Ben Collins-Sussman <sussman@newton.collab.net> writes:
> Perhaps I'll chase this bug down after all... it might not be in your
> editor, Greg. It's possible that it could be in my editor-driver
> (adm_crawler.c). I'll keep you posted.
OK, I just carefully stepped through adm_crawler, and it seems to be
driving the editor correctly.
Then I did another experiment: I made the commit command use the
"test-editor" in tests-common/, the one that prints plain english,
instead of the xml-output editor. The normally-segfaulting commit
now works fine, and looks kosher:
-----------------------------------
Attempting to commit.
REPLACE_ROOT: name '/blah', revision '37'
REPLACE_DIR: name 'A', ancestor 'anni/A' revision 1
REPLACE_DIR: name 'D', ancestor 'anni/A/D' revision 1
REPLACE_DIR: name 'H', ancestor 'anni/A/D/H' revision 1
REPLACE_FILE: name 'omega', ancestor 'anni/A/D/H/omega' revision 1
REPLACE_FILE: name 'psi', ancestor 'anni/A/D/H/psi' revision 1
CLOSE_DIR '/blah/A/D/H'
CLOSE_DIR '/blah/A/D'
CLOSE_DIR '/blah/A'
CLOSE_DIR '/blah'
TEXT-DELTA on file 'omega':
txdelta window: source text: offset 0, length 25
txdelta window: new text (2 bytes): z
end of windows
CLOSE_FILE 'omega'
TEXT-DELTA on file 'psi':
txdelta window: source text: offset 0, length 23
txdelta window: new text (16 bytes): new text for pi
end of windows
CLOSE_FILE 'psi'
EDIT COMPLETE.
------------------------------------
(Man, I *love* being able to just swap editors in a pinch!)
Anyway -- Greg H. -- I'm now pretty convinced that there's something wrong
in xml_output.c. The ball's in your court.
Enough of this; I've got other stuff to do. :)
Received on Sat Oct 21 14:36:15 2006