Enable page caching

It might be possible to do this by calling drupal_page_is_cacheable(TRUE) after the drupal_page_is_cacheable(FALSE) happens in the captcha module. If that doesn't work, then it would require a change to the captcha module to make it so that add-on captcha modules, such as hidden_captcha, can disable the call to drupal_page_is_cacheable(FALSE). Patch welcome.

Enable cacheable captcha support (once 2449209 is committed):

I just implemented that for captcha, so the fix is really simple now.

A custom validation function is needed, because the solution is not actually available (and the random value inserted is returned instead), but as its the same empty response always, this works fine.

The simpler the patch, the more bugs crop in ...

=== should be used instead of == to ensure only really empty '' responses are counted.

The simpler the patch, the more bugs crop in ...

=== should be used instead of == to ensure only really empty '' responses are counted.

Looks good to me

The contents of hidden_captcha_captcha_validation() can be just:
return $response === '';

I don't understand the reasoning for creating hidden_captcha_captcha_validation().

+1 to RTBC.

Minor code clean up suggestion:

+++ b/hidden_captcha.module
@@ -73,6 +78,17 @@ function hidden_captcha_captcha($op, $captcha_type = '') {
+  if ($response === '') {
+    return TRUE;
+  }
+
+  return FALSE;

Could be just one liner:

return $response === '';

Sorry for duplicate comment @Liam Morland, clearly I didn't read #6. Here is the updated patch.

I still don't understand the reasoning for creating hidden_captcha_captcha_validation(). This wasn't needed before, why now?

I think it's Form API to regenerate the hash, so that we cache the form without allowing to submit duplicate token.

I don't understand. It should be explained in a comment.

Sorry, it wasn't clear comment. I tried to explain better in code comment in this page.

The reason why the new function is needed is because 'sid' is empty for cached pages, so a new sid is generated, but the default value for a new sid is a random token. Hence it would try to compare an empty input (correct) with a random token (incorrect).

Adding a new function for validation circumvents this as we are not depending on the state of the captcha (random token if sid is not set) at all.

Back to RTBC, but parent issue still needs to be committed first.

Postponing until parent is committed.

Parent is committed. Back to RTBC.

Thanks!

Thank you!