This commit is largely based on work by Thiago Arrais (@thiagoarrais)
and Shane Bradfield (@l33tLumberjack). I wasn't really able to get their
PR reasonably merged onto the latest master, so I reapplied their changes
manually here and did some further improvements on that.
This adds a set of store modification events, and uses them for flow settings.
This addresses a bug where settings could persist even after flows were deleted.
- Handles upstream CONNECT and regular requests, plus HTTP Basic for reverse
proxy
- Add some tests to make sure we can rely on the .via attribute on server
connections.
- CONNECT requests do not generate the usual http events. Instead, they
generate the http_connect event and handlers then have the option of setting an
error response to abort the connect.
- The connect handler is called for both upstream proxy and regular proxy CONNECTs.
Also:
- Remove duplicate and unused code
- Tighten scope of HttpReadDisconnect handler - we only want to ignore this for
the initial read, not for the entire block that includes things like the expect
handling.
Instead of a (file, mode) tuple, move to "streamfile" as the path, and
"streamfile_append" to specify if we should append or not.
Leave the command-line options intact for now.
This is a tad complicated. The underlying sorted list implementation expects
the sort key to be stable for the lifetime of the object. However, if we sort
by size, for instance, the sort order changes as the flow progresses through
its lifecycle. We address this through two means:
- Let order keys cache the sort value by flow ID.
- Add a facility to refresh items in the list by removing and re-adding them
when they are updated.
Marking now works differently - it's no longer just another filter, it's
applied in addition to the filter. This means you can apply a filter, mark some
flows, and then toggle between marked and unmarked flows matching the filter.
I'm leaving the filter for marked flows intact - it will come in handy in other
situations.
