When it comes to the speed of the pipeline scan, one of the things we have found with Veracode is that it's very fast with Java-based applications but a bit slow with C/C++ based applications. So we have implemented the pipeline scan only for Java-based applications not for the C/C++ applications.
For C++ based languages, or languages where there is a platform dependency—for example, if I write C language code it is dependent on whether I'm executing that on Windows, or on Linux, or another platform—and with some of these platforms-specific languages, Veracode makes something called debug symbols that are introduced into the code. That gets cumbersome. They could improve that or possibly automate. If Veracode could quickly analyze the code and make file-line flags, that would be great. It is easy to do for Java, Python, and Pearl, but not so easy for C++. So when it comes to the debug symbols, guidance or automation could be improved.
Also, scan completion, as well scanning progress, is not reported accurately. Sometimes the scan says it will complete in two to three hours but it will take four or five hours. That is one of the areas where they can give a more accurate estimate.