Re: trunk failing tests on Windows XP (32 bit): prop-tests.py 33, stat-tests.py 5, upgrade-tests.py 11

On Sun, Oct 10, 2010 at 3:07 PM, Johan Corveleyn <jcorvel_at_gmail.com> wrote:
> On Sat, Oct 2, 2010 at 11:24 AM, Bert Huijben <bert_at_qqmail.nl> wrote:
>>> -----Original Message-----
>>> From: Paul Burba [mailto:ptburba_at_gmail.com]
>>> Sent: vrijdag 1 oktober 2010 15:46
>>> To: Bert Huijben
>>> Cc: Johan Corveleyn; Subversion Development
>>> Subject: Re: trunk failing tests on Windows XP (32 bit): prop-tests.py
>>> 33, stat-tests.py 5, upgrade-tests.py 11
>>>
>>> On Thu, Sep 30, 2010 at 8:10 PM, Bert Huijben <bert_at_qqmail.nl> wrote:
>>> >> -----Original Message-----
>>> >> From: Johan Corveleyn [mailto:jcorvel_at_gmail.com]
>>> >> Sent: vrijdag 1 oktober 2010 1:51
>>> >> To: Subversion Development
>>> >> Subject: trunk failing tests on Windows XP (32 bit): prop-tests.py
>>> 33,
>>> >> stat-tests.py 5, upgrade-tests.py 11
>>> >>
>>> >> Hi devs,
>>> >>
>>> >> The following tests fail on my machine (Windows XP 32-bit, built
>>> with
>>> >> VCE 2008, ra_local):
>>> >>
>>> >> - prop-tests.py 33: test properties of obstructed subdirectories
>>> >> svn: Can't open directory
>>> >>
>>> 'C:\research\svn\client_build\svn_trunk\Release\subversion\tests\cmdlin
>>> >> e\svn-test-work\working_copies\prop_tests-33\A\C':
>>> >> The directory name is invalid.
>>> >>
>>> >> - stat-tests.py 5: status on versioned items whose type has changed
>>> >> svn: Can't open directory
>>> >>
>>> 'C:\research\svn\client_build\svn_trunk\Release\subversion\tests\cmdlin
>>> >> e\svn-test-work\working_copies\stat_tests-5\A':
>>> >> The directory name is invalid.
>>> >>
>>> >> - upgrade_tests.py 11: missing directories and obstructing files
>>> >> svn: Can't open directory
>>> >>
>>> 'C:\research\svn\client_build\svn_trunk\Release\subversion\tests\cmdlin
>>> >> e\svn-test-work\working_copies\upgrade_tests-11\A\D':
>>> >> The directory name is invalid.
>>> >>
>>> >> They all seem to try to open some path as a directory, but it isn't
>>> a
>>> >> directory (but an empty file or something).
>>> >>
>>> >> Am I the only one seeing this? Is this known/normal?
>>> >
>>> > This seems to be related to the APR version you use. (Paul Burba
>>> reported
>>> > this same problem earlier this week).
>>>
>>> Hi Johan,
>>>
>>> Bert is correct, I saw these failure earlier this week while using APR
>>> 1.3.12.  All the failures occur when a file unexpectedly obstructs a
>>> directory.  The APR macro APR_STATUS_IS_ENOTDIR in 1.3.x does not
>>> handle the "ERROR_DIRECTORY 267 The directory name is invalid" that
>>> Windows raises in this case.  This was fixed in APR in
>>> http://svn.apache.org/viewvc?view=revision&revision=821306 and is in
>>> APR 1.4.2.
>>
>> I think we should just add this error to the SVN__APR_STATUS_IS_ENOTDIR() macro to fix compatibility with older apr versions, but I would like to know why this didn't fail a few weeks ago. (I haven't been very active for the last week, but I haven't seen any relevant commits on the commit list in the last few weeks)
>
> FWIW:
> I don't know if anybody else has tried to pinpoint when/why this
> started to fail on trunk, but I've taken a look. It seems to have
> started already quite some time ago: it started with r991236 (31
> August), which was the enabling of single-DB mode.
>
> I didn't have time yet to investigate further with SVN_WC__SINGLE_DB
> and SINGLE_DB
> defined to see which commit caused it with single-db already enabled
> before r991236. If nobody beats me to it, I'll continue the search
> when I have some more time next week.
>
> Gut feeling: maybe it has something to do with the actual removal of
> directories-scheduled-for-deletion in single-db, as opposed to the
> wc-1 behavior?

I saw that the problem was fixed in r102170 by Bert, by changing the
SVN__APR_STATUS_IS_ENOTDIR() macro. Thanks, the tests all pass now,
with APR 1.3.8 on my WinXP.

Now, some root-cause analysis :-), in case this is still useful...

Going further back than r991236, but with SVN_WC__SINGLE_DB and
SINGLE_DB explicitly defined, I found the following:

1) prop_tests.py 33
- First fails in r980406 with "(Can't open directory ... : The
directory name is invalid.)"
- Before r980406, it fails with a more normal failure: "EXCEPTION:
SVNTreeUnequal".

2) stat_tests.py 5
- First fails in r990818 with "(Can't open directory ... : The
directory name is invalid.)"
- Before r990818, the test succeeded.

3) upgrade_tests.py 11
- First fails in r991160 with "(Can't open directory ... : The
directory name is invalid.)"
- Before r991160, it fails with an assertion: "(svn: In file
'..\..\..\subversion\libsvn_wc\wc_db.c'
line 6558: assertion failed ((pdh)->wcroot != NULL &&
(pdh)->wcroot->format == SVN_WC__VERSION))".

Now it's up to someone else to take a closer look at those revisions,
and try to figure out why they introduced the "directory name is
invalid" failures. That's a bit out of my league :-).

Cheers,

-- 
Johan