When using the program slicer of Frama-C version Oxygen, I have the problem that the resulting slice uses undeclared variables.
I searched for existing postings to this topic before and found this:
http://bts.frama-c.com/print_bug_page.php?bug_id=806
There it is mentioned that the bug was fixed in the Nitrogen version of Frama-C. Maybe this change was not carried over to Oxygen? Like in the description of the existing posting it only happens for blocks with just one statement.
I cannot attach the example source code since it is from a customer project.
I have checked the exact steps described in the bug report you mention with Frama-C Oxygen (and csmith 2.0.0 for Csmith's runtime library), and everything works fine: it's very likely that you're experiencing another issue, but without the code, it's impossible to say more.
Related
I am getting the error here in this Travis build, and I cannot reproduce it locally. Yes, I realize that I do not have a minimal reproducible example, but I do know that it happens within tidyselect::vars_select(). Has anyone else encountered this before? I cannot find any mention of lang_unnamespace() in a Google search.
See if modifying your travis yaml to include the dev versions of the tidyverse packages help.
It looks like you're not calling rlang::lang_unnamespace() directly, correct? That function appears to have been removed, but tidyselect::vars_select() is still trying to call it.
I arbitrarily went back to the beginning of the year to see this old version.
The file doesn't even exist in today's version.
I'm stuck using the 4.0 version of lua which does not seem to support the os library. Is there a way to include this library into my project?
Or get another way to use the functionality contained within pertaining to date time calculations?
Preferably by using a *.lua file and not a *.c file since I don't have complete access to the code.
When I run the following line,
print(os.time{year=1970, month=1, day=1, hour=0})
I get an error stating:
attempt to index global 'os'(a nil value)
As #Colonel Thirty Two said it's not possible to use the os library. So the time() funciton is not available for me.
Adding to the (totally correct) currently accepted answer (that if "os" access was not allowed to you, you're generally done), there's some very slight chance the Original Programmer may have provided you with some alternative facilities to do your thing (fingers crossed). In a perfect world, those would be described in some kind of a User's Manual for your scripting environment. But if the manual was lost to time (or never existed in the first place), you might possibly try your luck at exploring any preloaded libraries by digging through the result of the globals() Basic Function. (At least I hope that's how it was done in 4.0 too.) That is, if the Original Programmer didn't block globals() for you too...
As you can guess I am at Qt developer and in the interest of getting up and running with libspotify as quickly as possible I am looking for a Qt wrapper.
I did come across this link https://github.com/romnes/libqspotify but as you can see the source is two years old. I am guessing libspotify has moved on a lot since then.
Does such a wrapper even exist?
Thanks
QSpot appears still to be in development and is based on libqspotify (they have copied the libqspotify sources into their qspotify_src directory). There are some recent commits (August 2012) to that directory, so I'd guess their classes are fresher than the ones you found on GitHub.
The sources of QSpot are found here.
If that doesn't work for you, there is also MeeSpot which is based on a library called libQtSpotify, located in MeeSpot's sources.
There's also Tomahawk. It's also open source
I am using the Nitrogen version of Frama-c on Mac, and can't seem able to use
the "set" logic, as documented in the ACSL manual, e.g., I can't declare
a ghost variable as in "//# ghost set<integer> someSet;".
The frama-c program always complains about a syntax error in the line where a set is declared, no matter what.
I also tried "Set" instead of "set", other types in place of "integer" (e.g. "char*")and specifying "//# open set;" to import the module.
Maybe I need to specify some command line option? Executing "frama-c -kernel-help" it's not clear what that would be though.
Or maybe the Mac version (I downloaded the Intel binary version) is outdated and I should compile the latest source code ?
Thanks, best regards,
Eduardo
ACSL is an annotation language that exists independently of Frama-C, although some of the same persons work on both. From the point of view of usage of ACSL in a Frama-C plug-in, there are three levels of definition/implementation, and you need all three to be able to use a feature:
The feature must be part of the ACSL language.
It must be made available by the current Frama-C front-end. Not all features of the ACSL language are immediately implemented in the front-end.
The plug-in you intend to use must take advantage of it.
Another explanation of the same distinction is here.
I can't declare a ghost variable as in "//# ghost set someSet;".
In your case, it appears that the partially implemented feature is not so much sets (which seem implemented in the front-end after a quick look) but ghost code, which can currently only use C constructs and types.
Or maybe the Mac version (I downloaded the Intel binary version) is
outdated and I should compile the latest source code ?
You have the latest version at this time.
I have some dll's for which i want to run code coverage using TestDriven.net (ncover) with vs2010. There are some methods which i don't want to be included in the coverage result since those methods are already written by some other programmer the option of using CoverageExclusion attribute is not feasible.
Is there any other way out ?
Unfortunately, as far as I can tell, Testdriven.NET does not yet support using other NCover arguments, like //ea.
I found this post, where it was last discussed and Jamie Cansdale at Testdriven confirmed this:
https://groups.google.com/group/testdrivenusers/browse_thread/thread/a0dc80c40c5f8815?pli=1
I found an interesting post about using NCover 3.4.x with Testdriven, just FYI:
http://thepursuitofalife.com/how-to-connect-ncover-3-x-with-testdriven-net/
So, my best advice is to contact support#testdriven.net to find out the latest on configuring and using NCover parameters.
Best Regards,
NCover Support
NCover has a command line parameter //ea which stands for 'exclude attributes'. It allows you to provide a list of attributes marking classes or methods to exclude from coverage.
There is a blog post about it here.
This feature works with the version of NCover that comes with TestDriven.NET (1.5.8).