Re: Current httpd24threading branch problems?
Jan Kaluža <[email protected]>
| Newsgroups | gmane.comp.apache.mod-perl.devel |
|---|---|
| Message-ID | <[email protected]> |
On 05/14/2014 02:07 PM, Steve Hay wrote: > On 14 May 2014 09:27, Jan Kaluža <[email protected]> wrote: >> On 05/13/2014 07:29 PM, Steve Hay wrote: >>> >>> On 29 April 2014 11:40, Jan Kaluža <[email protected]> wrote: >>>> >>>> On 04/17/2014 09:15 AM, Steve Hay wrote: >>>>> >>>>> >>>>> On 17 April 2014 07:46, Jan Kaluža <[email protected]> wrote: >>>>>> >>>>>> >>>>>> Hi Fred, >>>>>> >>>>>> could you please sum up what's the current state of httpd24threading >>>>>> branch? >>>>>> I think you're the only one currently who knows what has to be done >>>>>> before >>>>>> we can release it somehow. >>>>>> >>>>>> I've tried to read the mailing list to refresh my memory, but I'm not >>>>>> sure >>>>>> what has been fixed/done already during these 3 months. >>>>>> >>>>> >>>>> I think it's mostly been me and you working on it, actually! >>>>> >>>>> I've said before that I need to fix some Windows test failures and >>>>> then (assuming *nix doesn't have any outstanding of its own) I'll be >>>>> happy to merge this branch into trunk and we can get a release going. >>>>> >>>>> My current list of failures is: >>>>> >>>>> Test Summary Report >>>>> ------------------- >>>>> t\compat\conn_rec.t (Wstat: 0 Tests: 2 Failed: 0) >>>>> [new failure for 2.4.x] >>>>> Parse errors: Bad plan. You planned 4 tests but ran 2. >>>>> t\modperl\local_env.t (Wstat: 0 Tests: 6 Failed: 1) >>>>> Failed test: 6 >>>>> t\modperl\merge.t (Wstat: 0 Tests: 10 Failed: 3) >>>>> Failed tests: 3, 6, 9 >>>>> t\modperl\merge2.t (Wstat: 0 Tests: 10 Failed: 3) >>>>> Failed tests: 3, 6, 9 >>>>> t\modperl\merge3.t (Wstat: 0 Tests: 10 Failed: 3) >>>>> Failed tests: 3, 6, 9 >>>>> t\modules\cgi.t (Wstat: 0 Tests: 5 Failed: 5) >>>>> Failed tests: 1-5 >>>>> t\modules\cgi2.t (Wstat: 0 Tests: 5 Failed: 5) >>>>> Failed tests: 1-5 >>>>> t\modules\cgipost.t (Wstat: 0 Tests: 6 Failed: 5) >>>>> Failed tests: 2-6 >>>>> t\modules\cgipost2.t (Wstat: 0 Tests: 6 Failed: 5) >>>>> Failed tests: 2-6 >>>>> t\modules\cgiupload.t (Wstat: 0 Tests: 2 Failed: 2) >>>>> [with LWP only] >>>>> Failed tests: 1-2 >>>>> t\modules\cgiupload2.t (Wstat: 0 Tests: 2 Failed: 2) >>>>> [with LWP only] >>>>> Failed tests: 1-2 >>>>> t\protocol\echo_block.t (Wstat: 0 Tests: 3 Failed: 2) >>>>> [new failure for 2.4.x] >>>>> Failed tests: 2-3 >>>>> t\protocol\echo_nonblock.t (Wstat: 0 Tests: 3 Failed: 1) >>>>> [new failure for 2.4.x] >>>>> Failed test: 2 >>>>> t\protocol\echo_timeout.t (Wstat: 0 Tests: 5 Failed: 4) >>>>> [new failure for 2.4.x] >>>>> Failed tests: 2-5 >>>>> t\protocol\pseudo_http.t (Wstat: 0 Tests: 13 Failed: 9) >>>>> [new failure for 2.4.x] >>>>> Failed tests: 3-8, 11-13 >>>>> Files=252, Tests=2938, 848 wallclock secs ( 2.26 usr + 0.36 sys = 2.62 >>>>> CPU) >>>>> >>>>> The ones marked 'new failure for 2.4.x' fail with 2.4.x but not with >>>>> 2.2.x so they are particularly worrying for 2.4.x support, but note >>>>> also that *none* of the above tests fail with 2.2.x when using trunk >>>>> (or mod_perl-2.08), which is also quite worrying. >>>>> >>>>> What is the current state of play on *nix? Which tests, if any, fail >>>>> using: >>>> >>>> >>>> >>>> Hi, >>>> >>>> I've had finally some time to do the tests. It's run on Fedora 20, >>>> httpd-2.4.9 and httpd-2.2.23. See below. >>>> >>>>> (1) httpd24threading + httpd-2.4.x >>>> >>>> >>>> >>>> t/compat/conn_rec.t (Wstat: 0 Tests: 2 Failed: 0) >>>> >>>> Parse errors: Bad plan. You planned 4 tests but ran 2. >>>> >>>> ^ This is expected problem for now. It's caused by renamed "remote_ip" >>>> and >>>> "remote_addr" in httpd-2.4. I was not able to put proper version check in >>>> this test (probably my lack of Perl knowledge). >>>> >>>> t/modules/cgi.t (Wstat: 0 Tests: 5 Failed: 2) >>>> Failed tests: 2, 5 >>>> t/modules/cgi2.t (Wstat: 0 Tests: 5 Failed: 1) >>>> Failed test: 3 >>>> t/modules/cgipost.t (Wstat: 0 Tests: 6 Failed: 1) >>>> Failed test: 4 >>>> t/modules/cgipost2.t (Wstat: 0 Tests: 6 Failed: 1) >>>> Failed test: 4 >>>> t/modules/cgiupload.t (Wstat: 0 Tests: 2 Failed: 1) >>>> Failed test: 1 >>>> t/modules/cgiupload2.t (Wstat: 0 Tests: 2 Failed: 1) >>>> Failed test: 2 >>>> >>>> ^ All these are caused by r1491887 - >>>> /perl/modperl/trunk/t/modperl/local_env.t. Should we revert this commit? >>>> >>>>> (2) httpd24threading + httpd-2.2.x >>>> >>>> >>>> >>>> t/api/server_const.t (Wstat: 0 Tests: 6 Failed: 2) >>>> Failed tests: 5-6 >>>> t/modperl/setupenv2.t (Wstat: 0 Tests: 23 Failed: 7) >>>> Failed tests: 17-23 >>>> t/modules/cgi.t (Wstat: 0 Tests: 5 Failed: 2) >>>> Failed tests: 2, 5 >>>> t/modules/cgi2.t (Wstat: 0 Tests: 5 Failed: 1) >>>> Failed test: 3 >>>> t/modules/cgipost.t (Wstat: 0 Tests: 6 Failed: 1) >>>> Failed test: 4 >>>> t/modules/cgipost2.t (Wstat: 0 Tests: 6 Failed: 1) >>>> Failed test: 4 >>>> t/modules/cgiupload.t (Wstat: 0 Tests: 2 Failed: 1) >>>> Failed test: 1 >>>> t/modules/cgiupload2.t (Wstat: 0 Tests: 2 Failed: 1) >>>> Failed test: 2 >>>> >>>> CGI tests fails for the same reason as in the previous case. Other tests >>>> fail *also* for trunk + httpd-2.2.x (see below). >>>> >>>> >>>>> (3) trunk (or mod_perl-2.08) + httpd 2.2.x >>>> >>>> >>>> >>>> trunk + httpd-2.2.23: >>>> >>>> t/api/server_const.t (Wstat: 0 Tests: 6 Failed: 2) >>>> Failed tests: 5-6 >>>> t/modperl/setupenv2.t (Wstat: 0 Tests: 23 Failed: 7) >>>> Failed tests: 17-23 >>>> t/modules/cgi.t (Wstat: 0 Tests: 5 Failed: 2) >>>> Failed tests: 2, 5 >>>> t/modules/cgi2.t (Wstat: 0 Tests: 5 Failed: 1) >>>> Failed test: 3 >>>> t/modules/cgipost.t (Wstat: 0 Tests: 6 Failed: 1) >>>> Failed test: 4 >>>> t/modules/cgipost2.t (Wstat: 0 Tests: 6 Failed: 1) >>>> Failed test: 4 >>>> t/modules/cgiupload.t (Wstat: 0 Tests: 2 Failed: 1) >>>> Failed test: 1 >>>> t/modules/cgiupload2.t (Wstat: 0 Tests: 2 Failed: 1) >>>> Failed test: 2 >>>> >>>> Same as httpd-2.2.23 + httpd24threading branch. >>>> >>>> >>>>> ? >>>>> >>>>> If you have failures in (1) that aren't in (2), or failures in either >>>>> that aren't in (3) then we need eyes on them to try to get them fixed. >>>>> Perhaps *nix people could focus on failures that are common between >>>>> *nix and Windows, and I'll focus on Windows failures that aren't seen >>>>> on *nix? >>>> >>>> >>>> >>>> From the tests above it looks to me that tests failing for >>>> httpd24threading >>>> branch are not httpd-2.4.x related on Linux. The same tests failing with >>>> httpd-2.2.x and trunk fail also with httpd-2.4.x and httpd24therading >>>> branch. >>>> >>>> I think from my point of view on Linux, it should be possible to merge >>>> httpd24threading branch with trunk without introducing any regression >>>> which >>>> would be visible on our test-suite. >>>> >>>> >>>>> If we can't fix things any time soon then perhaps we should take a >>>>> vote on whether to release as-is, with documentation updated to note >>>>> the currently known problems? >>>> >>>> >>>> >>>> I'm not able to help with Windows specific bugs because of lack of >>>> knowledge >>>> and time for that. But if we fix the bugs mentioned above, maybe we could >>>> release some alpha mod_perl with initial httpd-2.4 support and see what >>>> happens. >>>> >>>> I have no knowledge to evaluate how hard it will be to fix mod_perl with >>>> httpd-2.4 on Windows. If I remember well, even httpd-2.4.x itself had few >>>> 2.4.x releases with some Windows related bugs... >>>> >>>> >>>>> I am keen to see a mod_perl-2.09 with httpd-2.4.x support, but not if >>>>> it's in a buggy state (especially on *nix) that will just generate a >>>>> ton of bug reports and disappointed users. >>>> >>>> >>>> >>>> If the tests are passing and we don't have more people testing trunk, we >>>> will have to consider releasing something to get more testers and more >>>> attention. >>>> >>>> We have httpd24threading branch in current Fedora for some time and there >>>> are no bug reports yet. Some people within Red Hat are using mod_perl >>>> with >>>> httpd-2.4 too. >>>> >>> >>> Thanks for testing, and sorry I've taken so long to get back. >>> >>> So it looks like you're happy on Linux, except for one known failure >>> (compat/conn_rec.t) and a bunch of failures caused by r1491887. >>> >>> I will try testing with that change reverted and see what results I >>> get. Presumably it will fix the same things that it fixes for you; I'm >>> curious to see whether it also fixes anything else. >>> >>> Aside from that, we have two sets of failures on Windows only: >>> >>> - modperl/local_env.t & merge*.t failing with either httpd-2.2 or 2.4, >>> but they work with trunk (on httpd-2.2, obviously) >>> >>> - protocol/echo_block.t, echo_nonblock.t, echo_timeout.t & >>> pseudo_httpd.t failing with httpd-2.4 only >>> >>> I've looked at all of these before and made little headway. I will >>> look again while I play with reverting r1491887, but otherwise I fear >>> we're going to have to release 2.0.9 with these known Windows failures >>> so that it doesn't hold up an otherwise good Linux release any longer. >>> >>> So I will probably start merging httpd24threading back to trunk soon >>> if there are no objections... >> >> >> That sounds great :). From my point of view it's good plan. I would vote for >> releasing some alpha version so we could get some more testing like the one >> provided by Alexander earlier this week. >> >> For me the Linux version is working properly on Fedora, but on different >> distributions and systems we could find out more bugs like that one, but >> without anyone else testing httpd24threading branch it will be quite hard to >> move with httpd-2.4 support. >> >> Regards, >> Jan Kaluza >> > > With r1491887 reverted I now have all tests passing when using > httpd-2.2 (yay! that makes me much happier), and with httpd-2.4 I only > have the expected failure plus one of the sets of Windows-specific > failures (protocol/echo_block.t, echo_nonblock.t, echo_timeout.t & > pseudo_httpd.t). > > That would be acceptable for release (I really wasn't happy with a new > release breaking 2.2 support...) so I think we'll have to revert > r1491887 for now, and then I'll get merging stuff back to trunk after > looking at that remaining group of failures again. > Great, will you do the reverting part too? Jan Kaluza