Is there a document anywhere that specifies how to generate
and interpret logs in the scrolls format? The phrase “a key=value
structure” is pretty open-ended.
I took a stab at defining an unambiguous format in this style,
using JSON as a baseline and tweaking it until it looks right for
logs. The idea is to take advantage of people's existing mental
model for JSON data and just define a different encoding.
One thing I'm concerned with is round-trip accuracy: encoding
a log message, then decoding it again should produce the same
data you started with.
What do you think of this? Does it strike a good balance between
flexibility and comprehensibility? Do you think it would be worth
changing scrolls to explicitly stick to this format (even if we have
to tweak the definition)?
https://gist.github.com/0e8d5ee4b954ce604bb2
Is there a document anywhere that specifies how to generate
and interpret logs in the scrolls format? The phrase “a key=value
structure” is pretty open-ended.
I took a stab at defining an unambiguous format in this style,
using JSON as a baseline and tweaking it until it looks right for
logs. The idea is to take advantage of people's existing mental
model for JSON data and just define a different encoding.
One thing I'm concerned with is round-trip accuracy: encoding
a log message, then decoding it again should produce the same
data you started with.
What do you think of this? Does it strike a good balance between
flexibility and comprehensibility? Do you think it would be worth
changing scrolls to explicitly stick to this format (even if we have
to tweak the definition)?
https://gist.github.com/0e8d5ee4b954ce604bb2