1. 30 Nov, 2017 1 commit
  2. 19 Nov, 2017 1 commit
    • Ilya Tregubov's avatar
      Bug 1685049: Remote file system modification · 78c87713
      Ilya Tregubov authored
      Enables Mahara to save files to an external file system
      - object storage (such as AWS's S3) -
      which can reduce the cost of storage
      Change-Id: I76822612f2922ba0ef2a0b7a4efb9cd2b96979a6
  3. 30 Mar, 2017 1 commit
    • Kristina Hoeppner's avatar
      Bug 1677087: Various lang string updates · 1881391a
      Kristina Hoeppner authored
      Some string IDs changed so that they are
      picked up by the translators while others
      aren't changed when they are just changes
      for English.
      Change-Id: I9afb8980492937f361e6e35361245c689b5a4413
  4. 22 Dec, 2016 1 commit
  5. 08 Aug, 2016 1 commit
    • Aaron Wells's avatar
      Bug 1570221 Don't print parameter values to logs when in production mode · 9a297249
      Aaron Wells authored
      The best way to prevent sensitive data from being printed to the logs
      is to avoid printing the value of *any* parameter. For instance, a
      password parameter may have an unusual name, or it may be passed
      through a general-purpose function like "strlen()".
      Since parameter values are useful for debugging, we can still print
      them when not in production mode (although with known password
      params still scrubbed out).
      Note this patch both scrubs likely password params, and hides their
      scrubbed value. That's mostly because I'm lazy, but it also obscures
      the password's actual length.
      Change-Id: I4a1ab4c89a169c6b29a7b63384c2412cee761ab7
      behatnotneeded: Can't test with behat
  6. 05 Jul, 2016 1 commit
  7. 08 Jun, 2016 1 commit
    • Aaron Wells's avatar
      Bug 1588613: Later session start so we can use DB config table · 12cb73cf
      Aaron Wells authored
      The session init code relies on $CFG->session_timeout, which is
      normally defined in the config table. So, we need to start the
      session after opening the database connection.
      (In the event that there's an earlier session start, for instance
      due to an error message, this will cause the session for that
      page load to disregard any database config values. But that's not
      a show-stopper, and there's no easy way to fix it.)
      Change-Id: Iffbeebc8e92929970a558ff0fbc726719bb92741
      behatnotneeded: Covered by existing tests
  8. 12 May, 2016 1 commit
  9. 21 Apr, 2016 1 commit
    • Robert Lyon's avatar
      Bug 1565546: Allowing $cfg setting to be json · af64abec
      Robert Lyon authored
      1) Allowing $CFG to accept json encoded strings and to be decoded back to php
      2) Making the $cfg->openbadgedisplayer_source a json encoded string
      3) Allowing the openbadges have the defaults only on one place and warn when they are missing
      Change-Id: Ica0349d6343d9f608b2272117d7412b288799278
      Signed-off-by: Robert Lyon's avatarRobert Lyon <robertl@catalyst.net.nz>
  10. 18 Apr, 2016 1 commit
    • Aaron Wells's avatar
      Bug 1570744: Fixing session bugs · 83ec33f2
      Aaron Wells authored
      This patch does 2 things:
      1. It loads the session much earlier during init.php. We wind
      up creating one on *every* script load anyway, due to LiveUser's
      constructor. Sometimes it gets created earlier if other code
      tries to use it before then, which adds some unpredictability
      to things. Moving it up to the top of init.php reduces that
      2. It turns out that in PHP 5.3, using header_remove('Set-Cookie')
      to only doesn't remove session headers. But header_remove()
      (with no params) to remove *all* cookies does remove them. So
      I'm changing remove_duplicate_cookies() to use that instead.
      3. Also in PHP 5.3, session headers are visible in headers_list().
      In situations where your session id changes (due to session_destroy()
      and session_regenerate_id()), our use of array_unique() meant we
      would preserve the old and new session IDs and send both back
      to the browser. This patch makes remove_duplicate_cookies() aware
      of the current session ID, and it only preserves that one.
      Change-Id: I7a90b8692a5f97429415aa9a17451a44cd2109dd
      behatnotneeded: Covered by existing tests
  11. 21 Mar, 2016 1 commit
  12. 28 Feb, 2016 1 commit
    • Aaron Wells's avatar
      Adding some HTTP headers for security (Bug 1531987) · 29656f03
      Aaron Wells authored
      X-XSS-Protection: Tells the browser not to disable XSS protection
      X-Content-Type-Options: Tells the browser not to try to guess at
      mimetypes of downloads
      X-Permitted-Cross-Domain-Policies: Tells Flash & PDF not to trust
      alternate crossdomain.xml files (which set the permissions on whether
      this site allows itself to be accessed by scripts in Flash & PDF).
      Prevents an attacker from uploading a more permissive crossdomain.xml
      X-Powered-By: PHP by default sends this header with the current full
      PHP version.
      behatnotneeded: Selenium can't examine HTTP response headers
      Change-Id: Ia2a6de971fc62b7d8806ad010aa0fbe37c1a7357
  13. 18 Feb, 2016 1 commit
  14. 10 Jan, 2016 2 commits
  15. 15 Dec, 2015 1 commit
    • Aaron Wells's avatar
      Removing obsolete "disablelogin" setting · 0284f9ab
      Aaron Wells authored
      Bug 1526076: I believe the initial intent was that
      Mahara core, and/or each plugin, could add a value
      to its version.php file indicating "disablelogin"
      true or false. And in this way, an upgrade could
      indicate whether it was a small enough upgrade that
      users did not need to log out for it.
      However, in practice this is not practical because
      we don't know what version of Mahara the user is
      upgrading from, and that is what determines whether
      or not it's a "stable" upgrade.
      Additionally, the core disablelogin has been set to true
      for the past 7 years, and the plugin disablelogin
      setting no longer has any effect.
      Removing disablelogin should hopefully make our
      maze of init.php auth_setup() login stuff a little
      bit easier to follow.
      behatnotneeded: Covered by existing tests
      Change-Id: I5f8a2b4faa95b9225bb926de6a54a622ea1a9618
  16. 14 Dec, 2015 1 commit
    • Aaron Wells's avatar
      Rename $CFG->siteclosed to $CFG->siteclosedforupgrade · 1404fe80
      Aaron Wells authored
      Bug 1526101: This should help make it clearer what's going
      on in init.php and the related auth code, by making the
      distinction between $CFG->siteclosed and $CFG->siteclosedbyadmin
      behatnotneeded: Covered by existing tests
      Change-Id: I8bc728622ae965ce25b55ee4b55278771fc1eedc
  17. 12 Dec, 2015 1 commit
  18. 25 Nov, 2015 1 commit
  19. 11 Nov, 2015 1 commit
    • Jono Mingard's avatar
      Remove unused and superfluous JavaScript (Bug #1323920) · a4dc90b3
      Jono Mingard authored
      Some of these files aren't loaded on any pages, some (ie. debug.js)
      are no longer necessary with modern debugging tools, and some have
      been replaced by Bootstrap functionality
      behatnotneeded: should be functionally identical
      Change-Id: I6d1b3874de5d42ccc00a8c0d2bb0e8bc162747d4
  20. 23 Sep, 2015 1 commit
  21. 20 Jun, 2015 1 commit
  22. 15 Jun, 2015 4 commits
  23. 19 May, 2015 1 commit
  24. 27 Mar, 2015 1 commit
    • Aaron Wells's avatar
      Bug 1427901: Performance improvements for cron · b4c1755f
      Aaron Wells authored
      - Cron doesn't need to run auth_setup()
      - Don't run cron when site is closed for upgrade
      - Get rid of forcelocalupgrade() option because it's no longer needed
      Change-Id: I1718b13337c50fadc0573d04f5b3d6b20bc842c2
  25. 02 Mar, 2015 1 commit
  26. 04 Dec, 2014 1 commit
  27. 03 Dec, 2014 1 commit
  28. 30 Oct, 2014 1 commit
  29. 15 Oct, 2014 1 commit
    • Aaron Wells's avatar
      Removing the redundant new local function for style overrides · 20da6bf8
      Aaron Wells authored
      Bug 1346926: This bug replicates the functionality of the
      newly enhanced (in bug 1328310) local/theme directory. I'm
      updating the documentation of local/theme at the same time,
      to try to avoid this kind of replication in future.
      Change-Id: Ia36442ac264f5e0740278592e734ddc0838bb80f
  30. 22 Jul, 2014 1 commit
  31. 30 Jun, 2014 1 commit
  32. 16 Jun, 2014 1 commit
  33. 04 Jun, 2014 1 commit
  34. 29 Jan, 2014 1 commit
  35. 21 Jan, 2014 1 commit
  36. 13 Jan, 2014 1 commit
    • Aaron Wells's avatar
      Silence most E_STRICT errors · 8d17e071
      Aaron Wells authored
      Bug 1268746: In PHP 5.4 E_ALL changed to include E_STRICT, causing Mahara to throw
      a lot of strict standards errors. This should silence most of them.
      HOWEVER, because most strict standards happen at compile-time, this will have no
      effect on strict standards errors caused in the files that have already been
      loaded by the time we call error_reporting() and set_error_handler(), which includes:
       - The file invoked directly by the URL
       - init.php
       - errors.php
       - config.php
       - config-defaults.php
      Change-Id: I7a7fdf7facb1f30e186a0e8a27f1c3b7473200da