I've been reading MySQL 5.5 source code, and got confused by the macro HAVE_PSI_INTERFACE, which is appeared in many source file of the whole project.
For example, in the source file storage/example/ha_example.cc, there is the following code:
#ifdef HAVE_PSI_INTERFACE
static PSI_mutex_key ex_key_mutex_example, ex_key_mutex_EXAMPLE_SHARE_mutex;
static PSI_mutex_info all_example_mutexes[]=
{
{ &ex_key_mutex_example, "example", PSI_FLAG_GLOBAL},
{ &ex_key_mutex_EXAMPLE_SHARE_mutex, "EXAMPLE_SHARE::mutex", 0}
};
static void init_example_psi_keys()
{
const char* category= "example";
int count;
if (PSI_server == NULL)
return;
count= array_elements(all_example_mutexes);
PSI_server->register_mutex(category, all_example_mutexes, count);
}
#endif
So what does the HAVE_PSI_INTERFACE mean? Specifically, what does the PSI stand for? And what is the macro HAVE_PSI_INTERFACE used for?
Thanks.
PSI stands for: Performance schema instrumentation interface.
You can find a psi.h file here (with comments)
Related
im new to oracle agile plm can you solve me this.I refer Agile sdk guide but it not at all solving can you make this efficient search
In many documentations, you may not find code samples of every feature that there is. Instead, you are provided with one sample code which can be used with little modification for similar scenarios. In this case, you have code sample for reading the BOM table as below:
private void printBOM(IItem item, int level) throws APIException {
ITable bom = item.getTable(ItemConstants.TABLE_BOM);
Iterator i = bom.getReferentIterator();
while (i.hasNext()) {
IItem bomItem = (IItem)i.next();
System.out.print(indent(level));
System.out.println(bomItem.getName());
printBOM(bomItem, level + 1);
}
}
private String indent(int level) {
if (level <= 0) {
return "";
}
char c[] = new char[level*2];
Arrays.fill(c, ' ');
return new String(c);
}
All you need to do is change TABLE_BOM to TABLE_MANUFACTURERS, update relevant attribute names and fetch data from required cells.
Hope this helps.
Also, here's the link for latest documentation: SDK Developer Guide
I've set up a webserver running on ESP8266 thats currently hosting 7 sites. The sites is written in plain HTML in each diffrent tab in the arduino ide. I have installed the library Pagebuilder to help with making everything look nice and run.
Except one thing. I have a button connected to my ESP8266 which by the time being imitates a sensor input. basicly when the button is pressed my integer "x" increments with 1. I also managed to make a string that replicates "x" and increments with the same value.
I also have a problem with Printing the IPadresse of the server, but thats not as important as the other.
My plan then was writing the string "score" (which contains x) into the HTML tab where it should be output. this obviously didnt work.
Things I've tried:
Splitting up the HTML code where I want the string to be printed and using client.println("");
This didnt work because the two libraries does not cooperate and WiFiClient does not find Pagebuilders server. (basicly, the client.println does nothing when I used it with Pagebuilder).
Reconstructing the HTML page as a literal really long string, and adding in the String with x like this: "html"+score+"html" and adding it into where the HTML page const char were. (basicly replacing the variable with the text that were in the variable).
This did neighter work because the argument "PageElement" from Pagebuilder does only expect one string, and errors out because theres an additional string inside the HTML string.
I've tried sending it as a post req. but this did not output the value either.
I have run out of Ideas to try.
//root page
#if defined(ARDUINO_ARCH_ESP8266)
#include <ESP8266WiFi.h>
#include <ESP8266WebServer.h>
#include <WiFiClient.h>
#elif defined(ARDUINO_ARCH_ESP32)
#include <WiFi.h>
#include <WebServer.h>
#endif
#include "PageBuilder.h"
#include "currentGame.h" //tab 1
#if defined(ARDUINO_ARCH_ESP8266)
ESP8266WebServer Server;
ESP8266WebServer server;
#endif
int sensorPin = 2; // button input
int sensorValue = 0;
int x = 0; // the int x
String score=""; //the string x will be in
PageElement CURRENT_GAME_ELEMENT(htmlPage1);
PageBuilder CURRENT_GAME("/current-game", {CURRENT_GAME_ELEMENT}); // this //only showes on href /current-game
void button() {
sensorValue = analogRead(sensorPin); //read the voltage
score="Team 1: "+String((int)x+1); //"make" x a string
if (sensorValue <= 10) { // check if button is pressed
x++; // increment x
Serial.println(x);
Serial.println(score);
delay(100);
}
}
void setup() {
Serial.begin(115200);
pinMode(2, INPUT);
WiFi.softAP("SSID", "PASS");
delay(100);
CURRENT_GAME.insert(Server);
Server.begin();
}
void loop() {
Server.handleClient();
button();
}
// tab 1
const char htmlPage1[] PROGMEM = R"=====(
/*
alot of HTML, basicly the whole website...
..............................................
*/
<div class="jumbotron">
<div align="center">
<h1 class="display-4"> score </h1> // <--- this is where
//I want to print the
//string:
</div>
</div>
)=====";
what I want to do is getting the value of the string score displayed on the website. If I put "score" directly into the HTML, the word score will be displayed, not the value. I want the value displayed.
Edit:
I have figured out how to make the string(score) be printed in the HTML code, thus, I only have to convert the HTML code string back to a char. explanation is in comment below.
Edit 2: (-------------------------solution-------------------------)
Many thanks for the help I've gotten and sorry for being so ignorant, its just so hard being so close and that thing doesnt work. but anyways, What I did was following Pagebuilders example, and making another element to print in current game..
String test(PageArgument& args) {
return score;
}
const char html[] = "<div class=\"jumbotron\"><div align=\"center\"><h1 class=\"display-4\">{{NAME}}</h1></div></div>";
PageElement FRAMEWORK_PAGE_ELEMENT(htmlPage0);
PageBuilder FRAMEWORK_PAGE("/", {FRAMEWORK_PAGE_ELEMENT});
PageElement body_elem(html, { {"NAME", test} });
PageElement CURRENT_GAME_ELEMENT(htmlPage1);
PageBuilder CURRENT_GAME("/current-game", { CURRENT_GAME_ELEMENT, body_elem});
suprisingly easy when I first understood it.. Thanks again.
You could try building your string first, then converting it to a const char
like this: const char * c = str.c_str();
if you can't use a pointer you could try this:
string s = "yourHTML" + score + "moreHTML";
int n = s.length();
char char_array[n + 1];
strcpy(char_array, s.c_str());
additionally you could try the stringstream standard library
This sort of thing is often done using magic tags in your markup that are detected by the server code before it serves the HTML and filled in by executing some sort of callback or filling in a variable, or whatever.
So with this in mind and hoping for the best, I nipped over to: PageBuilder on github and looked to see if there was something similar here. Good news! In one of the examples:
const char html[] = "hello <b>{{NAME}}</b>, <br>Good {{DAYTIME}}.";
...
PageElement body_elem(html, { {"NAME", AsName}, {"DAYTIME", AsDayTime} });
Where {{NAME}} and {{DAYTIME}} are magic tokens. AsName and AsDayTime are functions to be called when the respective tag is encountered while the page is being served.
EDIT: in response to a request to explain differently, I'm not convinced I can do a better job of explaining the code than the example on the library's own github page, so I'll try a wordy description instead:
When you want to serve a webpage to a client, the code needs to know what you want to serve. In the simplest case, it's a static page: the same every time. You can just write the HTML, stick it in a string an be done.
whole_page = "<html>My fixed content</html>";
webserver.serve(whole_page);
But you want some dynamic element(s). As noted, you can do it in a few ways, such as serving some static HTML, then the dynamic bit, then some more static HTML. It seems you've not had much luck like this, and it's rather clunky anyway.
Or you can pre-build a new string out of the three bits and serve that in one chunk, but that's also pretty clunky.
(Aside: taking big strings and adding them together is likely to be slow and memory intensive, two things you really don't want on a little CPU like the ESP8266).
So instead, you allow 'magic' markers in the HTML, using a marker in place of the dynamic content, and serve that instead.
whole_page = "<html>My dynamic content. Value is {{my_value}}</html>";
webserver.serve(whole_page, ...);
The clever bit is that as the page is being served, the webserver is watching the text go by, and when it sees a magic tag, it stops and asks you to fill in the blank, then carries on as before.
Obviously, there is some processing overhead with watching for tags, and some programming overhead with telling it what tags to watch for and how to ask you for the data it needs.
I got advice from a friend who told me I should make a unique argument where I wanted the string(x) and then using some syntax to replace it. I also took inspiration from you Jelle..
what I did was make a unique argument "VAR_CURRENT_SCORE" put that into the HTML where I want the score output, then convert htmlPage1 from a char to a string, use string.replace() and replace "VAR_CURRENT_SCORE" with the string(x) score. this workes as I can see in the serial monitor output.
This is what I did:
//root page
String HTMLstring(htmlstringPage);
delay(100);
HTMLstring.replace("VAR_CURRENT_SCORE", score);
delay(50);
Serial.println("string:");
Serial.println(HTMLstring);
//tab 1 char htmlstringPage[] PROGMEM = R"=====(
<div class="jumbotron">
<div align="center">
<h1 class="display-4">VAR_CURRENT_SCORE</h1>
</div>
</div>
)=====";
However, I still have a small problem left which is converting the string back to char to post it to the website.
To convert the string back:
request->send_P(200, "text/html", HTMLstring.c_str());
I have used a fifo pipe to read in some data (weather data) into a char variable. The console will display this variable correctly. However, when I try to display it through HTML on the CGI page, it simply does not display. Code below -
int main(void) {
int fd;
char *myfifo = "pressure.txt";
char buff[BUFFER];
long fTemp;
//open and read message
fd = open(myfifo, O_RDONLY);
read(fd, buff, BUFFER);
printf("Received: %s\n", buff);
close(fd);
printf("Content-type: text/html\n\n");
puts("<HTML>");
puts("<BODY>");
printf("Data is: %s", buff);
puts("</BODY>");
puts("</HTML>");
return EXIT_SUCCESS;
}
As you can see in the console is displays correctly -
Received: 2014-08-13 16:54:57
25.0 DegC, 1018.7 mBar
Content-type: text/html
<HTML>
<BODY>
Data is 2014-08-13 16:54:57
25.0 DegC, 1018.7 mBar
</BODY>
</HTML>
logout
But on the CGI webpage it does not display the weather data, but it does display "data is".
Two important things when writing a CGI program:
the program will be run by the webserver, which is normally
started as a different user (the 'www' user for example).
it's possible that the program is started from within another
directory, which can cause different behaviour if you don't
specify the full path of a file you want to open.
Since both these things can cause problems, it can be helpful
to add some debug information. Of course, it's always a good idea
to check return values of functions you use.
To make it easier to display debug or error messages, I'd first
move the following code up, so that all output that comes after
it will be rendered by the browser:
printf("Content-type: text/html\r\n\r\n");
puts("<HTML>");
puts("<BODY>");
It may be useful to know what the webserver uses as the directory
from which the program is started. The getcwd
call can help here. Let's use a buffer of size BUFFER to store
the result in, and check if it worked:
char curpath[BUFFER];
if (getcwd(curpath, BUFFER) == NULL)
printf("Can't get current path: %s<BR>\n", strerror(errno));
else
printf("Current path is: %s<BR>\n", curpath);
The getcwd function returns NULL in case of an error, and sets the value
of errno to a number which indicates what went wrong. To convert this
value to something readable, the strerror
function is used. For example, if BUFFER was not large enough to be
able to store the path, you'll see something like
Can't get current path: Numerical result out of range
The open call returns a negative number
if it didn't work, and sets errno again. So, to check if this worked:
fd = open(myfifo, O_RDONLY);
if (fd < 0)
printf("Can't open file: %s<BR>\n", strerror(errno));
In case the file can be found, but the webserver does not have permission
to open it, you'll see
Can't open file: Permission denied
If the program is started from another directory than you think, and
it's unable to locate the file, you would get:
Can't open file: No such file or directory
Adding such debug info should make it more clear what's going on, and more
importantly, what's going wrong.
To make sure the actual data is read without problems as well, the return
value of the read function should be
checked and appropriate actions should be taken. If read fails,
a negative number is returned. To handle this:
numread = read(fd, buff, BUFFER);
if (numread < 0)
printf("Error reading from file: %s<BR>\n", strerror(errno));
Another value indicates success, and returns the number of bytes that were
read. If really BUFFER bytes were read, it's not at all certain that the
last byte in buff is a 0, which is needed for printf to know when the
string ended. To make sure it is in fact null-terminated, the last byte in
buff is set to 0:
if (numread == BUFFER)
buff[BUFFER-1] = 0;
Note that this actually overwrites one of the bytes that were read in this
case.
If fewer bytes were read, it's still not certain that the last byte that was
read was a 0, but now we can place our own 0 after the bytes that were read
so none of them are overwritten:
else
buff[numread] = 0;
To make everything work, you may need the following additional include files:
#include <unistd.h>
#include <string.h>
#include <errno.h>
The complete code of what I described is shown below:
int main(void)
{
int fd, numread;
char *myfifo = "pressure.txt";
char buff[BUFFER];
char curpath[BUFFER];
long fTemp;
// Let's make sure all text output (even error/debug messages)
// will be visible in the web page
printf("Content-type: text/html\r\n\r\n");
puts("<HTML>");
puts("<BODY>");
// Some debug info: print the current path
if (getcwd(curpath, BUFFER) == NULL)
printf("Can't get current path: %s<BR>\n", strerror(errno));
else
printf("Current path is: %s<BR>\n", curpath);
// Open the file
fd = open(myfifo, O_RDONLY);
if (fd < 0)
{
// An error occurs, let's see what it is
printf("Can't open file: %s<BR>\n", strerror(errno));
}
else
{
// Try to read 'BUFFER' bytes from the file
numread = read(fd, buff, BUFFER);
if (numread < 0)
{
printf("Error reading from file: %s<BR>\n", strerror(errno));
}
else
{
if (numread == BUFFER)
{
// Make sure the last byte in 'buff' is 0, so that the
// string is null-terminated
buff[BUFFER-1] = 0;
}
else
{
// Fewer bytes were read, make sure a 0 is placed after
// them
buff[numread] = 0;
}
printf("Data is: %s<BR>\n", buff);
}
close(fd);
}
puts("</BODY>");
puts("</HTML>");
return EXIT_SUCCESS;
}
In Windows Store Apps, C++(C# is similar though), doing something like
IAsyncAction^ Action = CurrentAppSimulator::ReloadSimulatorAsync(proxyFile);
create_task( Action ).then([this]()
{
}.wait();
results in an unhandled exception. Usually it's
Microsoft C++ exception: Concurrency::invalid_operation at memory location 0x0531eb58
And I kind of need for that action to finish to get my In App Purchase information before trying to use it.
The weird thing here is that anything else besides IAsyncAction waits just fine. IAsyncOperation and IAsyncOperationWithProgress worked just fine, but this ? Exception and then crash.
To be honest, I have no idea what's the difference between an IAsyncOperation and an IAsyncAction, they seem similar to me.
UPDATE :
By analyzing this page http://msdn.microsoft.com/en-us/library/vstudio/hh750082.aspx you can figure out that IAsyncAction is just an IAsyncOperation without a return type. But, you can then see that most IAsyncAction-s are waitable. The real problem though is that certain Windows functions just want to execute on a particular thread (for some reason). ReloadSimulatorAsync is one such fine example.
Using code like this :
void WaitForAsync( IAsyncAction ^A )
{
while(A->Status == AsyncStatus::Started)
{
std::chrono::milliseconds milis( 1 );
std::this_thread::sleep_for( milis );
}
AsyncStatus S = A->Status;
}
results in an infinite loop. If called upon other functions it actually works. The problem here is why does a task need to be executed on a particular thread if everything is Async ? Instead of Async it should be RunOn(Main/UI)Thread or similar.
SOLVED, see answer;
Calling wait on the concurrency::task after you create it completely defeats the point of having tasks in the first place.
What you have to realize is that in the Windows Runtime, there are many asynchronous operations that cannot (or should not) be run on (or waited for on) the UI thread; you've found one of them, and now you're trying to wait on it. Instead of potentially causing a deadlock, you're getting an exception.
To remedy this, you need to use a continuation. You're most of the way there; you're already defining a continuation function:
IAsyncAction^ Action = CurrentAppSimulator::ReloadSimulatorAsync(proxyFile);
create_task( Action ).then([this]()
{
}).wait();
// do important things once the simulator has reloaded
important_things();
...but you're not using it. The purpose of the function you pass into then is to be called off the UI thread once the task is complete. So, instead, you should do this:
IAsyncAction^ Action = CurrentAppSimulator::ReloadSimulatorAsync(proxyFile);
create_task( Action ).then([this]()
{
// do important things once the simulator has reloaded
important_things();
});
Your important post-reload code won't run until the task is complete, and it will run on a background thread so it doesn't block or deadlock the UI.
This is the magical fix that gets the job done :
void WaitForAsync( IAsyncAction ^A )
{
while(A->Status == Windows::Foundation::AsyncStatus::Started)
{
CoreWindow::GetForCurrentThread()->Dispatcher->ProcessEvents(CoreProcessEventsOption::ProcessAllIfPresent);
}
Windows::Foundation::AsyncStatus S = A->Status;
}
In general you should use the continuations (.then(...)), like Adam's answer says, and not block. But lets say you want to do a wait for some reason (for testing some code?), you can trigger an event from the last continuation (to use C# parlance):
TEST_METHOD(AsyncOnThreadPoolUsingEvent)
{
std::shared_ptr<Concurrency::event> _completed = std::make_shared<Concurrency::event>();
int i;
auto workItem = ref new WorkItemHandler(
[_completed, &i](Windows::Foundation::IAsyncAction^ workItem)
{
Windows::Storage::StorageFolder^ _picturesLibrary = Windows::Storage::KnownFolders::PicturesLibrary;
Concurrency::task<Windows::Storage::StorageFile^> _getFileObjectTask(_picturesLibrary->GetFileAsync(L"art.bmp"));
auto _task2 = _getFileObjectTask.then([_completed, &i](Windows::Storage::StorageFile^ file)
{
i = 90210;
_completed->set();
});
});
auto asyncAction = ThreadPool::RunAsync(workItem);
_completed->wait();
int j = i;
}
By the way, for some reason this method causes an exception when after it is over in visual studio tests. I've tested it in an app too though and it worked with no problem. I'm not quite sure what the problem is with the test.
If anybody wants a C++/Wrl example then I have that too.
Update 07/08/2017: As requested here is a C++/Wrl example. I have just run this in a Universal Windows (10) Test project in Visual Studio 2017. The key thing here is the weird part Callback<Implements< RuntimeClassFlags<ClassicCom >, IWorkItemHandler , FtmBase >> , as opposed to just Callback<IWorkItemHandler> . When I had the latter, the program jammmed except for when it was in a .exe project. I found this solution here: https://social.msdn.microsoft.com/Forums/windowsapps/en-US/ef6f84f6-ad4d-44f0-a107-3922d56662e6/thread-pool-task-blocking-ui-thread . See "agile objects" for more information.
#include "pch.h"
#include "CppUnitTest.h"
#include <Windows.Foundation.h>
#include <wrl\wrappers\corewrappers.h>
#include <wrl\client.h>
#include <wrl/event.h>
#include <memory>
#include "concrt.h"
#include <Windows.System.Threading.h>
using namespace Microsoft::VisualStudio::CppUnitTestFramework;
using namespace ABI::Windows::Foundation;
using namespace Microsoft::WRL;
using namespace Microsoft::WRL::Wrappers;
using namespace Windows::System::Threading;
using namespace ABI::Windows::Foundation;
using namespace ABI::Windows::System::Threading;
namespace TestWinRtAsync10
{
TEST_CLASS(UnitTest1)
{
public:
TEST_METHOD(AsyncOnThreadPoolUsingEvent10Wrl)
{
HRESULT hr = BasicThreadpoolTestWithAgileCallback();
Assert::AreEqual(hr, S_OK);
}
HRESULT BasicThreadpoolTestWithAgileCallback()
{
std::shared_ptr<Concurrency::event> _completed = std::make_shared<Concurrency::event>();
ComPtr<ABI::Windows::System::Threading::IThreadPoolStatics> _threadPool;
HRESULT hr = GetActivationFactory(HStringReference(RuntimeClass_Windows_System_Threading_ThreadPool).Get(), &_threadPool);
ComPtr<IAsyncAction> asyncAction;
hr = _threadPool->RunAsync(Callback<Implements<RuntimeClassFlags<ClassicCom>, IWorkItemHandler, FtmBase>>([&_completed](IAsyncAction* asyncAction) -> HRESULT
{
// Prints a message in debug run of this test
std::ostringstream ss;
ss << "Threadpool work item running.\n";
std::string _string = ss.str();
std::wstring stemp = std::wstring(_string.begin(), _string.end());
OutputDebugString(stemp.c_str());
//
_completed->set();
return S_OK;
}).Get(), &asyncAction);
_completed->wait();
return S_OK;
}
};
}
Update 08/08/2017: More example, per the comments.
#include "pch.h"
#include "CppUnitTest.h"
#include <wrl\wrappers\corewrappers.h>
#include <wrl\client.h>
#include <wrl/event.h>
#include <memory>
#include "concrt.h"
#include <Windows.System.Threading.h>
#include <Windows.ApplicationModel.Core.h>
using namespace ABI::Windows::Foundation;
using namespace Microsoft::WRL;
namespace TestWinRtAsync10
{
TEST_CLASS(TestWinRtAsync_WrlAsyncTesting)
{
public:
TEST_METHOD(PackageClassTest)
{
ComPtr<ABI::Windows::ApplicationModel::IPackageStatics> _pPackageStatics;
HRESULT hr = GetActivationFactory(Microsoft::WRL::Wrappers::HStringReference(RuntimeClass_Windows_ApplicationModel_Package).Get(), &_pPackageStatics);
ComPtr<ABI::Windows::ApplicationModel::IPackage> _pIPackage;
hr = _pPackageStatics->get_Current(&_pIPackage);
ComPtr<ABI::Windows::ApplicationModel::IPackage3> _pIPackage3;
hr = _pIPackage->QueryInterface(__uuidof(ABI::Windows::ApplicationModel::IPackage3), &_pIPackage3);
ComPtr<__FIAsyncOperation_1___FIVectorView_1_Windows__CApplicationModel__CCore__CAppListEntry> _pAsyncOperation;
hr = _pIPackage3->GetAppListEntriesAsync(&_pAsyncOperation);
std::shared_ptr<Concurrency::event> _completed = std::make_shared<Concurrency::event>();
_pAsyncOperation->put_Completed(Microsoft::WRL::Callback<Implements<RuntimeClassFlags<ClassicCom>, ABI::Windows::Foundation::IAsyncOperationCompletedHandler <__FIVectorView_1_Windows__CApplicationModel__CCore__CAppListEntry*>, FtmBase >>
([&_completed](ABI::Windows::Foundation::IAsyncOperation<__FIVectorView_1_Windows__CApplicationModel__CCore__CAppListEntry*>* pHandler, AsyncStatus status)
{
__FIVectorView_1_Windows__CApplicationModel__CCore__CAppListEntry* _pResults = nullptr;
HRESULT hr = pHandler->GetResults(&_pResults);
ComPtr<ABI::Windows::ApplicationModel::Core::IAppListEntry> _pIAppListEntry;
unsigned int _actual;
hr = _pResults->GetMany(0, 1, &_pIAppListEntry, &_actual);
ComPtr<ABI::Windows::ApplicationModel::IAppDisplayInfo> _pDisplayInfo;
hr = _pIAppListEntry->get_DisplayInfo(&_pDisplayInfo);
Microsoft::WRL::Wrappers::HString _HStrDisplayName;
hr = _pDisplayInfo->get_DisplayName(_HStrDisplayName.GetAddressOf());
const wchar_t * _pWchar_displayName = _HStrDisplayName.GetRawBuffer(&_actual);
OutputDebugString(_pWchar_displayName);
_completed->set();
return hr;
}).Get());
_completed->wait();
};
};
}
This outputted:
TestWinRtAsync10
Just in case anyone needs here is solution in C++/WinRT. Say you have function ProcessFeedAsync() that would return IAsyncAction, just need following simple code:
winrt::init_apartment();
auto processOp{ ProcessFeedAsync() };
// do other work while the feed is being printed.
processOp.get(); // no more work to do; call get() so that we see the printout before the application exits.
source
I use the following code for Text-To-Speech application controls for blind persons in C++ Builder (most likely similar example can be used in Delphi). Main form has KeyPreview property checked to enable key F11 preview to start speaking active (focused) control. The code as it is works but there are some problems. This example is in C++ Builder code but from what I've found, Delphi suffers from same problem and the solution I found is the same. If you have Delphi solution, feel free to post it, it is similar anyway.
#include <sapi.h>
#include <WTypes.h>
//---------------------------------------------------------------------------
// Speak text string (synchronous function)
//---------------------------------------------------------------------------
bool SpeakText(UnicodeString Text)
{
ISpVoice* pVoice = NULL;
if (FAILED(::CoInitialize(NULL))) return false;
Word Saved8087CW = Default8087CW; // Disable floating point division by zero exception caused by Speak
Set8087CW(0x133f);
HRESULT hr = CoCreateInstance(CLSID_SpVoice, NULL, CLSCTX_ALL, IID_ISpVoice, (void **)&pVoice);
if (SUCCEEDED(hr))
{
//pVoice->SpeakCompleteEvent()
//pVoice->SetSyncSpeakTimeout(1000);
hr = pVoice->Speak(WideString(Text).c_bstr(), SPF_DEFAULT, NULL);
pVoice->Release();
pVoice = NULL;
}
Set8087CW(Saved8087CW);
::CoUninitialize();
return true;
}
//---------------------------------------------------------------------------
void __fastcall TForm1::FormKeyUp(TObject *Sender, WORD &Key, TShiftState Shift)
{
UnicodeString Speaker;
if (Key == VK_F11)
{
if (Screen->ActiveControl->InheritsFrom(__classid(TButton))) { Speaker += "Button, " + static_cast<TButton*>(Screen->ActiveControl)->Caption + "."; }
else if (Screen->ActiveControl->InheritsFrom(__classid(TEdit))) { Speaker += "Edit box, " + static_cast<TEdit*>(Screen->ActiveControl)->Text + "."; }
}
if (Speaker != "") SpeakText(Speaker);
}
//---------------------------------------------------------------------------
Problems:
pVoice->Speak causes Floating point division by zero if I don't override the exception using the Set8087CW function. This happens only on Windows 7 (possibly Vista and Windows 8 too) but not on Windows XP in the same program (compiled exe). Is there a solution without using Set8087CW? Removing these lines will cause the problem and exception. I have BCB2010.
Function is synchronous and won't shut up or return control to program until it finishes reading text. This is a problem for longer text. It also blocks program events. Is there a way to make it asynchronous or introduce an event to periodically check for F11 key status and if F11 is pressed again it stops reading and uninitializes object? For example poll every 300 ms (or after each word etc.) for key-press F11 and if pressed, stop speaking? Or run it threaded?
Does SAPI has memory leaks as some write on various sites?
Can above code use OleCheck instead of CoCreateInstance and CoUninitialize?
UPDATE for those looking for solution as suggested by Remy Lebeau:
SavedCW = Get8087CW();
Set8087CW(SavedCW | 0x4);
hr = pVoice->Speak(WideString(Text).c_bstr(), SPF_DEFAULT | SPF_ASYNC, NULL);
pVoice->WaitUntilDone(-1); // Waits until text is done... if F11 is pressed simply go out of scope and speech will stop
Set8087CW(SavedCW);
Also found detailed example in CodeRage 4 session: http://cc.embarcadero.com/item/27264
The error does occur in Vista as well. Masking floating point exceptions is the only solution.
To make Speak() run asynchronously, you need to include the SPF_ASYNC flag when calling it. If you need to detect when asynchronous speaking is finished, you can use ISpVoice::WaitUntilDone(), or call ISpVoice::SpeakCompleteEvent() and pass the returned HANDLE to one of the WaitFor...() family of functions, like WaitForSingleObject().
What kind of leaks do other sites talk about?
Not instead of, no. OleCheck() merely checks the value of an HRESULT value and throws an exception if it is an error value. You still have to call COM functions that return the actual HRESULT values in the first place. If anything, OleCheck() would be a replacement for SUCCEEDED() instead.
For what you are attempting, I would suggest the following approach instead:
struct s8087CW
{
Word Saved8087CW;
s8087CW(Word NewCW)
{
Saved8087CW = Default8087CW;
Set8087CW(NewCW);
// alternatively, the VCL documentation says to use SetExceptionMask() instead of Set8087CW() directly...
}
~s8087CW()
{
Set8087CW(Saved8087CW);
}
};
//---------------------------------------------------------------------------
__fastcall TForm1::TForm1(TComponent *Owner)
: TForm(Owner)
{
::CoInitialize(NULL);
}
//---------------------------------------------------------------------------
__fastcall TForm1::~TForm1()
{
if (pVoice) pVoice->Release();
::CoUninitialize();
}
//---------------------------------------------------------------------------
void __fastcall TForm1::FormKeyUp(TObject *Sender, WORD &Key, TShiftState Shift)
{
if (Key == VK_F11)
{
TWinControl *Ctrl = Screen->ActiveControl;
if (Ctrl)
{
TButton *btn;
TEdit *edit;
if ((btn = dynamic_cast<TButton*>(Ctrl)) != NULL)
SpeakText("Button, " + btn->Caption);
else if ((edit = dynamic_cast<TEdit*>(Ctrl)) != NULL)
SpeakText("Edit box, " + edit->Text);
}
}
}
//---------------------------------------------------------------------------
ISpVoice* pVoice = NULL;
bool __fastcall TForm1::SpeakText(const String &Text)
{
s8087CW cw(0x133f);
if (!pVoice)
{
if (FAILED(CoCreateInstance(CLSID_SpVoice, NULL, CLSCTX_ALL, IID_ISpVoice, (void **)&pVoice)))
return false;
}
SPVOICESTATUS stat;
pVoice->GetStatus(&stat, NULL);
while (stat.dwRunningState == SPRS_IS_SPEAKING)
{
ULONG skipped;
pVoice->Skip(L"SENTENCE", 1000, &skipped);
pVoice->GetStatus(&stat, NULL);
}
return SUCCEEDED(pVoice->Speak(WideString(Text).c_bstr(), SPF_ASYNC, NULL));
}