Le 15/01/2012 09:16, Daniel Shahaf a écrit :
Forwarding back to list.
dcz wrote on Sun, Jan 15, 2012 at 09:02:06 +0100:
Le dimanche 15 janvier 2012 00:48:01, Daniel Shahaf a écrit :
dcz wrote on Tue, Jan 10, 2012 at 13:55:52 +0100:
In file
'D:\Development\SVN\Releases\TortoiseSVN-1.7.3\ext\subversion\subversion\libsvn_client\diff.c'
line 1651: assertion failed (*target1&& *target2)
Those errors virtually always represent a bug in Subversion.
I do not find a record of this particular error in the last 6 months of
issues@/users@ mail. Please search the issue tracker, and if there is
not already a bug logged about this problem, then file one. If you can,
please include a complete reproduction recipe (starting from creating
a repository:
https://svn.apache.org/repos/asf/subversion/site/publish/docs/community-guide/repro-template.bat).
Thanks!
Apparently, there is no report on this :
http://subversion.tigris.org/issues/buglist.cgi?long_desc_type=fulltext&long_desc=1651
(I tried many keywords combos before narrowing down to the line #)
Though unfortunately, I honestly have no idea of the actual svn
commands triggered by tortoise on this particular operation, so I
don't know where to start to fill the issue template.
I already asked for this in the tortoise svn ML, with no luck so far.
The step by step procedure on tortoise is :
1 - open the revision graph
2 - select tag& last trunk rev
3 - compare revision
4 - select all files listed
5 - unselect some of them (like I said, few files from within two
directories)
6 - ask for the diff for this subset of file (note, this was not a
feature in tortoise 1.6.x)
7 - boom, get the error message
Problem is, I'm not even 100% sure that the operation was not in the
end successful since the error does not break the operation and the
diff is in the end displayed.
The working copy went upgraded from latest 1.6 to 1.7.3 with no
errors, and I was able to commit / update many times before I
eventually ran into this issue.
I will cross post this mail in the hope someone from tortoise will
provide with more details.
Cheers
Fabrice
Oups, I hit the wrong button sorry.
Anyway, I investigated a bit more on this and found out something new.
The described operation indeed works perfectly if there are no deleted
files between the two rev (the tag part does not seems to matter).
When it fails, it actually only fails when comparing from the comparison
list, if I directly ask for a diff from the rev graph, it works without
errors.
When using the comparison window, it actually fails even if I select one
single deleted file and ask for the diff, but if I select this same file
and ask for a compare, it opens my editor in compare mode with one empty
side (as expected).
When I ask for a diff on all file listed, I get what seems to be a valid
diff but with no reference to the deleted files.
And, I could not make a compare between the diff obtained while
selecting all files (the one with errors) and the same diff obtained
from the rev graph because the file order was no the same, but it really
looks like only the deleted files diffs are missing.
Hope this helps
Fabrice
