[cfe-dev] Will libclang ever be able to step through classes defined by template instantiations?

Justin Bogner via cfe-dev cfe-dev at lists.llvm.org
Wed Dec 20 08:50:05 PST 2017

(Forgot to reply to list, sorry for the duplicate)

> On Dec 20, 2017, at 07:01, Wills, Samuel - US via cfe-dev <cfe-dev at lists.llvm.org> wrote:
> Hello,
> Last month, I posted a question to the Clang developers. So far, I have not gotten a response from any Clang developers, and I still have no answer to my question. My plan is to continue to post this question to the boards every two weeks, until I get a response from a developer. (The response does not have to answer my question, it just has to let me know that you have seen it and will try to let me know of an answer in a certain amount of time.) If there is a better or more appropriate way to get a response than reposting the question every two weeks, please let me know.

It’s much better to bump a thread by replying to the the original thread than to start a new one. This makes it much easier for people to follow the context.

That said, simply pinging a thread doesn’t guarantee you’ll get any more response than you did the first time, as that can very much depend on the question.

> In short, my question is:
> Are there any plans to allow libclang to step through classes defined by template instantiations in the future? (Please take a look at my original post below for details.)

This style of question is tricky, because if many people look at it and think, “I don’t personally plan on doing that”, they aren’t necessarily going to be confident to say, “there are no such plans”. I think the lack of responses is probably a good indicator that nobody is looking at this right now.

One way you could potentially move forward with this (if you’re willing and have time to put in the work) would be to propose changes, ideally with a patch, that extend the API in the way you need. That approach is more likely to get traction with people, as they can evaluate it on its own value and compatibility with their use cases.

> Here is my original post:
> Hello,
> I have been working on a project for my company over the last few months that uses libclang. Over the last couple of months, I have been working to get information about classes created by template instantiations in our programs. Unfortunately, libclang skips classes defined by template instantiations in its walk of the AST. I know that Clang has other functions that are able to step through template instantiations (such as RecursiveASTVisitor), but those functions are considered unstable. My company wants something stable, which libclang is. I am currently working on a workaround that will solve most of our problems, but will not solve all of them. Are there any plans to allow libclang to step through classes defined by template instantiations in the future? At that point, I could rewrite the code of our company's tool to do everything that we want it to do.
> This is my first time posting to this email list. Also, this is my first year at a programming job, so I am somewhat new to this industry.
> Thank you very much!
> Samuel Wills
> _______________________________________________
> cfe-dev mailing list
> cfe-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20171220/ea6a6179/attachment.html>

More information about the cfe-dev mailing list