[andrew.bentley]

I apologize for being unclear, I missed the “public:” declaration. The parent functions in this case are indeed public so are accessible to the derived classes. The code compiles and runs correctly, it’s only intellisense that gives me issues on this.

[andrew.bentley]

I appreciate you pointing that out. For our existing projects that conversion may be a tough sell. I will experiment with CMake though, as it seems to offer some advantages.

Thanks,
Andrew

[andrew.bentley]

support,

This is very close to what I are looking for. I have a couple of comments.

• While filtering like this makes it much easier to identify which relevant files are being tested, filtering does not allow recalculation of the overall  statistics. I would like to exclude system files from those statistics so we can get a quick summary. If it can recalculate on the fly that would be great, but I would be content even if it required rebuilding and rerunning the test suite.
• I really like being able to right-click and add files. Excellent feature.
• There are an awful lot of files being added to the exclusion list. For example, the boost library contains dozens of different directories. Allowing wildcards to automatically filter subdirectories would shorten this list, although with the right-click feature above perhaps that’s not critical.

Thanks,
Andrew

Attachments:

You must be logged in to view attached files.

[andrew.bentley]

That sounds like a good approach to things.

Once concern I have with it is building and executing tests on a remote server. I haven’t worked a whole lot with gcov, but it appears to produce a separate gcov file for each source file (as opposed to one for each object file). I can see the project gcov files get copied back into the source file directory, but what about gcov output for the corresponding system files, where do those get copied to? I want to make sure the remote use case is covered by whatever solution we come up with.

Thanks,
Andrew

[Author]

Posts