authenticators and chained session factories. We can share a lot more code this
way and the implicit requirement that the default authenticator/factory had to
be configured before anyone else configured a chanied author/factory is gone.
The other big change is that Credentials doesn't require a username
(UsernamePasswordCreds inherits that username so most derived classes don't
notice any difference). Instead we require that a canonical authentication
username be determined and configured in AuthingConnection during the
authentication process. This canonical username is then used to resolve the
client session and map everything in the client manager.
This is pretty much exactly what was going on before except that we were doing
it all in an ad hoc way by jamming a new name into Credentials during the
authentication process and also doing jiggery pokery in
PresentsSession.assignStartingUsername. That all goes away and/or becomes
cleaner and more explicit.
This is going to impact some Yohoho jiggery pokery, which I will shortly commit
a patch for, but we're going to need to test it. Omelets, eggs, etc.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@5828 542714f4-19e9-0310-aa3c-eee0fc999fb1
will not name, was trying to allow certain authentications to proceed directly
on the authenticator thread instead of making a round trip through the
invoker. This resulted in those authentications trying to post their results
from the authenticator thread, which is bad because all downstream messages
should be posted from the distributed object thread.
The only authentications that would "benefit" from this optimization were peer
authentications of which there would be extremely few (currently one in our
largest installation) compared to the hundreds of thousands of normal
authentications, which have to make the invoker thread round trip as a matter
of course.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4655 542714f4-19e9-0310-aa3c-eee0fc999fb1
condition between the omgr thread and the conmgr thread. Now when the omgr
thread processes an event that is going out to the clients, it flattents
the message itself for each client that is to receive the message and the
flattened data is posted to the conmgr outgoing queue.
This means that once an event is finished processing, no further
modifications to any of the data associated with the event can effect the
data queued up to be sent to the client. This is a good thing, it will
eliminate or illuminate a very baffling class of bugs that we've sort of
been ignoring because we knew this could be the cause.
We used to take an event and flatten it directly into the direct buffer
from which we would do our socket write. Now we flatten it into a
temporary byte array. This means a metric shitload more garbage generation
and collection. We used to do the flattening on the conmgr thread, now we
do it on the omgr thread. This means a big redistribution of CPU demand.
Either of those things could result in a significant negative impact on
our performance, but we'll just have to deploy this stuff and find out.
Whee! If it turns out to be a serious problem, there are potential
optimizations that could be done by keeping a pool of direct buffers
around and flattening messages into them, relying on the fact that the
outgoing conmgr queue generally doesn't grow too large and we could
allocate tens to a hundred megabytes of memory for the outgoing queue if
we really needed to.
I'd also like to test the overflow handling stuff more. It didn't really
change in that everything just deals with arrays of bytes now instead of
unflattened messages, but I'll be more comfortable once I've seen all this
in action on ice where there may be few users, but they are just as likely
to experience lag and receive an overflow queue as users on the higher
traffic servers. There is code to log when overflow queues are created and
finally flushed and how much use they got while they were around, so that
should give us an indication of whether things are operating properly.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@3419 542714f4-19e9-0310-aa3c-eee0fc999fb1
process so that it can be done by other entities than just the client
management services. Coordination between these parties is managed so that
no toes are stepped on in the course of loading and unloading clients and
everything is generally much nicer.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@1086 542714f4-19e9-0310-aa3c-eee0fc999fb1
longer maintained. Implementations can choose to create their own
authentication thread if they wish or use some existing combination of the
invoker and dobjmgr threads. Also added an invoker to the base server
class.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@1085 542714f4-19e9-0310-aa3c-eee0fc999fb1
changed to Presents and Party changed to Crowd. Whee!
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@431 542714f4-19e9-0310-aa3c-eee0fc999fb1
packages. These contain all the message codes as well as error response
codes that are used by a particular invocation service.
Also changed client.LocationManager to client.LocationDirector and
chat.ChatManager to chat.ChatDirector to go along with the new philosophy
of naming the client-side managing entity for an invocation service a
director.
Also elimitated cocktail.util.Codes since it's no longer used as a central
repository for codes (instead they are in InvocationCodes and its
derivatives).
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@368 542714f4-19e9-0310-aa3c-eee0fc999fb1
Michael Bayne