Conference bulova::decw_jan-89_to_nov-90

2677.0. "Weird clipboard problem on multihead system" by EPIK::J_JOSEPH (Living in the Pasture) Fri Apr 27 1990 10:50

I'm running into some trouble with the clipboard.  This trouble only seems to
occur when cutting and pasting across a two headed system.  This trouble also
only seems to occur when the data takes up a large number of bytes.  But since
I am cutting and pasting color images stored in DDIF format, this happens
often.

Here is what happens.  First, I copy a ddif frame to the clipboard - this
works fine.  Then, from another instance of my program (paint) I copy the
data from the clipboard (using DwtInquireNextPasteLength and
DwtCopyFromClipboard).  If the instance of paint which is doing the copying
is on the same head of the two headed system everything works fine.  If it
is on the other head, and the data is < (roughly 2^14 bytes) it works fine.  If
however, it is on the other head and the data is > 2^14 bytes, then
DwtCopyFromClipboard returns a status code of ClipboardNoData.

Now, I could survive with this as a limitation, but when this happens -
something quite unexpected happens, the paint which attempted the copy goes
about its normal business after informing the user that there was nothing on
the clipboard to paste (that was expected), but the instance of paint which
copied the data to the clipboard in the first place is now hung!!??!?!
This instance of paint remains hung until the other instance is exited, at
which point, some Xlib errors are generated (ones that say: not a valid
window ID).

Also, I tried doing a paste a second time from the unhung paint, and that
one hung as well.

Can anyone help me out here.  This looks like a clipboard bug to me.

-Jonathan

The way I came up with 2^14 bytes is that one instance where the size of the
data was 16270 bytes (as well as many instances where there was less data)
worked.  An instance where the data was 16894 bytes (as well as many
instances where there was more data) failed.  2^14 = 16384 which is between
those two.

I also should point out that in my test cases, one of the heads has always
been a monochrome device, and the other an 8 plane color device.  The
problem can be seen in both directions.