llvm/clang static analyzer reports
hoes.maarten at gmail.com
Thu Oct 15 17:48:33 UTC 2020
On Thu, Oct 15, 2020 at 6:51 PM Stephan Bergmann <sbergman at redhat.com>
> On 15/10/2020 10:58, Stephan Bergmann wrote:
> > On 15/10/2020 09:51, Maarten Hoes wrote:
> >> Just one more thing I need to mention. At some point during the
> >> build/analysis, I run into the following:
> >> The analyzer (or make, I can't tell which) goes looking for
> >> '/usr/bin/clang-10++' (which isn't there) instead of
> >> '/usr/bin/clang++-10'. In other words, it looks for clang-version++
> >> instead of clang++-version. I haven't really investigated much, but
> >> chose to be pragmatic and just add a symlink to work around the issue.
> >> If this is an acceptable workaround, I'll add a README that mentions
> >> this. On the other hand, if this is a dealbreaker and people want this
> >> 'fixed' before continuing, then I am going to need a lot of help in
> >> finding the root cause (so I can submit a bug report with the project
> >> that causes the behaviour).
> > We can have a look once you have sent something to Gerrit.
> <https://gerrit.libreoffice.org/c/core/+/104383> "Rename CLANG_CC,
> CLANG_CXX configuration vars (avoid clash with scan-build)" should get
> that fixed
Well, that was quick. Thanks for helping out. I personally assUmed (but you
know what happens when we assume) that someone made a mistake/typo in
either 'scan-build' or 'c++-analyzer' but I don't speak perl, and couldn't
find any obvious mistakes myself (and then incorrectly proceeded to wonder
why something like that could have gone unnoticed for so long and in so
many releases of clang/llvm). Thanks again.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the LibreOffice