I have a patch that passes make check. But a test suite can't test
everything (although the status code is probably one of the best covered
due to it's part in almost every cmdline-test). Posting only the changes
What I'm wondering about:
* I'm using SVN_INVALID_REVNUM() for determining if read_info() gives us
a valid revision. If it does I assume that we don't have to check for
base_shadowed, addition or deletion. If the path used as IN parameter
to read_info() is deleted, we will get a valid revision. Correct?
* svn_wc__db_scan_deletion() checks work_del_abspath. That works for all
the cases in the test suite. I assumed that a plain delete would set
the work_del_abspath to the root but this part of the doc comment of
scan_deletion() says something else. Am I misinterpreting this (e.g.
it's a part of the earlier paragraphs and has more operations done) or
can it really be that a plain delete does not set the work_del_abspath?
If B/W/D does not exist in the WORKING tree (we're only talking about a
deletion of nodes of the BASE tree), then deleting B/W/D would have
marked the subtree for deletion. BASE_DEL_ABSPATH will refer to B/W/D,
BASE_REPLACED will be FALSE, MOVED_TO_ABSPATH will be NULL, and
WORK_DEL_ABSPATH will be NULL.
* For all cases where read_info() says we have an addition (and by that
read_info() means copied-here and moved-here too) I've set the
revision to -1. Correct? Until committed, no revision exists is my
Received on 2010-04-19 18:15:28 CEST