[Bug 17195] NamedPipe datagrams need to be _really_ datagrams

[wine-bugs at winehq.org]

http://bugs.winehq.org/show_bug.cgi?id=17195





--- Comment #16 from Luke Kenneth Casson Leighton <lkcl at lkcl.net>  2009-02-07 08:39:25 ---
tests TODO:

1) thread multiwrite, thread multiread.

  this test will show whether it's possible to have multiple concurrent readers
and writers at the _same_ time.  i.e.:

  wth-1 : write starts (e.g. 2mb of data)
  wth-2 : write starts (e.g. 2mb of data)
  rth-1 : read starts whilst wth-1 and wth-2 are still writing.
  rth-2 : read starts of SECOND message whilst wth-1 and wth-2 are still
          writing, and rth-1 is still reading message number 1.

2) start off in non-message-mode and switch to message-mode with data still
unread

 this test will be interesting.  basically, the SetNamedPipeHandleState is to
be delayed until a few messages have been exchanged, and some data left in the
pipe. a special focus will be on determining whether the data already in the
pipe is:

  a) thrown away
  b) "turned into" a message-mode message
  c) prepended to the next message that's written.

3) start off in message-mode and switch to non-message-mode with data
outstanding and not outstanding, as two separate tests.

4) do a CreateNamedPipe in message-mode on server-end and don't set
message-mode on the client, and see what happens

5) do a CreateNamedPipe _not_ in message-mode on server-end, and set
message-mode on the server, and see what happens.

6) more - yet to be thought of.


-- 
Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email
Do not reply to this email, post in Bugzilla using the
above URL to reply.
------- You are receiving this mail because: -------
You are watching all bug changes.