DB error: Out of range value for column 'objectindex'

Updating issue tags...

Hi,
I have a similar issue when doing a site install where I i18n-string import some panel-related strings, where the i18n_string module also returns an error.

After some debugging and inspecting the DB, it seems that i18n_string converts the panels UUID in "objectid" into an integer to fill in the "objectkey".

protected function save_string($string, $update = FALSE) {
    if (!$string->get_source()) {
      // Create source string so we get an lid
      $this->save_source($string);
    }
    if (!isset($string->objectkey)) {
      $string->objectkey = (int)$string->objectid; <----- UUID to INT goes wrong.
    }
    if (!isset($string->format)) {
      $string->format = '';
    }

    $status = db_merge('i18n_string')
      ->key(array('lid' => $string->lid))
      ->fields(array(
          'textgroup' => $string->textgroup,
          'context' => $string->context,
          'objectid' => $string->objectid,
          'type' => $string->type,
          'property' => $string->property,
          'objectindex' => $string->objectkey,
          'format' => $string->format,
      ))
      ->execute();
    return $status;
  }

I updated the module with a rather weak patch to set the "objectkey" to 0 in case of UUID type strings to solve my problem, but I think there should be better ways to improve this.

making the objectindex from int to bigint solves the problem

@GiorgosK - agreed - it should be easy to patch the install file for this. I should probably do this when I have time!

Patch based on #8

- update column objectindex to bigint
- check if objectid is numeric before converting to objectkey
- make sure objectkey is numeric before saving

- update column objectindex length to 20.
Please review my patch.

- update column objectindex length to 20.
Please review my second patch.

Fixed problem in the class.

#15 Nice catch thanks for rerolling

Anyone willing to RTBC this patch under oath that it's regression free?

Will this work with postgresql as well?
I imagine the abastraction layer takes care of this to make it compatible with Postgresql using it's driver and mssql using the sqlsrv driver

#18 I ran this on Postgres without a problem

ok, I'll commit this to 7.x-1.x-dev sometime soon thanks.

Thanks again for all the hard work. Without having actually tested it, just reviewing the code visually, the patch looks really well written.

Yahooo
"ok, I'll commit this to 7.x-1.x-dev sometime soon thanks."
@joseph.olstad Thanks )))

fixed in 7.x-1.x dev

Please try to break this asap.

We should plan to release this in 2 weeks.
'release' meaning, tag a new release of i18n in 2 weeks.

unless of course someone objects.

released in 7.x-1.22

There have been reports of regressions.

#2926290: Database schema inconsistent i18n_string table on clean install i18n module.

I have not seen this myself however enough people have raised issues to make it a valid concern. Trying to determine what versions of database servers and what type of database server (MySQL vX.X / Postgresql X.X / MSSQL Server x.x / sqlite version X, etc)

looking for the version info

followup in

#2926290: Database schema inconsistent i18n_string table on clean install i18n module.