On 29 June 2014 22:56, Bert Huijben <bert_at_qqmail.nl> wrote:
>
>
>> -----Original Message-----
>> From: ivan_at_apache.org [mailto:ivan_at_apache.org]
>> Sent: vrijdag 27 juni 2014 20:45
>> To: commits_at_subversion.apache.org
>> Subject: svn commit: r1606182 - in /subversion/trunk/subversion: svn/
>> tests/cmdline/
>>
>> Author: ivan
>> Date: Fri Jun 27 18:44:51 2014
>> New Revision: 1606182
>>
>> URL: http://svn.apache.org/r1606182
>> Log:
>> Follow-up to r1604179: Print "Committing transaction" progress notification
>> for tree changes only commits.
>>
>> Found by: julianfoad
>
> I don't understand why (yet), but this patch breaks quite a few tests on most buildbots.
>
> The pattern I don't understand is:
> [[[
[...]
> FAIL: merge_tests.py 59: merge -rold(before rename):head renamed file
> ]]]
>
> The expected and actual output appear to be identical here.
>
Fixed in r1606648. Thanks!
> If we are going to fix these cases anyway I would prefer an additional patch like
> [[
> Index: subversion/svn/notify.c
> ===================================================================
> --- subversion/svn/notify.c (revision 1606560)
> +++ subversion/svn/notify.c (working copy)
> @@ -1061,7 +1061,7 @@ notify_body(struct notify_baton *nb,
> {
> SVN_ERR(svn_cmdline_printf(pool, _("done\n")));
> }
> - SVN_ERR(svn_cmdline_printf(pool, _("Committing transaction...")));
> + SVN_ERR(svn_cmdline_printf(pool, _("Committing transaction...\n")));
> break;
>
> default:
> Index: subversion/svn/util.c
> ===================================================================
> --- subversion/svn/util.c (revision 1606560)
> +++ subversion/svn/util.c (working copy)
> @@ -76,7 +76,7 @@ svn_cl__print_commit_info(const svn_commit_info_t
> apr_pool_t *pool)
> {
> if (SVN_IS_VALID_REVNUM(commit_info->revision))
> - SVN_ERR(svn_cmdline_printf(pool, _("\nCommitted revision %ld%s.\n"),
> + SVN_ERR(svn_cmdline_printf(pool, _("Committed revision %ld%s.\n"),
> commit_info->revision,
> commit_info->revision == 42 &&
> getenv("SVN_I_LOVE_PANGALACTIC_GARGLE_BLASTERS")
> ]]
> To remove the magic coupling of the EOL handling between these two functions.
> (This will need additional test fixups to remove several empty lines that we expect in older cases)
>
I considered such change, but found some other coupling between
different notifications and decided to look to this later.
--
Ivan Zhakov
Received on 2014-06-30 11:03:28 CEST