[cfe-dev] ignoring*() and hasParent() matchers

via cfe-dev cfe-dev at lists.llvm.org
Wed Jul 11 08:54:23 PDT 2018

This would be an awesome topic for a talk or tutorial at one of the semi-annual LLVM Developer Meetings.  Just saying.

From: cfe-dev [mailto:cfe-dev-bounces at lists.llvm.org] On Behalf Of Michael Jabbour via cfe-dev
Sent: Wednesday, July 11, 2018 11:07 AM
To: Manuel Klimek
Cc: Clang Dev
Subject: Re: [cfe-dev] ignoring*() and hasParent() matchers

I am certainly planning to invest some time writing blog posts/scripts to make clang-tidy checks easier to use. Giving a talk seems a bit impossible in the near future, as there isn't much interest in clang in my local community (hope that this changes).

Also, I have written some<https://github.com/micjabbour/pronto-clang_format> scripts<https://github.com/micjabbour/pronto-clang_tidy> (that are meant to be used in continuous integration) to get clang-tidy/clang-format output as comments in popular web-based git repo managers (e.g. github, gitlab, bitbucket, ...). This can help enforce some coding conventions when reviewing code changes. Here is a screenshot on gitlab (I'll be deploying this into production soon):
[cid:image001.png at 01D4190D.E6E73B90]

Sorry for the long email, Just wanted to share my enthusiasm :D
Thanks again,
On 07/07/2018 6:04 PM, Manuel Klimek wrote:
Thanks! You can pay that back by giving a talk about clang tooling from a user's point of view at some point, as you seem to have mastered the subject :)
On Sat, 7 Jul 2018, 17:23 Michael Jabbour, <micjabbour at gmail.com<mailto:micjabbour at gmail.com>> wrote:

Thanks for the confirmation. I have learned a lot from your great talks/presentations :-)



On 06/07/2018 11:03 AM, Manuel Klimek wrote:
That's what I'd do, too.

On Fri, 6 Jul 2018, 00:50 Michael Jabbour via cfe-dev, <cfe-dev at lists.llvm.org<mailto:cfe-dev at lists.llvm.org>> wrote:

I ended up writing my matcher like this:

  expr().bind("literal"),       //inline bind call using a redundant node matcher
  hasAncestor(                  //climb up the AST to the nearest varDecl ancestor
    varDecl(hasInitializer(ignoringParenCasts(equalsBoundNode("literal"))))    //check that this varDecl satisfies the original condition

Any feedback is appreciated, as I am completely new to LibTooling...

P.S.: The actual matcher I am working on is far more complicated. Inverting it (i.e. using my very first example) would result in much more difficult problems. Although this seems like a roundabout hack, it might be effective when nested inside matchers that have already been tested



On 05/07/2018 7:31 PM, Michael Jabbour wrote:

Hello all,

When using has*() it is sometimes desirable to use ignoringParenCasts(), for example (from the docs):


Is there any way to get the same effect when hasParent(), e.g.:


Obviously this does not work, but what is the best way to achieve a similar effect?



cfe-dev mailing list
cfe-dev at lists.llvm.org<mailto:cfe-dev at lists.llvm.org>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20180711/af2aa557/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.png
Type: image/png
Size: 37965 bytes
Desc: image001.png
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20180711/af2aa557/attachment.png>

More information about the cfe-dev mailing list