hgcummings / hypermeter Goto Github PK

Need to do this when all request have completed asynchronously (use promises)

Probably under a new checker section in config

Some of the tests are a bit too big (run the whole process rather than unit testing one module)

Need to consider whether the response time includes the redirect or is just the time of the last hop (probably the latter)

i.e. use promises throughout

The plotly interaction could probably be pushed out into a separate module, as could the code for resolving environment variables in config.

Probably implemented as an optional checker.

Need to think about at what point responses are saved for comparison.

The cURL-based client appears to sometimes return some/all of the response instead of the actual response code, resulting in spurious failures.

e.g. using loglevel

The X axis type is 'category' by default, which can cause results to be displayed in the wrong order (leading to some very oddly shaped plots).

Publish to npm

A bit long and repetitive at the moment. Would probably also benefit from factoring the plotly code out of the graph module (see #14)

Add proper documentation

In the case where an endpoint fails (so isn't included in the output) then reappears, the new value is appended to the list of values without any null value in between to represent the skipped build.

Note that the JSON reporter should possibly output all URLs rather than just failing ones, which would mitigate this (but it would still need solving anyway).

Should include:

• Timings (in ms) for each URL
• List of failed URLs

e.g. if no config file specified, or config file contains no URLs.

Graph name should be read from config file. Values should be appended to existing graph.