Added using an own instance of ThreadPool
Using the global ThreadPool (getSharedOptimalPool()) can lead to
problems when more than one usage executes and one of them
already calls waitUntilEmpty() what of course influences the
other usage. Thus I added an own instance of ThreadPool for
async loading of chart models in writer
Change-Id: I4bea64af0d36e87081abec95c75574966d0fe5b9
diff --git a/sw/source/core/ole/ndole.cxx b/sw/source/core/ole/ndole.cxx
index bc8ba1a..a565f40 100644
--- a/sw/source/core/ole/ndole.cxx
+++ b/sw/source/core/ole/ndole.cxx
@@ -644,6 +644,38 @@
}
//////////////////////////////////////////////////////////////////////////////
+// due to some problems in test cases with the SharedOptimalPool, I have to
+// use an own iinstance of comphelper::ThreadPool. Prtoblem is that other
+// usages of getSharedOptimalPool() may interfere when more than one pool
+// user calls waitUntilEmpty().
+//
+// It gets created on-demand and will be available during LO's lifetime for
+// loading chart models used in writer in parallel.
+// It would be possible to add a usage count, then trigger
+// a timer and clean it up (due to lifetime issues), but that's unnecessarily
+// complicated. It gets created on demand, is ready for global reuse and makes
+// no harm (not much ressources needed)
+
+static comphelper::ThreadPool* pLocalPool = 0;
+
+comphelper::ThreadPool* getLocalThreadPool()
+{
+ if (pLocalPool)
+ {
+ return pLocalPool;
+ }
+
+ if (0 == comphelper::ThreadPool::getSharedOptimalPool().getWorkerCount())
+ {
+ return nullptr;
+ }
+
+ pLocalPool = new comphelper::ThreadPool(comphelper::ThreadPool::getSharedOptimalPool().getWorkerCount());
+ return pLocalPool;
+}
+
+//////////////////////////////////////////////////////////////////////////////
+// class holding local data for a parellel executed task to load a chart model
class DeflateData
{
@@ -703,6 +735,7 @@
};
//////////////////////////////////////////////////////////////////////////////
+// class defining a parellel executed task to load a chart model
class DeflateThread : public comphelper::ThreadTask
{
@@ -1048,13 +1081,16 @@
if(aXModel.is())
{
- // due to some problems in test cases with the SharedOptimalPool,
- // I need to deactivate this for now
- static bool bAnynchronousLoadingAllowed = false;
+ // added using own instance of comphelper::ThreadPool, see
+ // getLocalThreadPool(). I ran the UnitTests and the problems from
+ // before seem solved, so I optionally allow asynchronous loading
+ // for now. The static bool below is an anchor point to allow
+ // deactivating this feature quickly if trouble surfaces somewhere
+ static bool bAnynchronousLoadingAllowed = true;
if(bSynchron ||
!bAnynchronousLoadingAllowed ||
- 0 == comphelper::ThreadPool::getSharedOptimalPool().getWorkerCount())
+ nullptr == getLocalThreadPool())
{
// load chart synchron in this Thread
m_aPrimitive2DSequence = ChartHelper::tryToGetChartContentAsPrimitive2DSequence(
@@ -1070,7 +1106,7 @@
{
m_pDeflateData = new DeflateData(aXModel);
DeflateThread* pNew = new DeflateThread(*m_pDeflateData);
- comphelper::ThreadPool::getSharedOptimalPool().pushTask(pNew);
+ getLocalThreadPool()->pushTask(pNew);
}
}
}