aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
...
* improve the interface of component_future a bitJesse Luehrs2019-10-179-260/+286
| | | | make it feel more like the standard futures::Poll/futures::Async
* refactor config handlingJesse Luehrs2019-10-179-326/+688
|
* improve logging a bitJesse Luehrs2019-10-173-28/+25
|
* hide all logs except error logs for interactive subcommandsJesse Luehrs2019-10-173-7/+16
| | | | | this should let us use the logger more idiomatically without interfering with normal command output
* pull some more hardcoded values into constantsJesse Luehrs2019-10-172-7/+10
|
* also factor out client auth id savingJesse Luehrs2019-10-172-9/+16
|
* allow restricting the accepted auth type listJesse Luehrs2019-10-175-8/+94
|
* no need to reconnect the list client when the watch client disconnectsJesse Luehrs2019-10-171-1/+1
|
* add tests for protocol type handlingJesse Luehrs2019-10-171-2/+44
|
* also use an enum for message type instead of constsJesse Luehrs2019-10-171-108/+84
|
* the server stored tokens should also be data, not cacheJesse Luehrs2019-10-173-15/+6
|
* factor out caching of the client oauth idJesse Luehrs2019-10-173-22/+25
|
* use a real enum for auth typesJesse Luehrs2019-10-174-26/+76
|
* fix session list testsJesse Luehrs2019-10-171-0/+1
|
* display number of watchers per stream in the watch uiJesse Luehrs2019-10-163-7/+51
|
* optimize the protocol a bit moreJesse Luehrs2019-10-162-39/+55
|
* handle resets and clears separatelyJesse Luehrs2019-10-161-23/+44
|
* use different logic for buffering on the client side vs serverJesse Luehrs2019-10-164-68/+154
| | | | | they have different needs and performance profiles, so try to be a bit more intelligent
* ensure that all bytes coming from the terminal actually get writtenJesse Luehrs2019-10-164-94/+135
| | | | | | | | | it's not actually safe to just skip over bytes when we see a reset escape sequence, because there might be other escape sequences changing the terminal state that don't get cleared by a reset. for instance, starting vim sends "switch to alternate screen" followed by "reset", and if those two come in the same packet, we were previously dropping the "switch to alternate screen" escape sequence, causing incorrect display.
* simplifyJesse Luehrs2019-10-163-14/+13
|
* simplifyJesse Luehrs2019-10-162-22/+21
|
* remove some unnecessary movesJesse Luehrs2019-10-161-2/+2
|
* cache refresh tokens and use them when the existJesse Luehrs2019-10-1612-53/+331
| | | | to avoid needing to go through the browser auth flow every time
* further generalize the oauth flowJesse Luehrs2019-10-162-35/+69
|
* better error handling in client oauth login flowJesse Luehrs2019-10-162-17/+35
|
* move address default out of the util functionJesse Luehrs2019-10-163-6/+5
|
* add a bit more debug logging to the clientJesse Luehrs2019-10-161-5/+22
|
* don't block the main thread when waiting for oauth responseJesse Luehrs2019-10-162-72/+142
|
* rename some enum variantsJesse Luehrs2019-10-161-7/+7
| | | | more consistency with the server side
* commentJesse Luehrs2019-10-161-1/+1
|
* pull address out into a constJesse Luehrs2019-10-161-3/+10
|
* relax the http request parsing a bitJesse Luehrs2019-10-161-1/+1
|
* start implementing oauth authenticationJesse Luehrs2019-10-1511-55/+1308
|
* don't send terminal output messages unless we're connectedJesse Luehrs2019-10-151-1/+5
| | | | | the server won't be expecting terminal output messages during the process of a login flow, for instance
* need to reset rsock after closing the connection tooJesse Luehrs2019-10-151-0/+1
| | | | | since the future that produced it completed at this point, so we don't want to poll it again (which could happen since closing is deferred)
* reset reconnect timer after logging inJesse Luehrs2019-10-151-2/+1
| | | | | resetting after connect means that we could still potentially crashloop during a multi-phase login
* add an intermediate LoggingIn stateJesse Luehrs2019-10-151-3/+34
| | | | for login methods that require additional messages back and forth
* add a processing state to server connectionsJesse Luehrs2019-10-151-8/+50
| | | | this allows us to do long-lived actions in response to messages
* send initial terminal size on process startJesse Luehrs2019-10-154-18/+34
| | | | | | sending it on connect meant that the terminal size wouldn't be set at all if there was a problem connecting to the server, which confuses programs running in the pty
* add client message logging tooJesse Luehrs2019-10-151-0/+3
|
* wait to send on_connect messages until after loginJesse Luehrs2019-10-151-6/+6
|
* allow clippy::single_match everywhereJesse Luehrs2019-10-152-2/+1
|
* various client cleanupsJesse Luehrs2019-10-151-91/+99
|
* delay moving to the connected state until we get an ack from the serverJesse Luehrs2019-10-153-3/+35
| | | | | this should make adding additional handshake steps during login more straightforward
* store term type and size data on the clientJesse Luehrs2019-10-151-9/+15
|
* more simplificationJesse Luehrs2019-10-141-8/+11
|
* a few more tweaksJesse Luehrs2019-10-145-53/+56
|
* fix tls poll implementationJesse Luehrs2019-10-141-1/+1
|
* commentJesse Luehrs2019-10-141-0/+1
|
* and more simplifyingJesse Luehrs2019-10-141-63/+60
|