --- a/indra/llmessage/llavatarnamecache.cpp +++ b/indra/llmessage/llavatarnamecache.cpp @@ -495,13 +495,14 @@ void LLAvatarNameCache::idle() // By convention, start running at first idle() call sRunning = true; + // *TODO: Possibly re-enabled this based on People API load measurements // 100 ms is the threshold for "user speed" operations, so we can // stall for about that long to batch up requests. - const F32 SECS_BETWEEN_REQUESTS = 0.1f; - if (!sRequestTimer.checkExpirationAndReset(SECS_BETWEEN_REQUESTS)) - { - return; - } + //const F32 SECS_BETWEEN_REQUESTS = 0.1f; + //if (!sRequestTimer.checkExpirationAndReset(SECS_BETWEEN_REQUESTS)) + //{ + // return; + //} // Must be large relative to above const F32 ERASE_EXPIRED_TIMEOUT = 60.f; // seconds |