Re: Strange survey input

Joel Palmius <[email protected]> Wed, 5 Apr 2006 12:25:47 +0200 (CEST)
Newsgroups gmane.comp.apache.mod-survey.general
Message-ID <[email protected]>
The 999 thing is probably a session timeout. The current timeout for the 
session is one hour. This means that the user must have paused somewhere 
and done nothing for one hour and then continued.

The session timeout can be changed at line 102 in Session.pm. (The comment 
says 10 minutes but that's obsolete).

Otherwise 999 is used to denote "not seen" questions. Ie, a page skipped 
in a branching survey for example.

It's a bug that session timeouts do not produce user-visible errors.

The CUSTOM mustanswer thing.. Currently, I don't see any way to modify 
that error message. You could possibly think up a solution involving a 
submit-time perl snippet and intercepting the submit, but it might be 
tricky. Another way to get around it would be using javascript and the 
ACCESSJS/ACCESSFUN parameters of the SUBMIT tag.

   // Joel


On Tue, 4 Apr 2006, George Clark wrote:

>
> Hi all,
>
> I just received an unusual survey response and wondered if anyone has any
> suggestions on the cause.  I'm running mod_survey on apache 1, on gentoo linux.
> www-apache/mod_survey-3.2.3-r2
>
> *Every* field in the survey was filled with 999,  including the IP address and
> version fields.   I could see that someone might have entered 999 in every
> field, but I can't see how the survey version and IP address would be
> corrupted.  They are env variables and a constant.
>
> The only field that was not exactly 999 was the survey key, which appeared to
> be valid.
>
> I've checked logs, and the only operation I see in the log was a GET.  No POSTs
> occur matching the time of the record.  So I really can't explain what
> happened.  And not hints of other errors were in the logs.
>
> (I saw something similar today at work where around half of the fields of one
> survey respose were 999 - including things like lickert controls.)
>
> And on another topic,  I've put together some custom questions that are
> MUSTANSWER=yes.  However the error message only states that a custom question
> must be answered.   That doesn't leave the user with much of a clue on what is
> wrong.  Is there any way to provide an error response for the custom fields so
> that the user gets a hint about what is required, either through an
> undocumented variable or a perl snippet?
>
> Thanks,
> George
> Skickat av "George Clark" <[email protected]>
> till survey-discussion
>
Skickat av Joel Palmius <[email protected]>
till survey-discussion