Using M6 builds- 1999050508 for Win32 & 1999050513 on linux (we don't have a working Mac build yet to try. When you select Get Msg from the Messenger window, the count changes in the Sidebar when new msgs are retrieved but the message/s aren't displayed in the Thread window until you go to another folder then back to the inbox. This is regression from M5 builds. 1. Launch Messenger 2. Send yourself a message 3. Click Get Msg and see that the message count changes for the Inbox, but the message doesn't show up in the Thread pane. To see the message you have to close the application or change folders, then return to the Inbox.

This is an M6 stopper (regression), but not a build acceptance stopper since we can still test with this bug.

cc'ing waterson. This is happening because OnAssert in the composite datasource is suddenly asking if the assertion really exists before telling its observer to really assert this new value. We currently haven't implemented HasAssertion. I recognize this needs to get implemented, but is there a reason why we're being asked if our OnAssert is really valid? For the moment I'm just going to say that if the source is the type I'm expecting that the assertion is always true, but it seems like this extra step could be redundant in this situation.

OK, I can't do what I mentioned because you can't lie to HasAssertion. I'll have to make it actually determine whether or not the assertion is correct. This is more of an interface issue, but it seems like if I implement this correctly, this is going to slow us down if this function gets called a lot.

per pmock, also happens on Mac build 1999050515-M6

I have a fix in my tree for this that I need to clean up a bit. I'll probably check this in when the tree opens tomorrow, so this most likely won't be in tomorrow's build.

ACk. Sorry. I checked in a change to composite datasource that I was messing around with. The change is not really necessary. Scott: if this is causing you trouble I can back it out.

I implemented HasAssertions for the datasource that was having problems. It's probably a good thing for me to have implemented anyway. In this case it just seemed redundant since I was telling you I was adding it and then you verified it immediately afterwards. If it turns out you don't need it, it will speed things up to take it out.

I checked in the fix.

RE: linux build (1999-05-11-08-m6) This problem is fixed on the linux build

Esther - check on Win32 and post results here. We can't verify on the Mac due to bug 6106. After 6106 is fixed, then we'll mark this bug verified if all is well.

Using win32 build 1999051109M6 this is fixed. Waiting for 6106 to be fixed to retest on Mac.

Using 19990517 build on Mac Get Msg works multiple times, so I will verify this one. However, note that send doesn't seem to work twice in a row so I had to send messages from another system to see Get Msg work right. I will investigate and log a new bug for the send problem.