Hi!
We've made a great progress on fixing and cleaning up the unit tests,
I'd like to thank everybody involved for their contribution. We still
have some failures and tests with unclear status, so I think it's better
we take another week before we release the beta and try to work them
out, just to be sure. So I think it makes sense to postpone the beta to
15th.
Stanislav Malyshev, Software Architect
SugarCRM: http://www.sugarcrm.com/
(408)454-6900 ext. 227
Hi!
We've made a great progress on fixing and cleaning up the unit tests, I'd like to thank everybody involved for their contribution. We still have some failures and tests with unclear status, so I think it's better we take another week before we release the
beta and try to work them out, just to be sure. So I think it makes sense to postpone the beta to 15th.
As a heads-up, Felipe has updated gcov.php.net with a new page to show
expected failures. Currently only the 5.3 build has completed since
this change was merged:
http://gcov.php.net/viewer.php?version=PHP_5_3&func=expected_tests
For 5.4, there are still are a bunch of systematic failures on gcov
e.g. imap, pdo_firebird, pdo_odbc that need attention, along with less
obvious issues. Volunteers welcome.
Getting gcov tests resolved should be our aim. Even though gcov
builds are not continuously run or fully representative of all the
extensions and potential environments it is visible and is how the
world rightly or wrongly measures PHP
(http://epixa.com/2011/08/how-php-is-broken-and-how-it-can-be-fixed)
Chris
--
Email: christopher.jones@oracle.com
Tel: +1 650 506 8630
Blog: http://blogs.oracle.com/opal/
Christopher Jones wrote:
For 5.4, there are still are a bunch of systematic failures on gcov
e.g. imap, pdo_firebird, pdo_odbc that need attention, along with less
obvious issues. Volunteers welcome.
I've asked a couple of times now about fixing the PDO tests, most of the
pdo_firebird failures are due to different SQL formats which are actually not
handled by PDO at all, so how do we fix a pdo test designed for one database
when it fails in another? The tests do seem to have driver specific elements,
but that just make things more difficult to manage :(
The other question has to be which version of database the tests are being run
against ... like MySQL some things have changed from older versions which is the
reason for the interbase test failures. We DO need to think about a separate
firebird version of the driver, which is then used by pdo_firebird, since
interbase gives different results to firebird nowadays. fbird_x versions of the
functions have been aliases since the early days of PHP5 but now need to be a
separate extension. Since interbase is a licenced product not many of us even
have a copy to run, so firebird is the more normal test base.
--
Lester Caine - G8HFL
Contact - http://lsces.co.uk/wiki/?page=contact
L.S.Caine Electronic Services - http://lsces.co.uk
EnquirySolve - http://enquirysolve.com/
Model Engineers Digital Workshop - http://medw.co.uk//
Firebird - http://www.firebirdsql.org/index.php
Christopher Jones wrote:
For 5.4, there are still are a bunch of systematic failures on gcov
e.g. imap, pdo_firebird, pdo_odbc that need attention, along with less
obvious issues. Volunteers welcome.I've asked a couple of times now about fixing the PDO tests, most of the pdo_firebird failures are due to different SQL formats which are actually not handled by PDO at all, so how do we fix a pdo test designed for one database when it fails in another?
The tests do seem to have driver specific elements, but that just make things more difficult to manage :(The other question has to be which version of database the tests are being run against ... like MySQL some things have changed from older versions which is the reason for the interbase test failures. We DO need to think about a separate firebird version of
the driver, which is then used by pdo_firebird, since interbase gives different results to firebird nowadays. fbird_x versions of the functions have been aliases since the early days of PHP5 but now need to be a separate extension. Since interbase is a
licenced product not many of us even have a copy to run, so firebird is the more normal test base.
Lester,
File a bug and attach test patches that satisfy you (and run on gcov).
Or merge the patches yourself - but I'm not sure you have karma???
I've never heard of anyone else running these tests so your help would
be appreciated. I won't sidetrack this thread with long term comments
about PDO or the PHP test infrastructure. Let's work with what we
have access to.
For "generic" tests in ext/pdo/tests that should not run for Firebird,
update the SKIPIF section to skip them. For example,
ext/pdo/tests/bug_43130.phpt doesn't run on sqlite, Postgres or Oracle
DB. The SKIPIF includes:
if (!strncasecmp(getenv('PDOTEST_DSN'), 'sqlite', strlen('sqlite'))) die('skip not relevant for sqlite driver');
if (!strncasecmp(getenv('PDOTEST_DSN'), 'pgsql', strlen('pgsql'))) die('skip not relevant for pgsql driver');
if (!strncasecmp(getenv('PDOTEST_DSN'), 'oci', strlen('oci'))) die('skip not relevant for oci driver - Hyphen is not legal for bind names in Oracle DB');
Other tests may need SQL tweaks. For example bug_43139.phpt needs a
different query on Oracle:
$from = '';
if ($db->getAttribute(PDO::ATTR_DRIVER_NAME) == 'oci') {
$from = 'from dual';
}
var_dump($db->query("select 0 as abc, 1 as xyz, 2 as def $from")->fetchAll(PDO::FETCH_GROUP));
If a test is important but fails on Firebird due to an extension bug,
then consider skipping it in pdo/tests and making a copy in
ext/pdo_firebird with an --XFAIL-- section that explains why the test
fails on Firebird.
For tests that need to cater for multiple Firebird or Interbase
versions, make multiple copies with different SKIPIFs.
Don't forget to verfiy the results with 5.3, 5.4 and trunk.
Chris
--
Email: christopher.jones@oracle.com
Tel: +1 650 506 8630
Blog: http://blogs.oracle.com/opal/
Christopher Jones wrote:
File a bug and attach test patches that satisfy you (and run on gcov).
Or merge the patches yourself - but I'm not sure you have karma???
I've never heard of anyone else running these tests so your help would
be appreciated. I won't sidetrack this thread with long term comments
about PDO or the PHP test infrastructure. Let's work with what we
have access to.
Thanks for that feedback.
Part of the problem here is that PDO is not at all helpful when working with
Firebird. I use ADOdb exclusively for cross database support, and switching on
the PDO driver instead of the generic one throws up a lot more problems.
But what I do have is a crib sheet from 3 years ago that corrects many of the
reported errors - but no karma to apply them :)
--
Lester Caine - G8HFL
Contact - http://lsces.co.uk/wiki/?page=contact
L.S.Caine Electronic Services - http://lsces.co.uk
EnquirySolve - http://enquirysolve.com/
Model Engineers Digital Workshop - http://medw.co.uk//
Firebird - http://www.firebirdsql.org/index.php
Christopher Jones wrote:
File a bug and attach test patches that satisfy you (and run on gcov).
Or merge the patches yourself - but I'm not sure you have karma???
I've never heard of anyone else running these tests so your help would
be appreciated. I won't sidetrack this thread with long term comments
about PDO or the PHP test infrastructure. Let's work with what we
have access to.Thanks for that feedback.
Part of the problem here is that PDO is not at all helpful when working with Firebird. I use ADOdb exclusively for cross database support, and switching on the PDO driver instead of the generic one throws up a lot more problems.But what I do have is a crib sheet from 3 years ago that corrects many of the reported errors - but no karma to apply them :)
Log a bug with verified patches and I'm sure we can get them applied.
Chris
--
Email: christopher.jones@oracle.com
Tel: +1 650 506 8630
Blog: http://blogs.oracle.com/opal/
I would like to mention this segfault that i am having with 5.4 alpha 3:
https://bugs.php.net/bug.php?id=55578
The reproduce case is trivial (thanks to Felipe and Tyrael for improving it
even more)
greetings,
Benjamin
On Thu, Sep 8, 2011 at 8:01 PM, Stas Malyshev smalyshev@sugarcrm.comwrote:
Hi!
We've made a great progress on fixing and cleaning up the unit tests, I'd
like to thank everybody involved for their contribution. We still have some
failures and tests with unclear status, so I think it's better we take
another week before we release the beta and try to work them out, just to be
sure. So I think it makes sense to postpone the beta to 15th.Stanislav Malyshev, Software Architect
SugarCRM: http://www.sugarcrm.com/
(408)454-6900 ext. 227
Hi!
I would like to mention this segfault that i am having with 5.4 alpha 3:
https://bugs.php.net/bug.php?id=55578The reproduce case is trivial (thanks to Felipe and Tyrael for improving it
even more)
FWIW, was unable to reproduce it on Darwin with current SVN.
--
Stanislav Malyshev, Software Architect
SugarCRM: http://www.sugarcrm.com/
(408)454-6900 ext. 227
Hi!
We've made a great progress on fixing and cleaning up the unit tests, I'd like to thank everybody involved for their contribution. We still have some failures and tests with unclear status, so I think it's better we take another week before we release the
beta and try to work them out, just to be sure. So I think it makes sense to postpone the beta to 15th.
Is anyone looking at PHP 5.4's openssl failures on gcov? E.g.
http://gcov.php.net/viewer.php?version=PHP_5_4&func=tests&file=ext%2Fopenssl%2Ftests%2Fbug47828.phpt
This particular test always passes for me.
Chris
--
Email: christopher.jones@oracle.com
Tel: +1 650 506 8630
Blog: http://blogs.oracle.com/opal/