No, I definitely don't mind -- it gives me something to start with. ;)
Thanks for the info.
e
Ingo Weinhold wrote:
On 2003-08-26 at 18:38:04 [+0200], Erik Jaesler wrote:
Yeah, that sounds like exactly the thing. I monkeyed about some last
night with my messaging test to have the second looper reply to the
first and it just sat there. Clearly I have some debugging to do. ;)
I tried a bit more to get the RosterShell running again and fixed a few
warnings (I think, I'll enable warnings in the build by default), one or two
revealing actual bugs -- I hope you don't mind. The RosterShell still isn't
working, but I rather stopped here, before getting in your way.
The problem I ran into is that in BMessage::Unflatten() the dataLen read from
the flattened message seems to be invalid, which causes the memory allocation
to fail and a B_NO_MEMORY to be thrown (it probably shouldn't be thrown but
just returned (the same in two other places in Unflatten()) -- I left it that
way, since it is excellent for debugging :-).
If you want to reproduce it, you can run the RosterShell (in
test/servers/registrar; after the registrar) and type `l[ist]'. It will choke
on Unflatten()ing the reply message from the registrar, when a
BRoster::GetRunningAppInfo() is invoked (size is > 255).
