Closed yuyantingzero closed 9 years ago
The SQL statements provided in the json dashboard are still using the {{dataset_name}} and {{table_name}} placeholders.
@cmccoy :
@jmuharsky : The SQL statement in this PR uses {{dataset_name}}, {{table_name}}. In the README.md file, there are instructions to replace dataset_name, table_name with target dataset and table. I can use samples_mart, results directly if you think that is enough for most users. I will send you the deployed preview dashboards offline.
@yuyantingzero: Thanks. What is the criterion for including a result from a benchmark in this summary page?
@cmccoy I am trying to include 1 or 2 most important metric per benchmark. The dive-in dashboards contains all metics with parameters. Feel free to point out any different metric should be added into summary dashboard in your mind.
Otherwise, LGTM.
@yuyantingzero We shouldn't be varying from the samples_mart.results pattern, so I think fixing these is the right thing to do. We can instruct the user to change those values if they should upload to a different dataset/table, but our tooling should guide them towards these values.
Recommend a more descriptive title than _PerfkitBenchmarker -- something like PerfKit Benchmarker Summary Dashboard (Sample).
@jmuharsky Addressed all comments. @cmccoy Used pivots in summary dashboard now, but since query has a different set of results, the color schema is different in each widget, any suggestions?
LGTM modulo missing data.
Recommend moving the legends to top, bottom or inline/right, and providing enough "left" on the chart area to account for the x axis.
Addressed all comments, PTAL @jmuharsky
@jmuharsky Updated color and ordering.
LGTM