[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: 1.9.0-rc2 up for testing/signing

From: Branko Čibej <brane_at_wandisco.com>
Date: Wed, 03 Jun 2015 15:26:26 +0200

On 03.06.2015 14:57, Ivan Zhakov wrote:
> On 3 June 2015 at 15:35, Branko Čibej <brane_at_wandisco.com> wrote:
>> On 03.06.2015 14:24, Ivan Zhakov wrote:
>>> On 3 June 2015 at 14:37, Branko Čibej <brane_at_wandisco.com> wrote:
>>>> On 02.06.2015 20:05, Branko Čibej wrote:
>>>>> On 02.06.2015 12:45, Daniel Shahaf wrote:
>>>>>> Ben Reser wrote on Sun, May 31, 2015 at 14:28:39 -0700:
>>>>>>> The 1.9.0-rc2 release artifacts are now available for testing/signing.
>>>>>>> Please get the tarballs from
>>>>>>> https://dist.apache.org/repos/dist/dev/subversion
>>>>>>> and add your signatures there.
>>>>>>> Thanks!
>>>>>> Note that Evgeny reported a regression in svn_repos_verify_fs2() in
>>>>>> <http://svn.haxx.se/dev/archive-2015-05/0141.shtml>. No objections to
>>>>>> moving forward with rc2, but as that issue is a regression, we'll need
>>>>>> an rc3 that fixes it.
>>>>> Yes ... and the patch has been reviewed but not committed. I believe it
>>>>> only needs a couple tweaks (fixing an "if" condition and removing the
>>>>> unused error code).
>>>> I have a more complete fix based on Evgeny's patch, running tests now.
>>>> It turns out that we still need a new error code for the summary
>>>> results, but with a different meaning and therefore different name.
>>>> Renaming it had a positive side effect as it turned out that we were
>>>> emitting the SVN_ERR_REPOS_CORRUPTED error from FSFS and FSX instead of
>>>> the correct SVN_ERR_FS_CORRUPT.
>>> Another option would be to require notify_func for
>>> svn_repos_verify_fs3() and always report errors through notify_func.
>>> This would make error reporting consistent whether keep_going TRUE or
>>> FALSE. For svn_repos_verify_fs2() we could create compat notify_func
>>> handler that converts notifications to errors.
>> See r1683311; I believe error reporting is consistent now.
>> Adding complex logic to the deprecated function isn't such a good idea, IMO.
> Maybe, but it's better than adding complex logic to current
> (not-deprecated) function. IMO it's not svn_repos_verify_fs3()
> responsibility to generate verification summary -- it should be done
> at the UI level. I.e. in svnadmin or other application using
> svn_repos_verify_fs3() API.

But that would imply either returning a bunch of extra info from
svn_repos_verify_fs3, or counting notifications in the callers (which
means making the callers depend on implementation details of the API).

Note that the summary will only be generated if an error occurred during
verification in keep-going mode, when we have to return some kind of
error anyway; we may as well put as much info as we have available into
the error message.

-- Brane
Received on 2015-06-03 15:26:43 CEST

This is an archived mail posted to the Subversion Dev mailing list.