Review main Merged

8764[FEATURE] Allow authentication with partial OpenID identifier

Xavier Perseguers CR +2 V +2 Normal +80 −27 updated 06/03/13, 20:49

Commit message

Instead of forcing the user to enter her full OpenID identifier when authenticating,
a partial OpenID identifier, without the protocol, should be allowed.

Review

  • Code review CR +2
    • Steffen Ritter +2
    • Stefan Neufeind +1
    • Xavier Perseguers +1
    • Philipp Gampe +1
  • Verified V +2
    • Steffen Ritter +2
    • Xavier Perseguers +1
    • Philipp Gampe +1

Issues 1

Files 1

Review log 29

Show 4 earlier messages
  1. Xavier Perseguers commented on patch set 1 02/10/12, 12:44

    I would prefer that you didn't submit this Order of quoting is wrong: http://buzz.typo3.org/teams/security/article/correct-usage-of-typo3-database-api/
  2. Steffen Müller commented on patch set 2 02/10/12, 12:44

    (1 inline comment) just a minor glitch (see comments) but the rests looks good to me. maybe you give some feedback to my inline comment, then I'd give +1.
  3. Steffen Müller commented on patch set 2 1 comment 02/10/12, 12:44

    I would prefer that you didn't submit this see my inline comment
  4. Dmitry Dulepov commented on patch set 2 1 comment 02/10/12, 12:44

    I would prefer that you didn't submit this Going to publish another patch.
  1. Dmitry Dulepov commented on patch set 2 02/10/12, 12:44

    This work is based on the code we use at Snowflake for years. It is not an exact copy, because we have further enhancement, but pretty close.
  2. Dmitry Dulepov uploaded patch set 3 02/10/12, 12:44

  3. Dmitry Dulepov commented on patch set 3 02/10/12, 12:44

    (1 inline comment)
  4. Xavier Perseguers uploaded patch set 2 02/10/12, 12:44

  5. Xavier Perseguers commented on patch set 2 02/10/12, 12:44

    (1 inline comment)
  6. Xavier Perseguers commented on patch set 3 02/10/12, 12:44

    Verified; Looks good to me, but someone else must approve As the patch is not from me anymore, review for Dmitry's one
  7. Helmut Hummel commented on patch set 1 02/10/12, 12:44

    I would prefer that you didn't submit this As discussed via Skype my comments

    class.tx_openid_sv1.php:295

    treating a string with escapeStrForLike() and after that with quoteStr() is wrong. See: http://buzz.typo3.org/teams/security/article/correct-usage-of-typo3-database-api/

    class.tx_openid_sv1.php:301

    I would not use a LIKE query for that. Although some checks are done after that it feels dangerous to allow more than one result, where there should be only one

    class.tx_openid_sv1.php:307

    user input in the pattern part is not a good idea
  8. Philipp Gampe commented on patch set 3 02/10/12, 12:44

    (6 inline comments)

    class.tx_openid_sv1.php:455

    nitpick: uses space instead of tab

    class.tx_openid_sv1.php:458

    one more tab

    class.tx_openid_sv1.php:463

    one more tab

    class.tx_openid_sv1.php:482

    one more tab

    class.tx_openid_sv1.php:483

    space before second parameter

    class.tx_openid_sv1.php:610 Reply

    No, but a '\n' is required for all unix text files. Before the file ends with 0a 3f 3e. After the patch, the file ends with 0a 3f 3e 0a. This is totally valid as the next '\n' is always considered part of the "?>" by PHP.
  9. Dmitry Dulepov commented on patch set 4 02/10/12, 12:44

    (1 inline comment) Philipp, example is example, it is not a rule :) CGL does not tell whether we should space or tab. Existing code uses both. That's valid. So it is better not to change such thing because they add more work for other people to check your changes. Helmut, please, see my inline comment.
  10. Dmitry Dulepov commented on patch set 4 02/10/12, 12:44

    I would prefer that you didn't submit this Philipp, 2 of 5 changes are unnecessary or against CGL. > Line 455: * @return string Normalized OpenID identifier nitpick: uses space instead of tab That is ok. CGL does not require tabs there. Please, revert back this unnecessary change. > Line 610: ?> No, but a '\n' is required for all unix text files. No, it is not and it is against CGL, which clearly says: "There must be no empty lines after the closing PHP tag.". Please, revert this change. CGL, FYI: http://typo3.org/documentation/document-library/core-documentation/doc_core_cgl/4.5.1/view/1/3/
  11. Dmitry Dulepov commented on patch set 4 02/10/12, 12:44

    No score
  12. Stefan Neufeind commented on patch set 4 02/10/12, 12:44

    Looks good to me, but someone else must approve seems to make sense to my by reading
  13. Xavier Perseguers commented on patch set 4 02/10/12, 12:44

    Verified; Looks good to me, but someone else must approve
  14. Helmut Hummel commented on patch set 4 02/10/12, 12:44

    (2 inline comments)
  15. Philipp Gampe uploaded patch set 4 02/10/12, 12:44

  16. Philipp Gampe commented on patch set 4 02/10/12, 12:44

    Verified; Looks good to me, but someone else must approve Just fixed the mentioned parts (I only touched changed lines). BTW, I can not reproduce this bug. https://www.google.com/accounts/o8/id works for me even without this patch.
  17. Philipp Gampe commented on patch set 4 02/10/12, 12:44

    @Dmitry CGL says > There must be no empty lines after the closing PHP tag. Empty lines after closing tags break output compression in PHP and/or result in AJAX errors. And that what git tells you: -?> \ No newline at end of file +?> So there is no new line at the end of file. There is only a 0a ('\n') at the end of the last line. That was all what I was talking about. There is not even a way to tell vim to not add such a newline character at the end of the last line, because not having it might break some text program. According to the PHP doc, the trailing newline char is considered as part of the "?>". Not even diff can handle messing with the last char. This is the reason why git puts this comment in. As long as you do not change anything new the end of the file, you can just remove the last hunk by editing the patch by hand (git rebase -i or git add -e). This works as git follows a merge strategy and treads the rest of the file as binary stream. Maybe we should discuss this at the core list, because I do have a filter for git which can remove all 0a the the end of the file. The only problem is, that this filter works for any touched git files, so even if you pull. And then there is of course no why to reset the file ;) so this filter needs to be placed at upstream git. Conserning the method signature. I formated it according to the example in the CGL which uses a single space between the elements of the doc block.
  18. Philipp Gampe commented on patch set 4 02/10/12, 12:44

    I this can be merged.
  19. Philipp Gampe commented on patch set 4 02/10/12, 12:44

    I wrote "www.google.com/accounts/o8/id". Is gerrit too smart?
  20. Mr. Jenkins commented on patch set 1 02/10/12, 12:44

  21. Mr. Jenkins commented on patch set 2 02/10/12, 12:44

  22. Mr. Jenkins commented on patch set 3 02/10/12, 12:44

  23. Mr. Jenkins commented on patch set 4 02/10/12, 12:44

  24. Steffen Ritter commented on patch set 4 V 0 02/10/12, 12:44

    by team member; Looks good to me, approved
  25. Steffen Ritter merged patch set 4 02/10/12, 12:44