Hi Steve!
So I've been talking to my colleagues at Zillow, and there's some
functionality that they wanted implemented in HttpArchive. I know some
of these might not make sense in terms of what HttpArchive's goals
are, but I was hoping you could help me on figuring out a way to add
these on that makes sense.
1. Graphs for load time, document complete, and start render, fully
loaded
These are performance-related statistics that we were looking for, and
it looks like the method you use to generate these graphs is by using
the columns in the Pages table. As you're not looking for this data, I
believe that it probably wouldn't make sense to add new columns into
the table. What I was thinking about doing was adding another table
which just contains the timing values and a field for the page label,
so they can be linked, then keeping a separate .inc file that has the
methods with which to graph these. Extensible but a non-necessary part
of HttpArchive.
2. api for triggering a HttpArchive run on a specific URL.
We were imagining a method with which one could activate a run of a
specific site with an HTTP request, something along the lines of:
httparchive.org/run?url=ESCAPED_URL_HERE&key=AUTHENTICATION_KEY
The key would be some sort of hash, so not anyone could simply start a
run.
My thinking is this could be done along with the redesign of the
batch_process and batch_start methods, which you said needed to be
made more efficient.
I was also thinking about adding a page that would allow for simpler
configuration of HttpArchive, Possibly containing settings like which
graphs and trends to show, so people who want private instances of
HttpArchive can see information relevant to them.
Please let me know your thoughts on these. I know these aren't
necessarily in the scope of HttpArchive, but we feel like HttpArchive
can become a versatile performance testing tool as well as a great
look at the trends in internet technology.
Thanks! Sincerely,
-Yusuke