<div>
Platform? Compiler? Boost version?
</div>


<div>
FWIW, I tried this is Visual Studio 2017 on a Windows 10 machine with boost_1_62_0 and had no problems.
</div>


<div>
do you run this in debug mode or release? and is the project with clr and/or native c++?<br />
<br />
if 10,000,000 objects were created with gc you might get memory issues depending on your environment.<br />
<br />
you probably got some messages in the debug output window when the crash happens?<br />
<br />
Sara
</div>


<div>
Boost 1.59<br />
VS 2015, compiled with platform toolset v140, no clr, release version (so no debug output)<br />
The exception message says there is a &quot;System.AccessViolationExc<wbr />eption&quot;.
</div>


<div>
Have you considered trying the latest version of Boost? That version is pretty old. You can download precompiled binaries from the Boost website.
</div>


<div>
Ok, I downloaded boost 1_63_0 (got lot's of errors packing out the zip file, but the necessary files for uuid seem ok).<br />
<br />
No difference, still crashes...
</div>


<div>
I have tried with:<br />
<br />
#define BOOST_UUID_NO_SIMD<br />
<br />
before inclusion of the boost headers, no difference...
</div>


<div>
You built Boost yourself or used the precompiled binaries? I always use the precompiled binaries, which can be installed using the provided setup file.
</div>


<div>
<a href="https://sourceforge.net/projects/boost/files/boost-binaries/1.63.0/" rel="ugc">https://sourceforge.net/projects/boost/files/boost-binaries/1.63.0/</a>
</div>


<div>
Good point. It could be. To be completely honest I didn't check. Very weird you're seeing this crash though. I've used this in production a number of times without issue. Without being able to reproduce the issue, it's hard to know what else to check other than to recommend trying it on a completely different machine just to rule out environmental factors. Maybe also try checking you have the latest 2015 updates or even try VS2017.
</div>


<div>
I can't really switch compiler, this is a huge system with many components. We have never seen any problems (as far as we know) with this in normal use. But I got sceptical when I made this test and got a crash.<br />
<br />
Anyway, the main question is actually why should we use boost for this at all and not the rpc version instead?<br />
<br />
Thanks for trying to help!
</div>


<div>
I am going for dropping the boost version, since we don't use boost anywhere else. So even though the crash problem wasn't solved, I got help to reach my conclusion on what to do because of the comment about boost not really making any difference.<br />
<br />
Thank you!
</div>


<div>
You're welcome. If you do ever figure out why it crashes, I'd be curious to know the reason.<br />
<br />
All the best.
</div>


<div>
does it crash in debug mode as well? or if you decrease the loop count? access violation always means pointer error. as the destructor definitively would free all pointers before next loop cycle and next construction would/could use the same heap memory again, &nbsp;in my opinion the crash only could happen because of optimization the compiler has made for the loop. you could turn off optimization or print the generated guid to a file. both should at least change the behavior, perhaps even solve the issue.<br />
<br />

<blockquote>
why should we use boost for this at all and not the rpc version instead?
</blockquote>
i think you have answered the question to yourself, that you won't go to boost as long as the issue is not resolved. i would not see any reason to prefer boost even if it was faster.<br />
<br />
Sara
</div>


<div>
Thank you Sara! I had already tested streaming the guids to a file before I even asked this question. Things crashed right in the middle of it, and the file would have less guids than expected and the last one could even be shorter than supposed.<br />
<br />
I won't be spending any more time on this, since I kicked out boost.
</div>


<div>
<div class="content wysiwyg-content">
Hi!<br />
<br />
I am trying to compare speed of boost::uuid vs UUidCreate from rpc.h. But the boost version crashes quite quickly in this loop! Any ideas?<br />
<br />

<pre><code class="code-1 nolinks" id="code-10-29010949-1"><span># include &lt;boost/uuid/uuid.hpp&gt;</span>
<span># include &lt;boost/uuid/uuid_io.hpp&gt;</span>
<span># include &lt;boost/uuid/random_generator.hpp&gt;</span>
<span></span>
<span>  static boost::uuids::random_generator gen;</span>
<span>  for (int i = 0; i &lt; 10000000; ++i)</span>
<span>    to_string(gen());</span>
</code></pre>
<br />
Is there really any reason we should prefer the boost version?<br />
<br />
Thank you!
</div>
</div>


Hi!

I am trying to compare speed of boost::uuid vs UUidCreate from rpc.h. But the boost version crashes quite quickly in this loop! Any ideas?

(CODE)

Is there really any reason we should prefer the boost version?

Thank you!

boost::uuid crashes

C++ is an intermediate-level general-purpose programming language, not to be confused with C or C#. It was developed as a set of extensions to the C programming language to improve type-safety and add support for automatic resource management, object-orientation, generic programming, and exception handling, among other features.