2014-07-09 23:52 GMT+02:00 Chirayu Chiripal <chirayu.chiripal@gmail.com>:

On Thu, Jul 10, 2014 at 3:10 AM, Hugues Peccatte <hugues.peccatte@gmail.com> wrote:

2014-07-09 19:37 GMT+02:00 Hugues Peccatte <hugues.peccatte@gmail.com>:

2014-07-09 17:34 GMT+02:00 Isaac Bennetch <bennetch@gmail.com>:

Hello,

In the course of my testing the new input transformation feature, I've
encountered some trouble that Chirayu is unable to reproduce. We've had
a bit of discussion privately about it, but wanted to move discussion
back to the mailing list.

Note that running his branch[1] will change your pma__column_info table
structure; so you may wish to back it up for testing purposes.

The problem at the moment is the automatic table upgrade feature. What
is supposed to happen is that, when logging in,
PMA_checkRelationsParam() verifies that the correct table structure
exists for pma__column_info. If it doesn't, it attempts to run
./examples/upgrade_column_info_4_3_0+.sql (after substituting the
correct database and table names if needed). This is all quite good.

On my system, what happens, however, is that

$GLOBALS['dbi']->tryMultiQuery($query, $GLOBALS['controllink'])

doesn't work. It returns 1 (which I presume is just how print_r
represents TRUE), but the query doesn't execute ($query in this case is
essentially the contents of the .sql file; several SQL statements with
semicolons and newline characters and a few comments). It works properly
on Chirayu's system but silently fails on mine.

I'm hoping to get additional testing on the "silent upgrade" feature
(or, for that matter, thoughts about the problem). At this point, I'm
the only one for whom it fails, which is quite interesting. To test,
simply checkout the branch and open phpMyAdmin, then check your
pma__column_info table; it should have two new columns at the end (input_*).

Thanks!

Some more detailed information if you really want to read on:

Around line 610, instead of directly testing the result of the
tryMultiQuery in the if statement, I assign it to a variable then
print_r that variable and get a message about "Commands out of sync; you
can't run this command now".

            $ipb_test = $GLOBALS['dbi']->tryMultiQuery($query,
$GLOBALS['controllink'] );
die("<pre>" . print_r($GLOBALS['controllink']) . "</pre>");
            if ($ipb_test !== false ) {
                return true;

mysqli Object ( [affected_rows] => 0 [client_info] => 5.5.37
[client_version] => 50537 [connect_errno] => 0 [connect_error] =>
[errno] => 0 [error] => [error_list] => Array ( ) [field_count] => 0
[host_info] => Localhost via UNIX socket [info] => [insert_id] => 0
[server_info] => 5.5.37-0+wheezy1 [server_version] => 50537 [stat] =>
Commands out of sync; you can't run this command now [sqlstate] => HY000
[protocol_version] => 10 [thread_id] => 6069 [warning_count] => 0 )

I'm not familiar enough with tryMultiQuery to know why [stat] contains
this message.

According to [2], it appears as if there is a problem with executing the
code because either we haven't used mysql_free_result() or
mysql_use_result() or mysql_store_result() since a previous query.

We've looked at mysqli_multi_query()
(http://www.php.net/manual/en/mysqli.multi-query.php), which
is called by tryMultiQuery() depending on your DBI Extension (mine
is mysqli). It says that it returns false when First statement is
failed. See the return values section of it.

Note that my mysqli object claims 0 rows were affected but also didn't
return false. I expect that the 87 rows in my pma__column_info would
have been affected by the first line (ALTER...).

1 - https://github.com/D-storm/phpmyadmin/commits/FR-637
2 - https://dev.mysql.com/doc/refman/5.0/en/commands-out-of-sync.html

Hi,

It seems that I found the same information while reading your email and doing some research on internet about mysqli_multi_query and its return.

So maybe we need to improve PMA_DBI_Mysqli::realMultiQuery to return the real error status.

But first maybe, you can try something like this in this method (I didn't test it…):

if ($r = mysqli_multi_query($link, $query)) {

    do {

        if ($result = mysqli_store_result($link)) {

            echo 'ok.<br />';

            mysqli_free_result($result);

        } else {

            $r = false;

            echo 'KO<br />'; //Do you know a way get the query here? I didn't find one…

        }

    } while (mysqli_next_result($link));

}

return $r;

Then, you'll know the query which failed.

We're waiting for your feedback. (I'll try… to try…)

Hugues.

Sorry, I just did some research and it can work. mysqli_store_result can''t be used with INSERT, UPDATE, DELETE, ALTER statement. But we may try to use mysqli_error to really if there is an error or not.

I'll try to do other tests with mysqli_error, maybe.

Hugues.

Yeah, you are right. I tested it using this code,

mysqli_multi_query(
    $GLOBALS['controllink'], $query
);
do {
    var_dump(mysqli_error($GLOBALS['controllink']));
} while (mysqli_next_result($GLOBALS['controllink']));
var_dump(mysqli_error($GLOBALS['controllink']));

and for the last var_dump() I got the error which I intentionally did and for other queries which were executed successfully it returned an empty string.

--

Regards,
Chirayu Chiripal
phpMyAdmin Intern - Google Summer of Code 2014
https://chirayuchiripal.wordpress.com/

Oh thanks! I was about giving up!

So I defined the following method (and I think that should be in master) :

    public function realMultiQuery($link, $query)

    {

        if (false === mysqli_multi_query($link, $query)) {

            return false;

        }

        while (mysqli_next_result($link));

        return('' === mysqli_error($link));

    }

And it seems to be ok.

The only thing that I don't like is that the script is half executed but can't be reverted… But I'm not sure there is a way to manage it… There are some DB structure modifications and we can't revert this, so it would need to write a specific "undo" script. But even with this, how to automatically know if all the undo script has to be executed are only a part and which one ?

(If we know the number of queries, we might imagine, but not sure that would be a good solution…)

Hugues.