Review main Merged

55849[BUGFIX] Pass original extbase response object to form framework finishers

Benjamin Franzke CR +2 V +2 Easy +13 −2 updated 03/08/18, 01:54

Commit message

The original extbase response was passed as second parameter
to ObjectManager->get() – that did effectively nothing. The second get()
parameter is not a fallback parameter (as probably supposed by the author)
but the first constructor argument (but Response has no constructor).
Now, rather use a coalesce operator to use the original response or create
an own, if required.

Also handle the StopActionException in case no extbase response
is available (e.g. when rendered through FluidTemplateContentObject).

Due to this fixes the redirect finisher can be adapted to stop echo'ing
the response content on it's own. Directly echo'ing the response content
caused wrong Content-Length headers to be generated as the core didn't
know that content has been echo'ed. (The redirect response contains both a
header based redirect and html meta redirect; due to early echo that meta
redirect is prepended to the regular html output, and is missing in the
Content-Length calculation.)
That casused certain server environments (e.g. gzip, keep-alive,
proxy) and browsers (e.g. Android, Opera desktop) to display gzip
compressed data as garbage in the browser.

Review

  • Code review CR +2

    Imported before Forger kept the single votes, the next import adds them.

  • Verified V +2

    Imported before Forger kept the single votes, the next import adds them.

Issues 1

Files 2