Cache autocomplete results

app/assets/javascripts/gfm_auto_complete.js.coffee

@@ -4,7 +4,7 @@ window.GitLab ?= {}
 GitLab.GfmAutoComplete =
   dataLoading: false
   dataLoaded: false
-
+  cachedData: {}
   dataSource: ''
   # Emoji
@@ -55,7 +55,7 @@ GitLab.GfmAutoComplete =
     @setupAtWho()
     if @dataSource
-      if !@dataLoading
+      if not @dataLoading and not @cachedData
         @dataLoading = true
         # We should wait until initializations are done
@@ -70,6 +70,8 @@ GitLab.GfmAutoComplete =
             @loadData(data)
         , 1000)
+      if @cachedData?
+        @loadData(@cachedData)
   setupAtWho: ->
     # Emoji
@@ -205,6 +207,7 @@ GitLab.GfmAutoComplete =
     $.getJSON(dataSource)
   loadData: (data) ->
+    @cachedData = data
     @dataLoaded = true
     # load members

app/views/layouts/_init_auto_complete.html.haml

@@ -3,4 +3,5 @@
 - if @noteable
   :javascript
     GitLab.GfmAutoComplete.dataSource = "#{autocomplete_sources_namespace_project_path(project.namespace, project, type: @noteable.class, type_id: params[:id])}"
+    GitLab.GfmAutoComplete.cachedData = undefined;
     GitLab.GfmAutoComplete.setup();