Expect for Windows NT Buglist
Expect for Windows NT Buglist
Beta 1 Buglist
- October 1, 1998: At some point, output from Kermit95 is no longer
captured. Somehow, it seems to be missing the breakpoints that
capture data. It might be caused by exceptions occurring, or some
multithreaded issue might be coming into play. This one is going
to require some significant debugging, and I don't know when I'll
get to it.
- 100% of CPU could be used in some cases. This was due to a bug in
the event driven pipe code. Fixed in snapshot 21
- When using expectclib, there is an occasional problem when reading
data. It reads the data, but then it checks against NULL. I have
yet to track down when this happens. Possibly fixed in snapshot 19
- When using expectclib, the console reading thread would be started.
This caused bad interaction with calls to gets(). Fixed in snapshot 16.
Alpha 2 Buglist
- Serial communications may report an end of file condition before it
actually exists. Fixed in Beta 1.
- If more than 4K of data gets send to a logging function, expect crashes.
Fixed in Beta 1.
- If stdin/stdout handles are the same, Tcl would crash.
Fixed in Beta 1.
- Serial communication does not work properly. Fixed in Beta 1.
- send_user didn't work properly after the spawn_id was changed.
Fixed in Beta 1.
- With spawn -pipes, when stdout or stderr is closed in the slave, no
more output is read from the other. Fixed in Beta 1.
- Output from console based emacs could not be caught.
Support added in Beta 1.
- send -null always returns an error. Fixed in Beta 1.
- Pipe reading does not always drain all output from subprocess before
an eof. Fixed in Beta 1.
- Subprocesses of spawned processes do not always work correctly.
Fixed in Beta 1.
- eof condition is not properly detected when reading from spawned process.
Fixed in Beta 1.
- Expect breaks the Tcl close commmand. You will get a 'wrong # args'
error when using it. Fixed in Beta 1.