Thanks Evgeny, please use the ML to which I'm forwarding your last
message, that has a much wider audience. I cannot do it right now, but
I'll try to understand the issue.
ciao, lele.
--- Begin Message ---
Lele Gaifax <[EMAIL PROTECTED]> writes:
> For the Cvs.testCvsToSubversion problem, maybe the SF project changed
> something, it fails for me as well.
I've looked into it some more, and it appears that the
normalize_cvs_rev()/compare_cvs_rev() may not always Do The Right
Thing when comparing a branchpoint to a branch.
To repro:
* unpack the tailor-0.9.20 distribution.
* put the file ftrace.py (attached) into the tailor-0.9.20/vcpx
subdirectory
* patch the file tailor-0.9.20/vcpx/cvsps.py with the patch that follows:
diff -c /home/zhenya/tailor-0.9.20/vcpx/.\~/cvsps.py.\~1\~
/home/zhenya/tailor-0.9.20/vcpx/cvsps.py
--- /home/zhenya/tailor-0.9.20/vcpx/.~/cvsps.py.~1~ Fri Dec 23 08:08:00 2005
+++ /home/zhenya/tailor-0.9.20/vcpx/cvsps.py Wed Feb 22 18:56:53 2006
@@ -18,6 +18,10 @@
InvocationError
from target import SyncronizableTargetWorkingDir, TargetInitializationFailure
+from ftrace import trace, ltrace, untrace
+
+trace('compare_cvs_revs', 'normalize_cvs_rev')
+
class EmptyRepositoriesFoolsMe(Exception):
"Cannot handle empty repositories. Maybe wrong module/repository?"
@@ -352,8 +356,9 @@
for m in cs.entries:
info = entries.getFileInfo(m.name)
if info:
- odversion = info.cvs_version
- applied = compare_cvs_revs(odversion, m.new_revision) >= 0
+ print "m.name", m.name, "oldversion", info.cvs_version,
"m.new_revision", m.new_revision
+ oldversion = info.cvs_version
+ applied = compare_cvs_revs(oldversion, m.new_revision) >= 0
if not applied:
break
return applied
* run the tailor test case like this:
./tailor test -v Cvs.testCvsToSubversion 1>cvs2svn.stdout 2>cvs2svn.stderr
* Look at cvs2svn.stdout;
Here's a small chunk that, I believe, illustrates the problem:
m.name skins/CMFEditions/versions_history_form.pt.metadata oldversion 1.1
m.new_revision 1.1.2.1
compare_cvs_revs (revstr1='1.1', revstr2=u'1.1.2.1')
normalize_cvs_rev (rev='1.1')
normalize_cvs_rev (rev='1.1') ==> (1, 1)
normalize_cvs_rev (rev=u'1.1.2.1')
normalize_cvs_rev (rev=u'1.1.2.1') ==> (1, 1, 2, 1)
compare_cvs_revs (revstr1='1.1', revstr2=u'1.1.2.1') ==> -1
Isn't version 1.1 of this file the branch point from which 1.1.2.1 was
derived? And, if so, shouldn't the comparison return something other
than -1?
That is, doesn't the revision history look like this?
----- 1.1 -- 1.2 -- 1.3 --->
\
\--- 1.1.2.1 -- 1.1.2.2 -->
It's entirely possible I am just confused on this point (I haven't
studied the code in normalize_cvs_rev() and compare_cvs_revs() yet,
sorry!)
Thank you for your time!
--
Evgeny
ftrace.py
Description: A file to add a primitive/simple-minded \"tracing\" of functions to Python
--- End Message ---
_______________________________________________
Tailor mailing list
[email protected]
http://lists.zooko.com/mailman/listinfo/tailor
