Deprecated Catapult GitHub. Please instead use http://crbug.com "Speed>Benchmarks" component for bugs and https://chromium.googlesource.com/catapult for downloading and editing source code..
When running a benchmark we get some strange temp file names being printed on the output without much context nor explanation, e.g.:
$ tools/perf/run_benchmark memory.top_10_mobile
69 KB/s (3387 bytes in 0.047s)
[ RUN ] https_www_google_com_hl_en_q_science
/tmp/tmpSLAytK.html
[ OK ] https_www_google_com_hl_en_q_science (19514 ms)
[ RUN ] after_https_www_google_com_hl_en_q_science
/tmp/tmpHiafO9.html
[ OK ] after_https_www_google_com_hl_en_q_science (16478 ms)
[ RUN ] https_m_facebook_com_rihanna
/tmp/tmpoZbald.html
[ OK ] https_m_facebook_com_rihanna (15878 ms)
[ RUN ] after_https_m_facebook_com_rihanna
/tmp/tmpo1Slbw.html
[ OK ] after_https_m_facebook_com_rihanna (16371 ms)
...
After some bisecting through the code I eventually pinned those strange looking lines of output coming out of telemetry.value.trace. They started appearing after @randalnephew moved the trace to html conversion out of process. :-)
I'm considering just replacing that check_call to check_output so we just swallow the stdout of the subprocess. Does that sound reasonable?
nedn
commented
8 years ago
When running a benchmark we get some strange temp file names being printed on the output without much context nor explanation, e.g.:
After some bisecting through the code I eventually pinned those strange looking lines of output coming out of telemetry.value.trace. They started appearing after @randalnephew moved the trace to html conversion out of process. :-)
I'm considering just replacing that
check_calltocheck_outputso we just swallow the stdout of the subprocess. Does that sound reasonable?