Maybe i can help. Is there a test project? What patch/branch should I look at.
Hi, thanks, the test workspace is in this post
Re: CC upper limits ? BUT for sure BUGs (supplied by killerbot)
the patch is in this post:
Re: CC upper limits ? BUT for sure BUGs, it is against the CB trunk code, but I'm using GIT clone, so patch is in git format.
You need to open the workspace, and see the logs in both "Codeblocks log" and "Codeblocks debug log"(start C::B with debug-log options) panel.
Also, the CC option should be set to "one parser for the whole workspace" mode, it can be done by opening the CodeCompletion plugin option, and in the bottom of the c/c++ parser page, once changed, you should restart C::B(I guess it is a must step).
In the patch file, I have generally wrote the logic of how CC add files to the parser, it add projects one by one, there are three projects:
When finish parsing the Dummy project, all the files in the NamingService were parsed too. This means the files of NamingService were all belong to the Dummy project, so NamingService doesn't need to add files to the "thread task pool" to run the parsing tasks, also, under Linux, the "fetch of the GCC predefined macros" is always failed (you can see jens' reply before, I don't know why), as an conclusion, NamingService doesn't make the "thread task pool" to run, so all the later parsing jobs were canceled, because the start parsing of the next project only happens we receive a "thread task pool done" event.
My patch try to detect the condition that if both "file need to parse" and "fetched GCC predefined macros" are empty, we try to parse the next project, in killerbot's sample, it is the third project "Dummy2".
I'm not sure you understand my explanation, I don't have this issue under Windows. Thanks.