I propose to add an option to scan-build to ignore problem reports 
containing certain strings (user-specified list of strings to ignore) in 
certain parts of the problem report.

Rationale and explanation:

We use clang-scanbuild to find problems in our codebase.  But it finds 
problems related to a package we use, uthash that we have determined are 
not problems.  We are not prepared to change the uthash package, so we want 
to ignore those error reports.

I noticed that clang-scanbuild produces a html-file per error, containing 
the marked-up source code of the source file containing the error.  The 
actual error line is indicated, along with (potentially) a series of other 
lines where assumptions were made ("events") about the state of the program 
leading to the error.  In the uthash problem reports, the last one of these 
events always mentions a string related to uthash such as HASH_DEL.

So my proposed feature will allow us to specify a set of strings to ignore, 
and when those strings appear in the last of the series of "events" that 
lead to an error, then that error will not be included in the error summary.

Does this sound useful to the community?  Shall I upload it?  I still need 
to make some tests.

Testing

For testing, I was thinking of making some html files with and without the 
strings to find, and seeing if the summary contains the right errors ... 
does that sound like a good approach?  I will have to make a sample 
program, I should not use our unreleased code for this.  Any advice on the 
testing will be welcome.

bjb

-- 
You received this message because you are subscribed to the Google Groups 
"Jenkins Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to jenkinsci-dev+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/jenkinsci-dev/3f653c8b-5c6a-4986-984f-bbea96abadf8%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to