Ping

就是一个啊

https://github.com/coolsnowwolf/packages/commit/b09bcb347505050e83e26160f63463fef018c16b 把这单revert之后编译的可以启动起来

Here is some summary info generated from genhtml.json(`--profile` output), LCOV version 2.1-beta2.3-g6141622 without `--parallel` option: ```json {"test1": {"overall": 69.693657875061, "html": {"cnt": 884, "min": 0.00220179557800293, "max": 0.715844869613647, "sum": 37.969685792922974, "avg": 0.042952133249912865},...

All files are processed locally rather than retrieved from the network. As you concerned, disk storage might be slow, so this time the output directory was mounted as tmpfs, but...

Have written a script to generate sample project, the results may be reproducible. [gen_sample_proj.tar.gz](https://github.com/linux-test-project/lcov/files/14957791/gen_sample_proj.tar.gz) ``` # To generate sample project with 800 files, 100 functions per file ./gen_sample_proj.py 800 100...

In my case, several .info files were processed one by one, and "--parallel" increased the overall time. Generating the coverage report took almost half of the total time. I noticed...

By "processed one by one" I mean: ``` genhtml -o output1 --branch --prefix prefix --parallel -- 1.info genhtml -o output2 --branch --prefix prefix --parallel -- 2.info genhtml -o output3 --branch...