epics/pvData
0
0
Fork 0
Code Issues Pull Requests Projects Wiki Activity

modify test.

Browse Source 
update README
This commit is contained in:
Marty Kraimer
2010-08-10 13:23:17 -04:00
parent deb9b23391
commit d54893b18e
2 changed files with 13 additions and 33 deletions
Download Patch File Download Diff File Expand all files Collapse all files

41
README.html
View File

@@ -10,7 +10,7 @@
<body>
<h1 style="text-align: center">EPICS pvData C++<br />
Overview<br />
2010.08.02</h1>
2010.08.10</h1>
CONTENTS
<hr />
@@ -27,26 +27,9 @@ following is done:</p>
Scalar</dd>
<dt>test</dt>
<dd>A test of Scalar.</dd>
<dt>As mentioned below there are major problems with the current
<dt>As mentioned below there are problems with the current
implementation.</dt>
</dl>
<p>Since the last version the following are the main changes:</p>
<ul>
<li>The name space no longer has the trailing " namespace pvData {" </li>
<li>A typedef has been added to define a StringConst and all string
arguments are now a StringConst. See gargage collection below for
details.</li>
<li>typedefs have been added for FieldPtr, ScalarPtr, etc. These make the
boost stuff somewhat transparent.</li>
<li>The array arguments have been changed from using * to using [].</li>
<li>Other arguments were change to be more understandable.</li>
<li>All the toString methods were changed from something like:
<pre> static const std::string toString(...</pre>
to
<pre>static void toString(std::string &amp;buf,</pre>
This should be more efficent.</li>
</ul>
<hr />
<h2 style="text-align: center">Building</h2>
@@ -76,14 +59,14 @@ type:</p>
<p>The pure virtual classes defined in pvData.h now work. But there are still
some things that are not so nice. Amoung these are:</p>
<ul>
<li>User code sees pointers instead of references. Can this be changed
while keeping the boost::shared_ptr support?</li>
<li>In FieldCreateFactory.cpp look for "WHY DO I". It asks why the derived
class must also define methods defined in the base class. If it does not
then a compilation error occurs. WHY??? </li>
<li>The toString methods have an argument of "std::string &amp;buf" instead
of "std::string *". But in the test code I could not create a
"std::string &amp;" to work. I do not know why.</li>
of "std::string *". Does this seem correct?</li>
<li>Can arguments and return descriptions be defined better?</li>
<li>Is const present everywhere it should be? Remember that introspection
classes are immutable and are shared whereever possible.</li>
classes are immutable.</li>
<li>The code is NOT thread safe. When we decide for sure what thread/lock
support to choose I will fix this.</li>
</ul>
@@ -94,14 +77,6 @@ with c++</p>
<h2 style="text-align: center">Garbage Collection</h2>
<hr />
<p>boost::shared_ptr is currenly used for implementations of Field, Scalar,
etc. It is also used for all string arguments passed to methods. Since string
values are freely shared by pvData, pvAccess, etc, this is really important.
questions:</p>
<ol>
<li>Is the implementation using shared_ptr correctly?</li>
<li>Should other things currently defined use stared_ptr?</li>
</ol>
<p>Not yet implemented. Lets get class structure correct first.</p>
</body>
</html>

5
pvDataApp/test/test.cpp
View File

@@ -32,6 +32,11 @@ int main(int argc,char *argv[])
TypeFunc::toString(*myString,type);
printf("%s\n",myString->c_str());
myString->clear();
*myString += "fieldName ";
std::string const & fieldName = scalar.getFieldName();
*myString += fieldName;
printf("%s\n",myString->c_str());
myString->clear();
scalar.toString(*myString);
printf("%s\n",myString->c_str());
return(0);